2D FSV Tool User's Guide
Contents
1. Combined ADMc Ext E Combined ADMi Ext z Combined FOM average txt E Combined _FDMc txt Combined FDMi Ext Combined GDM average Fxt E Combined GDMc Ext E Combined GDMi Ext l Combined Analysis mat Wi Combined ADMc Fig Combined ADMi Fig a Combined FOMc Fig Combined _FDMi fig Wi Combined GDMc Fig Combined GOMi Fig Fig 7 4 K factor x folder for Frequency Domain Combined analysis The data file structure and content is as described in Section 7 1 Note that all the data files relating the input data are absent because they are saved in the top level of output folder Fig 7 3 moreover the data files related to offset calculations ODM are absent because the offset computation is only performed on input components 8 Output Results Consider two Peaks distributions as input data in which Peaks 2 1s the mirror image of Input data 1 with respect the y axis 49 Input Dataset 1 Amplitude ASCII X Y ASCII File Edit View Insert Tools Desktop Window Help Dss r anale o Amplitude Input Data a Ea E T Fig 8 1 Input data example The significant FSV outputs are e ADM is a figure of merit of the comparison of amplitudes and trends of the two data sets to compare the lower the ADM the better the comparison e FDM is a figure of merit of the comparison of details derivatives of the two datasets to be compared the lower the FDM the better the2. Continue Analysis Fig 3 13 Two value of K weighting factor added When all K are inserted is possible to run the analysis for each K by pressing Continue Analysis button 62 1D FSV 4 0 21 ARLI K weighting 1 of 2 bo 0 00 23 CO YA Started at 13 50 08 cpu time used 0 00 08 Fig 3 14 Progress bar of K weighting procedure The GUI prevents the insertion of a pair of K weighting factors that have already been added or where invalid values of K are being entered 0 KmagnitudesK phase lt 1 and displays these warnings Warning ej Warning A F value already present F must be O lt K lt 17 Fig 3 15 K Warning Dialogs At the end of the calculations a new prompt is issued as in the next figure 1D FSY Analysis Ended E m Do vou want Fig 3 16 Window at the end of a complete 1D FSV analysis of a pair of data sets 20 If it is answered Load New Data then the 1D FSV Tool back to Initial Window Fig 3 1 and is prepared for a new analysis If it is answered View Results the figures of the results are shown see Section 5 2 4 Input Data Loading The current 1D FSV release take can be chosen from X Y ASCII and Y only ASCII formats For the data structure of the input files see Chapter 6 In this examples we assume that input data are stored in X Y ASCII structure Then each set 1s given by one file 4 1 Domain and data requirements Consider a dataset with a
3. FDM aR Compute K_ADM amp k_FDM ej Do vou want GDM weighting for Dataset 17 Do you want GDM weighting for Dataset 27 Fig 3 10 Windows for selecting the GDM weighting in combined analysis 3 4 2 1 K weighting factor After the GDM weighting procedures FSV requests to insert one or more Kmagnitude Kphase Weighting factors see section 1 2 The user can choose the value of the K factors by means of a GUI dialog 18 K weighting factor m BR Enter a value beetwen O and 1 in the windows or move the slider K magnitude Add K Continue Analysis Fig 3 11 K weighting factor choice GUI it can be directly inserted the value of Kynagnitude Or Kpnase and automatically the other one is computed according to 1 15 The default value for Kmagnitude and Kphase 18 0 5 Press Add K button to add a new pair of K weighting factors K weighting factor lE Enter a value beetven O and 1 in the windows or move the slider K magnitude 1Kvalues added Add Continue Analysis Fig 3 12 One value of K weighting factor added At least one pair of K weighting factors must be entered in order to continue analysis then the Continue Analysis button is enabled The counter in the bottom left orange box gives the number of K weighting factors added 19 K weighting factor eL Enter a value beetwen and 1 in the windows or move the slider I maanitude E K phase u HL Hh 2Kvalues added Add
4. In the combined analysis two pair of data sets are compared 1 e 1 Magnitude 1 Magnitude 2 2 Phase 1 Phase 2 For each pair of data sets 1 and 2 the previous above two condition must be valid in addition to be correctly combined the domain of each Magnitude and Phase parts must be identical In the example Magnitude 1 2 domain must be identical to Phase 1 2 domain 4 2 Loading X Y ASCII format This format can be used when data coming from different sources have to be compared 1D FSV can import data stored in two column ASCII file see Section 6 1 An example of loading one set of data in X Y ASCII format 1 Select the file 1 e inputdatal txt containing the data set 1 as in Fig 4 3 Select Amplitude Data 1 X Y ASCII Cerca in 4 FSV1D 402 data examples liba Tinputdatat txt inputdataz txt inputdatayonly txt Mome file linputdatal xl Tipo file txt Annulla Fig 4 3 Window for domain files selection The same sequence of windows is proposed for the loading of the second set of data 22 4 3 Loading Y onlyASCII format 1D FSV can import data stored in one column ASCII file see Section 6 2 An example of loading one CST ASCII set 1 Select the Y only ASCII data file 1 e inputdatalyonly dat containing the set as in Fig 4 4 Select Amplitude Data 1 Y only ASCII Cerca in 4 FSv1D 402 hd data examples liba E inputdatal txt inp
5. Insert Tools Desktop Window Help Css b a e g OB m Amplitude Input Data 1 5 1 45 0 4 qi ma f 30 1D FSV 4 0 2L Results Data display Grade Spread chart Data Display 0 6 0 8 1 0 6 0 8 1 Edit Figure Export Figure Edit Fiqure Export Figure Time Domain Analysis Time Domain Analysis ADMi ADMc FDMi FDMc GDMi GDMc Hide legend Hide synthetic figures Hide legend Hide synthetic figures Fig 5 1 The original data window and main Data Display Tool window In the main window of the Results GUI there are two drawing areas e Drawing box area in which the graph are displayed e Export Figure button take a screenshot of the currently graph displayed The displayed graph is exported in PNG format The figure is saved in a file located in the subdirectory inside the selected directory for output data with the name of the displayed FSV variable e Edit Figure button open the current graph displayed in drawing box in a new window Fig 5 2 with the full figure toolbar to inspect and analyze the figure e Selection box select the graph to display in corresponding drawing box from ADMi FDMi GDMi ADMc FDMc and GDMc radio buttons Hide legend checkbox hide the legend textbox in the ADMi FDMi and GDMi graphs Hide synthetic figures checkbox hide the annotation textboxes in the ADMc FDMc and GDMc graphs 3l Fi
6. Low _inputdatal txt Magnitude Low inputdataz Ext Magnitude _ODMi tect Phase ADM average Ext E Phase ADMc Ext Phase ADMi Ext E Phase DC inputdatal Ext Phase DC inputdataz Ext Phase FDM average Ext Phase _FDMc txt E Phase FDMi Ext Phase GDM average txt Phase GDMc Ext Phase GDMi Ext Phase High_inputdatal txt Phase High _inputdataz txt Phase inputdatal txt Phase inpubdakaz txt Phase Low _inputdatal txt E Phase Low inputdata2 txt Phase ODMi Ext 4 Magnitude_ADMc fig 0 Magnitude_ADMi fig 0 Magnitude DC inputdata fig d Magnitude _FOMe fig Magnitude _FOMi fig 0 Magnitude_GDMc Fig 0 Magnitude GDMi fig 4 Maanitude High inputdata Fig 4 Magnitude_inputdata fig 0 Magnitude_Low_inputdata fig H Magnitude_ODMi fig 4 Phase_ADMCc fig 4 Phase_ADMi Fig Phase_DC_inputdata Fig 0 Phase_FDMc fig 4 Phase_FDMi fig Phase_GDMc fig 0 Phase_GDMi fig 0 Phase_High_inputdata fig 4 Phase_inputdata fig Phase_Low_inputdata fig Phase_ODMi Fig n Magnitude Analysis mat E Phase Analvsis mat Fig 7 3 Output data folder for Frequency Domain Combined analysis Combined domain contains the input domain after synchronization between all the four sets The K_factor x subdirectories have this structure x is the Kmagnituae value 48 jGsthrs 85 Combined ADM _ average txt E
7. Tool is an application embedded in the FSV Tool that allows the user to visualize the relevant FSV variables associated to the comparison of two datasets The Data Display Tool plots the data contained in the mat files generated by the FSV procedure It can be started from the initial window by selecting the Plot Results Fig 3 1 option without running an FSV analysis 1D FSV 4 0 2L Results Data Display 1 0 9 0 8 0 7 0 6 0 5 0 4 0 3 0 2 0 1 0 6 0 8 1 0 6 0 8 Edit Figure Export Figure Edit Figure Export Figure Time Domain Analysis Time Domain Analysis Load data Folder 1 Folder 2 Hidelegend Hide synthetic figures Hidelegend Hide synthetic figures Fig 5 6 The Data Display Tool window before selecting the directory in which the mat files are contained After the initial window of the Data Display Tool appears on the screen select the folders directories containing the FSV Tool output data to be displayed They are contained in mat files The User can select a directory for Folder 1 and a different directory for Folder 2 by pressing the corresponding button The selected plots will appear in the corresponding windows left side for Folder 1 right side for Folder 2 In this way it 1s possible to load and visualize two different FSV e non combined analysis select the automatically generated subdirectory inside the selected directo
8. X YASCIIDHHI OIQ uuu u aaa IR Ru RUD u as s 41 Gali sCONVEISON toe XY ASCIIHT6li al uuu ku uu imita axe ia bv ead eias 41 62 Yon ASCII FOIE ici cione L cuna dr esdadckE HD Aa aaae aa OE Vu aka sayay ARESE 44 62 1 CONVEISON tO Y only ASC IL BOTIDSTE sir erii yuyu yu uuu y uyu s 44 7 Output Data Structure amp Location T T T 45 Jl Data Smut Ue ics cadet 46 Z2 NOncOmbIBedqddaikuu uuu u yu a i nra C D RE su 2 47 Td GCombmed d i uu tu REREEREE GE u a NEM Sasu UE ENX MR EORR 47 8 OUIDUC Re SUIS E l u dne n dnd nna RR REOR EROR HC IF ai 49 9 BOAMP Siina aAA ROAD REOR TR E ETC OD E ER 52 TO MC GNSS M 52 FI MOMMA GUS uyu uyu muyun E E annuus 54 12 PAE KNOWS CG GING ING uuu uuu u yu says 54 13 Selected FSV Validation Bibliography 56 1 Introduction to the 1D Feature Selective Validation FSV Theory The 1D Feature Selective Validation FSV Tool is a standalone application that implements the two dimensional 1D FSV theory The 1D FSV Tool is as a joint project between Applied Electromagnetics Group De Monfort Univ Leicester UK http www eng dmu ac uk aeg UAq EMC Laboratory Univ of L Aquila L Aquila Italy http www diel univaq it labs emc UMR EMC Laboratory Univ of Missouri Rolla Rolla USA http www emclab umr edu The ID F
9. comparison e GDM Is a figure of merit of the combination of ADM and GDM These output can be at the end of the algorithm e xDMi values of x A F G DM for each pair of samples of the two datasets to be compared An example of ADMi and FDM1 is shown in Fig 8 2 50 1D FSV 4 0 2L Results Data display Grade Spread chart Data Display E ADMI 12 Excellent 0 0 1 Excellent 0 0 1 Very Good 0 1 0 2 Very Good 0 1 0 2 Good 0 2 0 4 Good 0 2 0 4 ds E Fair 0 4 0 8 r 0 8 1 8 Poor 0 8 1 6 Very Poor 1 6 inf Very Poor 1 5 jnf 1 a c T 2 a E 200 150 200 Edit Figure Export Figure Edit Figure Export Figure Time Domain Analysis Time Domain Analysis ADMI ADMc ADMI ADMc FDMi FDMc FDMi O FDMc C GDMi GDMc C GDMi C GDMc Hidelegend Hide synthetic figures Hidelegend Hide synthetic figures Fig 8 2 An example of ADMi and FDMi e xDMc percentage of points in x A F G DM1 that fall in each of the six classes of Table 1 1 described in Chapter 1 see Fig 8 3 1D FSV 4 0 2L Results Data display Grade Spread chart Data Display FE C 0 35 Edit Figure Export Figure Edit Figure Export Figure Time Domain Analysis Time Domain Analysis O ADMi ADMc O ADMi ADM FDMi FDMc C FDMi FDMc C GDMi GDMc C GDMi GDMc Hidelegend Hide synthetic figures Hide legend Hide synthetic fig
10. s x 1 l 3 The K factors ranges from 0 to 1 they are related by the following constrain K 1 esas l l 4 phase E The GDM will combine the ADM and FDM without the inclusion of a separate weighting factor this has already been included in 1 13 and 1 14 GD M combined ADM combine x T FDM pine CX 1 l 5 1 2 1 K weighting factor 11 K magnitude Kphase are the weighting factor 0 Kimagnitude Kphase lt 1 representing the relative subjective importance placed on the two terms The selected values of K are also an important factor in communicating the comparison because it relies on the engineers involved in rationalizing their subjective decision In certain applications only the magnitude of the electromagnetic variable is required in which case setting Kmagnitude 1 Kphase 0 reduces these equations to those previously described In other applications both the magnitude and phase are equally as important For example a wrong equivalent circuit could result when there is perfect agreement in the magnitude but a small difference in the phase In this case it would be anticipated that Kmagnitude Kpnase 0 5 would be appropriate The FDM is treated in the same way using the same weighting factor The GDM will combine the ADM and FDM as before without inclusion of a separate weighting factor this has already been included in the component measures The output variables of the Combined a
11. the data given by the vectors X Y Matlab Code 6 4 ploEQX Y The vectors X and Y are now ready to be cast in the proper format for FSV 1D This 1s described in Section 6 1 1 3 6 1 1 2 Case 2 Set of data generated outside MATLAB Given the function y f x 6 5 Which has been evaluated computed measured etc by the user outside MATLAB 42 The user has to build the correct association between the values of x and y in MATLAB format in order to plot the data 1 First set the domain values by defining the limits for the x axis Xmin Xmax The samples along x can be uniformly spaced or non uniformly spaced e samples uniformly spaced the step is Ax for the x values The x vector that represents the domain are generated as X axis Matlab Code KEN Axix 6 6 e samples nonuniformly spaced the user has to manually enter the values for x axis X axis Matlab Code 6 7 X X 35 X In this case FSV 1D during import procedure sets automatically the Ax for the x values to the minimum step between the x samples Then recalculate the x domain values and interpolates the y values 2 The values of y f x can be now computed An example of computation of y x is 6 15 Y f m f x fG 6 8 e At this point one can plot or store the data given by the vectors X Y by using the command Matlab Code photo Y eg The vectors X and Y are now ready to be cast in the proper format for FSV 1D Th
12. the radio buttons of the Select chart selection box Fig 5 11 39 1D FSV 4 0 2L Results Data display Grade Spread chart Grade Spread Chart 1 select chart C Folder 1 chart C Folder 2 chart Points to plot ADM FDM GDM Grade Spread thereshold Threshold af 85 Fig 5 11 Window of the Grade Spread charts 1D FSV 4 0 2L Results Grade Spread Chart 1 threshold 85 Kadm 1 Kfdm 1 Grade Spread Chart select chart C Folder 2 chart Points to plot ADM FDM GDM Grade Spread thereshold Threshold af 85 3 6 Spread Fig 5 12 Selecting Grade Spread chart of Folder data Selecting one chart is possible to independently update the threshold or the points to be displayed 40 1D FSV 4 0 2L Results Data display Grade Spread chart Grade Spread Chart Grade Spread Chart 2 threshold 85 Kadm 1 Kfdm 1 select chart C Folder 1 chart Folder 2 chart Points to plot ADM FDM GDM Grade Spread thereshold Threshold 6 Fig 5 13 Selecting Grade Spread chart of Folder 2 data Spread 6 Input Data Structure 6 1 X Y ASCII format ID FSV can import data stored in two column ASCII file as in Fig 6 1 Fig 6 1 Structure of the of the Data file The supp
13. 1D FSV Tool User s Guide Version 4 December 2007 TABLES OF CONTENTS 1 Introduction to the 1D Feature Selective Validation FSV Theory 4 11 The Feature Selective Validation 1D FSV method 5 12 The Combined Analysis for complex values 11 12 1 IK WEIGMENG Td Pu uu 2 l A usu Sinum Sua ER 11 2 How iIpet l suyusussamssmaguykakusausqauwaskasussasqaqacsustaussasthusasyana EVER RE VRAE REERV ERR 12 21 Sysenm Regull merts 4 u l cos Eua uma vu Qua SERERE eV va va CER vnd oov ERR Edw EG uU uas 12 22 Jalan one don din n nac VR khi o ER Pind C dn 12 3 RE FSV Oleo essi cw u CREER ER RENE METER FR MERI MRKR NER VF FEEENUURKESNEEE VERRNNVRE 13 3 1 Initial Window and analysis 13 32 O utputdakr kiem ERG ina endeavor a a ERE HUS 16 33 dInputdata LOA NG uyu aa TU RE ERU M ER ON RR S E ER 16 SA IDFESVCOompUutalblOli icu ori eria Ka Wi an RO I KR a Ru B OR bl i x AR RN T 17 341 JNolconmbinaedidalgug uuu y L Dyus bendi uum ads iMd fadi dota ale 17 BAZ JC OmbDIDed data iiid t s eL OR UO UI UU Ed ford um d oae id d vts d ad Up idle a wa 18 A Inp utDSta Loe GING etr 21 4 1 Domain and data requirements T T T J T J T T 21 42 Loading X
14. ADM FDM GDM and reported on a GRADE SPREAD chart 1 1 The Feature Selective Validation 1D FSV method The structure of the ID FSV involves reading the two data files to be compared and interpolating them over common domain so that the data points to be compared are coincident This approach ensures that like is being compared with like and will not affect the overall results unless the data are severely under sampled It must be remembered that the purpose of the FSV is to mimic a visual comparison and so long as any interpolation does not produce visually different results this approach is perfectly acceptable The actual comparison is based on decomposing the original data into trend and feature information This Is done by applying 1D Fourier Transform to the data and to window the transformed data to separate out the lower and higher portions The high and low portions are then inverse transformed back into the original domain Combinations of these filtered data sets and their derivatives are used to compute the Amplitude Difference Measure ADM and the Feature Difference Measure FDM which can be combined into the Global Difference Measure GDM More specifically the procedure 1s l Read data obtain the overlap window and interpolate the data if necessary in the overlap region to ensure coincident pairs of data points This ensures that the two data sets to be compared have the same number of data points located at t
15. GUI 10 FS Inputs Data Summary Seles Input Data Summary SET 1 FORMAT X ASCII DOMAIN FILE magnitude txt DATA FILE magnitude txt X AXIS per unit SET 2 FORMAT X ASCII DOMAIN FILE magnitude txt DATA FILE magnitude txt X AXIS per unit SET 3 FORMAT X ASCII DOMAIN FILE phase txt DATA FILE phase txt X AXIS per unit SET 4 FORMAT Y ASCII DOMAIN FILE phased txt DATA FILE phase txt MAXIE per unit If the data are correct press Continue otherwise press Main Menu to abort the analysis and go back to the initial window Fig 4 15 Summary GUI in combined analysis and an expanded version of Scaling of domain s dataset GUI 29 5 The Data Display Tool 5 1 The Results GUI The Data Display Tool is an application embedded in the FSV Tool that allows the user to visualize the relevant FSV variables associated with the comparison of two datasets The Data Display Tool plots the data contained in the mat files generated by the FSV procedure The Data Display Tool allows the visualization of the two original data sets and of the FSV variables associated at the FSV analysis performed The three buttons on top of the GUI are e Main menu back to the FSV main window analysis Fig 3 1 e Data Display display the main results GUI panel e Grade Spread chart display the Grade Spread chart panel 5 1 1 The Data Display panel Input Dataset 1 Amplitude A ASCII X Y ASCII File Edit View
16. Y ASC format u III UU uY 22 43 Loading Y onlyASCII formatio eicunasasenuse acad uu RECE 0 62 adu uaC aasan a RECEu n UT uA as waa n aS REGE Edo day Rckr ww 23 44 Time domain analysis T T 23 4 5 Frequency domain analysss 25 151 Moagntideggeuu ous susu ua uka Su usu 25 452 PHI E 22 uu Tun zl a 2 mamantana a a Stan mib uuu uni aU UU CEN aeu etd 25 AD COMMING l n uyashka w uathuy uapa sia atap yakum uqapas h pm 25 4 6 Invalid input data loadingd 27 AF Summary GUIW 28 5 Whe Dalal Display IO01l 2 uuu un uyum Feu iukaakusayakhuywkasaywhuskawayhuywkakawkuspauyaasahasaqas 30 Sb Me ReWkSGU k i i apse u u cus annals Ga sas SS unu S asua 30 5 11 Ne Data Dil p nela am uay maia aus C OubRucEbs 30 PZ Crad e SS CVI unu uu d aerea t a patas ru es t df hu dc Da 33 5 2 Displaing the results at the end of an FSV analysss 34 5 3 Displaing the results from the initial window 35 55 De Ca DIS eV ttc s 38 S532 Grde SO CCV Uda u uuu Sau uu des ota ior tV Cn buta uuu qupa 39 6 anputDalta SOUC Ilu uuu a a qawaspas nage Waaa uksy 41 6 1
17. cision 51 Get The first argument of dlmwrite is the output filename the second argument is the data to be stored in the ASCII file then 1s specified the delimiter character 1n this case the tabular character and the resolution and format of the data This ASCII file inputdatal txt define completely one set in 2D space without a defined x domain Repeating entire procedure for the second set to be compared you obtain the other file inputdata2 txt At this point you can run FSV 1D and make a comparison of the two set using as input files inputdatal txt and inputdata2 txt 7 Output Data Structure amp Location The output data files txt fig and mat and the screenshots generated by the user are stored in an automatically generated subdirectory inside the selected directory for output file see Section 4 4 The names of these subdirectories are Frequency Domain Analysis x or Time Domain Analysis x in which x is an incremental number to distinguish the subdirectories 45 7 1 Data Structure The main level of an output data folder appears like in Fig 7 1 d Amplitude_ADMc fig 0 amplitude_ADMi fig 4 Amplitude DC inputdata fig Amplitude ADM average txt A amplitude_FDMc Fig Amplitude ADMc Ext 0 amplitude_FDMi Fig Amplitude ADM txt Amplitude_GDMc fig Amplitude OC impukdatal txt 4 amplitude_GDMi fig Amplitude OC inputdataz Ext 4 Amplitude High inputdata fig Amplitud
18. d amp t a nagnikude 1 txt magnitudes txt B phase 1 txt B phase txt Mome file magnitude1 txt Tipo file bt Annulla Fig 4 8 Window for magnitude files selection 1 2 Select the second file 1 e magnitude2 txt containing the data of component 2 of first combined set Select Combined Magnitude Data 2 Xx Y ASCII Mome file magnitude2 Ext Tipo file txt Annulla Fig 4 9 Window for magnitude files selection 2 3 Select the third file 1 e phasel txt containing the data of component of second combined set 26 Select Combined Phase Data 1 4 ASCII Cerca In 4 combined k magnikuide1 tok E magnitude txt Mame File phase tet Tipo file txt Annulla Fig 4 10 Window for phase files selection 1 4 Select the fourth file i e Phase2 txt containing the data of component 2 of second combined set Select Combined Phase Data 2 X Y ASCII Cerca in B combined FE E magnitude txt E magnitude txt phased bk _Iphasez txt Mame File phases tut Tipo file but Annulla Fig 4 11 Window for phase files selection 2 4 6 Invalid input data loading e When input data files selected contain invalid data or structure an error dialog is displayed 1D FS 4 0 2 ERROR EOR INVALID INPUT DATASET Error using gt DataLoad gt tsvascil INVALID DATA FILE SELECTED Fig 4 12 Error window for i
19. domain on x axis of N points In the non combined analysis two data sets are compared first of all ID FSV calculates the overlay window of the input data by finding lower and upper bound of the two domains If the intersection of two domains isn t empty the resulting domain values are interpolated using the minimum of the punctual steps of the two original domains But there are two checks to perform 1 Condition 1 Absolute minimum size In order to correctly build the Low and High filter the data size 1s lower limited by a minimum value that depends by the DC cut off radius The total minimum size is given by 2 DC I2 1 If DC 4 the minimum data size for each dimension is 31 points The error messages of the GUI are 1D FS 4 0 2L ERROR Error using gt Interpolation Can t apply 10 F5 on this dataset The synchronized domain has not enough points on Asis X to define the LOW HI Filter Array Fig 4 1 Error Message GUI for insufficient points on x axis 2 Condition 2 High filter realizability The breakpoint value cannot be greater than a certain value to allow the building of high filter The error message of the GUI is 21 1D FS 4 0 2L ERROR Error using gt Fev d cutalf Can t apply 10 FS on this dataset The Breakpoint is too close to the edge of the domain The HI Filter Array cannot be defined Fig 4 2 Error Message GUI for breakpoint greater than BK Only for the combined data
20. e Domain Amplitude analysis x threshold 85 e Greadspread_ x txt contains the values and the ranges of Grade and Spread for each FSV variable ADM FDM and GDM e Analysis mode GradeSpreadChart x fig figure of the Grade Spread chart e Analysis mode GradeSpread x mat contains the Grade Spread data in MAT format 7 3 Non combined data In this case the output folder appears like in Fig 7 1 and the Grade Spread subdirectories like in Fig 72 7 3 Combined data For the Combined analysis the output data are saved in subdirectories of the Frequency Domain Analysis directory named with the value of the Kynagnitude parameter used 47 At the top level of the output folder are saved the data files of the two components forming the combined set as described in Section 7 1 except the Grade Spread subdirectories that are saved in relative K subdirectories EK Factor 0 3 K fFactor 0 5 _K_Factor 0 8 Combined Domain tt Combined InputPaths ExE Magnitude ADM average txt Magnitude_ADMc txt Magnitude _ADMi txt Magnitude OC _inpukdakal txt Magnitude OC _inpuktdaktaz txt Magnitude FOM average txt Magnitude FDMc Ext Magnitude _FDMi txt Magnitude GDM average txt Magnitude DIMc Ext Magnitude GDMi txt Magnitude High inputdatal txt Magnitude High inputdata txt Magnitude _inputdatat txt Magnitude _inpukdakaz txt Magnitude
21. e _Domain txt Amplitude_inputdata fig Amplitude FOM average txt 0 amplitude_Low_inputdata fig Amplitude FOMc txt J amplitude _ODMi Fig Amplitude _FDMI Ext E Amplitude Analysis mat E Amplitude GDM average txt Amplitude _GDPE Ext Z Amplitude _SDMi txt Amplitude High inputdatad txt E Amplitude High inputdataz txt Amplitude inputdatal txt Amplitude _inputdataz txt Amplitude InputkPaths ExE Amplitude _Low inpukdatal Ext Amplitude Low _inputdataz txt Z Amplitude ODMi txt Fig 7 1 Output data folder for Time Domain Amplitude analysis All data files are saved In the main level of the output folder except in the combined analysis mode Grade Spread data files are saved in appropriate subdirectories their name is given by GSthrs x x suffix is the threshold value used The txt files are ASCII files e Analysis mode inputdata contains the input data after synchronization e Analysis mode DC inputdata contains the DC Data used in the calculation of ADM FDM and GDM e Analysis mode Low inputdata contains the Low Data used in the calculation of ADM FDM and GDM e Analysis mode High inputdata contains the High Data used in the calculation of ADM FDM and GDM e Analysis mode x DMi contains the point by point values of FSV output e Analysis mode ODMi contains the point by point values of FSV output of offset components e Analysis mode
22. e domain is used a correction for left edge of the domain ag Lo i 1 Lo i dx l A where i represents the element of the vector and A 1s equal to 1 for right edge of the domain E w Lo i Lo i 1 X A for all internal point 29 Lo i 1 Lo i l dx A Calculate the point by point FDM The FDM 1s formed from three parts based on the derivatives calculated in step 9 Lo x Los x FD x 2 max 1 6 AL x Lo x 2 Hi x His Go a rb g Eio 72 lt lt 1 8 vs yr Del tin e His G9 e min FDMi 2 FD FD FD 1 9 Being min and max the lowest and highest components x in the data sets 11 Calculate the single value of FDM This is done in exactly the same way as for the ADM 12 Calculate the FDM confidence histogram This is done in exactly the same way as was done for the ADM 13 Obtain the point by point GDM value The GDM is premised on the ADM and FDM being largely independent which means that dans GDM 2 K apMi_ cm ADM x KypMi cm FDM x 1 10 the weighting coefficients kapy cm and Krpy cm as at point 20 14 Calculate the overall GDM value and the GDM confidence histogram This follows the same procedure as the ADM and FDM 15 Determine the synthetic figures of merit ADM FDM GDM These variables are the average of the point by point values of ADM FDM and GDM respectively 16 Determine the synthetic figur
23. eature Selective Validation FSV algorithm has been developed to compare two sets of bidimensional surface data not necessarily in the electrical engineering field and put them in an objective and comprehensible form Several motivations form the basis of FSV e The need to control variations between visual assessment results e The reduction of cost a skilled engineer is an expensive commodity e The desire to reduce ambiguities e The inability of humans to process and cache extremely large volumes of data The FSV theory was conceived as a technique to quantify the comparison of data sets by mirroring engineers visual perceptions Furthermore FSV allows automated comparisons of large volumes of complex data whilst reliably categorising the results into a common set of quality bands The FSV offers three figures of merit of the comparison of two data sets e ADM Amplitude Difference Measure and FDM Feature Difference Measure These are available as numerical values and can be converted to a natural language descriptor in a six level scale excellent very good good fair poor Very Poor These combine to give the GDM e GDM Global Difference Measure An overall single figure goodness of fit between the two data sets being compared This allows a simple decision to be made about the quality of a comparison This may be numerical or converted to a natural language descriptor These figures of merit can be further represented i
24. er and pushing the Update button The new values of GRADE and SPREAD are computed and the variables can be displayed on the chart 5 2 Displaing the results at the end of an FSV analysis At the end of an FSV analysis it is possible to visualize the results of the last FSV analysis performed by selecting view results in the end analysis GUI Fig 3 8 The Data Display Tool starts and automatically loads the data of last FSV analysis performed For combined analysis it loads the results relating the last K weighting factor used 34 Input Dataset 1 Amplitude A ASCII X Y ASCII File Edit View Insert Tools Desktop Window Help Dae b amp m 09 x9 sm Amplitude Input Data 1 5 1 45 4 Amplitude e e EM ME TE m hJ m Lu 1D FSV 4 0 2L Results Data display Grade Spread chart Data Display EFNI Excellent 0 0 1 Very Good 0 1 0 2 Good 0 2 0 4 Fair 0 4 0 8 Poor 0 8 1 6 Very Poor 1 5 int Amplitude 150 200 Edit Figure Export Figure Time Domain Analysis ADMi ADMcC FDMi FDMc GDMi Hidelegend Hide synthetic figures Edit Figure Export Figure Time Domain Analysis ADMI ADMc GDMi Hidelegend Hide synthetic figures Fig 5 5 The original data window and main Data Display Tool of last analysis performed 5 3 Displaing the results from the initial window 35 The Data Display
25. es of merit ADM cong FDM cong GDMcong These values are computed by weighting the number of samples of the point by point corresponding variables falling in the six classes of Table 1 1 with the associated weight of the visual six points scale in the same Table xDM 1 EX 2 VG 3 G 4 F 5 P 6 EP withx A F G 1 11 conf where is the number of elements belonging to a class The total value 1s then normalized to the length of the Low High array 17 Determine the equivalent visual scale values ADM FDMpy and GDM The FSV values can be scaled to the Visual six point scale in Table 1 1 The piece wise visual conversion for this is given in Table 1 2 where y is ADM on FDM jo or GDM io Table 1 2 Piece wise visual conversion If y lt 0 1 Then V 1 10y If y gt 0 1 and y lt 0 2 Then V 2 10 y 0 099 If y gt 0 2 and y lt 0 4 Then V 3 5 y 0 199 If y gt 0 4 and y lt 0 8 Then V 4 2 5 y 0 399 If y gt 0 8 and y lt 1 6 Then V 5 1 25 y 0 799 If y gt 1 6 Then V 6 The piece wise conversion approach is represented by the graph in Fig 1 1 Visual Rating Visual Rating EP Range of the classes X A F G D Mtot Fig 1 1 Piece wise visual conversion graph 17 Determine the GRADE value and range The GRADE value is computed by taking the number of classes starting from the best Excellent to the worst Extremely Poor which include a user defined amo
26. he range of values for the ADM and In fact the FDM and GDM can be divided into six categories each with a natural language descriptor Excellent Very Good Good Fair Poor And Very Poor These are the terms that are most often used in descriptions of the quality of comparisons The confidence histogram like a probability density function provides some intelligence as to how much emphasis can be placed on the single figure of merit There is some evidence to show that this mirrors the overall group assessment of any data pair by a number of engineers The determination of the histogram is simply a case of counting the proportion of points that fall into one of the categories according to the rule base in Table 1 1 Table 1 1 FSV INTERPRETATION SCALE FSV value FSV interpretation FSV Visual six quantitative qualitative point scale Less than 0 1 Between 0 1 and 0 2 Very good Greater than 1 6 Extremely Poor Calculate derivatives in preparation for the FDM calculation The following components need to be calculated The first derivatives of the Lo x and Hi x data sets and the second derivatives of the Hi x data sets The derivatives accentuate the high rate of change features in the original data and differences based on the derivatives are combined in the determination of the FDM The first and second derivatives are currently obtained by a simple difference approach Lo 1 Lo it1 Lo 1 1 At the edges of th
27. he same positions on the independent x axes 2 Fourier Transform both data sets Depending on the number of samples a Fast or Discrete two dimensional transformation is used 3 Calculate the low data sets using the transformed data a Ignore the first four data points in the transformed set in order to avoid DC and very low frequency components and sum the intensities of the remaining data b Obtain a 40 location by summing the data from the DC 1 point i e ignoring the near DC data until the total reaches 40 of the total value calculated in step 3a The 40 location used by the FSV is the lowest of the two resulting numbers from the two original data sets A break point five data points above this is returned a value that allows a comfortable transition window between the low and the high results c Window the transformed data for both data sets by taking a linearly decreasing envelope from two points below the break point to two points above it Essentially low pass filtering the transformed data d Inverse transform the windowed data to give the low region data for both original data sets they are named Lo x y and Lo2 x y Calculate the high data sets using the transformed data Repeat the process from 3c by applying the opposite envelope to the transformed data essentially high pass filtering it These data 1s then inverse transformed to give the high region data for both of the o
28. iates one of the six natural language descriptor categories to each range of values ADMc FDMc GDMc This 1s a graph with six vertical bars one for each six natural language descriptor categories In the top of graph is displayed xDMtot xDMconf xDMpw values x A F G in a yellow textboxes see Section 1 1 5 1 2 Grade Spread chart By selecting the Grade Spread chart button the FSV Grade Spread chart appears see Section 1 1 All variables are displayed on the chart for the default threshold of 85 33 1D FSV 4 0 2L Results Grade Spread Chart Grade Spread Chart 1 threshold 85 Kadm 1 Kfdm 1 Points to plot ADM FDM GDM Grade Spread thereshold Threshold 6 Fig 5 4 The GRADE SPREAD coloured chart Spread e The value of threshold currently used and the kapu and krpy used in forming GDM eq 1 11 are shown above the GRADE SPREAD chart e Export Figure button take a screenshot of the currently graph displayed The displayed graph is exported in PNG format The figure is saved in a file located in the subdirectory inside the selected directory for output data named with the value of the threshold e Points to plot checkboxes select the GRADE SPREAD points to plot from ADM FDM and GDM e Grade Spread threshold box from this section the threshold can be modified entering the new value in the appropriate text field or moving the slid
29. is is described in Section 6 1 1 3 6 1 1 3 Building the proper data format for FSV 1D Assume that the vectors X and Y have been defined as in the previous sections Now they should be cast for proper FSV 1D input FSV 1D requires for each one of the two set that should be compared one input file This file contains the domain x and the y values The structure of this file 1s described in the following figure 43 XI Yi Xo Y XN YN Fig 6 2 Structure of the of the Data file The X and Y domain values are placed side by side and then stored in ASCII format This file is created by the following MATLAB command that writes data into an ASCII file Matlab Code dimwrite iampurdatal ter x yl delim ter O 6 10 Att precision 1 88 The first argument of d1mwrite is the output filename the second argument is the data to be stored in the ASCII file then is specified the delimiter character in this case the tabular character and the resolution and format of the data This ASCII file inputdatal txt define completely one set in 2D space Repeating entire procedure for the second set to be compared you obtain the other file inputdata2 txt At this point you can run FSV 1D and make a comparison of the two set using as input files inputdatal txt and inputdata2 txt 6 2 Y only ASCII Format ID FSV can import data stored in one column ASCII file as in Errore L origine riferimento non stata trova
30. le Edit View Insert Tools Desktop Window Help Da Ee kaana El mE so ADMi Excellent D 0 1 very Good 0 1 0 2 Sood 0 2 0 4 Fair 0 4 0 8 Poor 0 6 1 6 very Poor 1 6 inf m a r Fig 5 2 ADM in the Edit Figure GUI Select the radio buttons on the bottom left frame to display one variable ADMI ADMc etc After this selection the radio buttons on the opposite bottom right frame are enabled for selection of another variable to be displayed 32 1D FSV 4 0 2L Results Data display Grade Spread chart ADMI Data Display Excellent 0 0 1 Very Good 0 1 0 2 Good 0 2 0 4 Fair 0 4 0 8 Poor 0 81 6 Very Poor 1 6 inf o k a 150 200 Edit Figure Export Figure Edit Fiqure Export Figure Time Domain Analysis Time Domain Analysis ADMI FDMi FDMc FDMi FDMc GDMi GDMc GDMIi GDMc Hidelegend Hide synthetic figures Hidelegend Hide synthetic figures Fig 5 3 Data Display Tool with graphs displayed ADMi FDMi GDMi At the left of each 3D graph ADMi FDMi GDM a colorbar is displayed The bottom of the color scale 1s associated with 0 Excellent values and the top with 1 6 Extremely poor values Then the labels of the colorbar are 0 Excellent 0 1 Very Good 0 2 Good 0 4 Fair 0 8 Poor and 1 6 Extremely Poor Moreover inside the graph a legend is displayed that assoc
31. mination 1977 Surface Science 62 pp 61 80 van Hove MA Tong SY and Elconin MH Surface Structure Refinements of 2H MoS 2H NbSe and W 100 p 2x1 O via new Reliability Factors for Surface Crystallography 1977 Surface Science 64 1 pp 85 95 A Martin 1999 Feature Selective Validation Thesis for Doctor of Philosophy De Montfort University Leicester G Antonini A Ciccomancini Scogna A Orlandi C Ritota A Duffy Applications of FSV to EMC and SI data IEEE Int Symp on EMC Chicago 2005 B Archambeault S Connor Comparing FSV and Human Responses to Data Comparisons IEEE Int Symp on EMC Chicago 2005 AJM Martin A Ruddle A Duffy Comparison of Measured and Computed Local Electric Field Distributions due to Vehicle Mounted Antennas using 1D FSV A Duffy A Martin G Antonini A Orlandi and C Ritota The feature selective validation FSV method in Proc of IEEE 2005 EMC Int Symp Chicago USA 8 12 August 2005 B Archambeault and S Connor Comparing FSV and human responses to data comparisons in Proc of IEEE 2005 EMC Int Symp Chicago USA 8 12 August 2005 56
32. n three different ways in order to quantify the quality of the comparison performed e GDMi ADM and FDM These are point by point comparisons of the amplitude differences the feature differences and the global differences This allows a user to analyze the resulting data in some detail probably with the aim of understanding the origin of the contributors to poor comparisons e GDM ADM FDM These give probability density functions which show the proportion of the point by point analyses of each of the components that falls into the six natural language descriptor categories This provides a measure of confidence in the single figure comparisons e GDMo ADM FDMtct GDM r ADMoont FOMcont GDMpw ADMpw FDM pw These are more synthetic figures of merits of the comparison and stem from an elaboration of the variables described in the previous points They are described in the next Chapter Based on these figures of merit the comparison of two data sets can be ranked The ranking useful for making a selection between multiple comparisons is given by considering two quality factors for each figure of merit The GRADE and the SPREAD e The GRADE is a direct indication of the quality of the comparison The smaller it 1s the better the comparison e The SPREAD indicates the level of reliability of the outputs The smaller it 1s the higher 1s the reliability of the results e GRADE and SPREAD are computed for each figure of merit
33. nalysis are the same than those illustrated in points 12 to 20 of the previous section Moreover the user can select more than one pair of K weighting factors for each analysis as is described in 3 4 2 1 Based on this algorithm the FSV Tool has been developed as a standalone application for Windows OS 2 How to Install 2 1 System Requirements The 1D FSV Tool is developed to run under WINDOWS 2000 or Windows XP It has not yet been tested on WINDOWS VISTA 2 2 Installation The 1D FSVxxx zip file xxx indicates the number of the version contains all the needed parts of the 1D FSV Tool Method 1 users that have a version of MATLAB installed 1 Check if a MATLAB compatible version is installed in your system type in matlab prompt version if the output is 7 1 0 246 R14 Service Pack 3 your machine is able to run ID FSV Otherwise see the Method 2 2 Unzip the content of the zip file ID FSVxxx zip in the installation directory i e ID FSV and double click on fsv D exe 3 In the zip package it is also provided an icon file 1co to create a desktop shortcut for the fsvI D exe file according to the standard Windows procedure 12 Method 2 users that do not have MATLAB installed or an earlier MATLAB version than from 7 1 0 To run 1D FSV to another development machine that does not have MATLAB 7 1 0 installed including a machine that has MATLAB but it is a different version of MATLAB 7 1 0 the
34. nvalid input data loading zy e When in the combined analysis the domain of each Magnitude and Phase parts of a given combined set is not identical in every sample this error dialog is displayed 1D FSV 4 0 2 ERROR EOIR Error using gt Interpolation Combined domain mismatch Fig 4 13 Error window for Combined domain mismatch After an error data loading occurs FSV stop the current task and returns back to the Main menu 3 1 4 7 Summary GUI At the end of the data loading procedure it is displayed a summary of the loaded data and domain s settings 1D FSV Inputs Data Summary Input Data Summary SET 1 FORMAT X ASCII DOMAIN FILE chartza new txt DATA FILE chart2a nieve txt X AXIS per unit SET 2 FORMAT X ASCII DOMAIN FILE chart2b news txt DATA FILE chart2b neve txt X AXIS per unit If the data are correct press Continue otherwise press Main Menu to abort the analysis and qo back to the initial window Fig 4 14 Summary GUI For each data set it 1s specified 28 FORMAT the format of input data DOMAIN FILE name of the file that containing the domain DATA FILE name of the file that containing the data X AXIS dimensional unit set for the X axis At this point the user can start the analysis pressing Continue or to go back to the initial window pressing Main Menu Combined analysis summary For the combined data analysis is displayed an expanded version of summary
35. orted delimiter between two columns are tabular character space and comma Also the exponential format 1s supported If the data isn t available in this format then is needed a conversion how 1s described in the section 6 1 1 6 1 1 Conversion to X Y ASCII format The aim of this section is to describe the X Y ASCII format of the input data for the use of FSV 1D from ver 4 0 0 on and to show how to generate this format by using MATLAB The 41 Two cases are considered e Case 1 the function y f x 1s computed directly in MATLAB e Case 2 function y f x is generated computed measured etc outside MATLAB In both cases MATLAB is used to recast the data in a proper format for input to FSV ID as described In Section 6 1 1 3 6 1 1 1 Case 1 Set of data computed directly in MATLAB Using MATLAB generate a curve y y f x 6 1 The first task is to define the range of the domain along the x directions and then to compute f x at these points Three steps are required 1 Set the limit values for the x axis Xmin Xmax Then define the incremental steps for x as Ax 2 Generate the x vector that represent the domain of the y f x function X axis Matlab Code gt X X nin AN A ae 6 The command lines 6 2 generate a row vector x from Xmin tO Xmax step Ax 3 The values of y f x can be now computed An example of computation of y x is Matlab Code iS Ke Q9 At this point one can plot or store
36. riginal data sets They are named Hi x y and Hi x y Calculate the DC data sets using the transformed data The first four data points in the transformed set in step 3a are inverse transformed and give the DC data for both of the original data sets They are named DC and DC Calculate the ADM on a point by point basis Each data set has N y points identified by the Cartesian coordinates xi 1 l N For sake of simplicity in the next it will be used the notation xi 1 At an arbitrary data point 1 the ADM is evaluated as in the following equation 1 1 This point by point variable is also abbreviated as ADM ADM i ADM c ODM i e OPM 1 1 where Lo G Lo G ADM i N l 1 2 xy 2 Lo i Lo i 1 2 i l DC i DC i amo ee l 1 3 pe Dc a DC i l Cm and Cm are weighting coefficients set equal to 1 1 4 ODM 1 represents the contribution to ADMi 1 of the difference of offset between the two original signals Calculate the single value of ADM A mean value of the ADM xi ADM 1 gives an overall single figure goodness of fit It is obtained from the following equation Y ADM i 1 5 Note a median value rather than a mean value has demonstrated some improvements in agreement with visual interpretations for 1D data although this has yet to be determined for 1D 8 10 Calculate the ADM confidence histogram T
37. ry for output file see Section 3 2 The names of these subdirectories are Frequency Domain Analysis x or Time Domain Analysis x in which x is an incremental number to distinguish the subdirectories 36 Sfoglia per cartelle Select Folder Far analyzed data 1 C Time Domain Analysis 57 i Time Domain Analysis 58 zJ Time Domain Analysis 59 j Time Domain Analysis 60 C Time Domain Analysis 61 zJ Time Domain Analysis 62 E F MEET Analvsis 63 73 asthrs 50 3 GSthrs 70 r lc gt Crea nuova cartella OK Annulla Fig 5 7 Selecting a Time Domain analysis output folder e combined analysis select the automatically generated subdirectory Frequency Domain Analysis x inside the selected directory for output file see Section 3 2 and then select the subdirectory K factor x named with the value of the Kynagnitude parameter used Sfoglia per cartelle Select Folder Far analyzed data 1 C3 Frequency Domain Analvsi Frequency Domain Analvsi 3 E Frequency Domain Analvsi E Co MU Gsthrs 85 K Factar 5 E Frequency Domain Analvsi k Factor n 4 7 E Factor 0 5 r _ kS l m Crea nuova cartella OK Annulla Fig 5 8 Selecting a combined Frequency Domain analysis output folder If the mat file is not found in the selected directory or an invalid mat file is found a warning is issued 37 Warning Mo valid mat file found in selec
38. sPIRED Fig 10 2 Trial expired GUI Pressing OK is displayed the License GUI dialog and from here is possible to load the license file To obtain a license contact us at orlandi ing univaq it and send the following informations e Your Name e Organization name e The authorization code that is displayed in the specific text box Use the key combination Ctrl C to copy it in the clipboard When you receive the license file you have to load it from the License GUI dialog Pressing Load license button is displayed the following GUI 53 Select the license file Cerca In 4 UA licenser libo U9 licenser SY Carlo_Polisini_1b FS li Uagproducts txt Nome fle Carlo_Polisini_1D FSV _license txt Tipo file tat Annulla Fig 10 3 License loading confirmation dialog From here browse to the license file and select it If the license is valid you get a confirmation message like this License Message Saks THE LICENSE IS SUCCESFULLY LOADED THIS LICENSE IS VALID SINCE 18 Dec 2007 TO 15 Dec 2006 Fig 10 4 License loading confirmation dialog 11 Contacts For any problem or comment please contact Dr A Duffy DMU Applied Electromagnetics Group apd dmu ac uk Prof A Orlandi UAq EMC Laboratory orlandi ing univaq it If you wish to be included in the FSV User List please e mail a short request to Prof A Orlandi UAq EMC Laboratory orlandi ing univaq it For downloading
39. sis x or Frequency Domain Analysis x in which the output files will be stored see Chapter 7 x is an incremental number to distinguish the subdirectories 3 3 Input data loading After that FSV requests the input data files to analyze according the analysis mode chosen in initial window FSV analyzes two set of data at a time in a normal computation and four set of combined data 1 e a pair of Magnitude Phase or real imaginary set of data in a combined analysis TIME DOMAIN e Amplitude Two set of data at a time are analyzed Amplitude 1 Amplitude 2 FREQUENCY DOMAIN e Magnitude two set of data at a time are analyzed Magnitude 1 Magnitude 2 e Phase two set of data at a time are analyzed Phase I Phase 2 e Combined four set of data at time are analyzed at a time Each dataset 1s given by two components Magnitude and Phase Magnitude 1 and Phase 1 Magnitude 2 and Phase 2 16 As described in section 1 2 1D FSV performs the comparison between the corresponding component 1 Magnitude 1 Magnitude 2 2 Phase I Phase 2 For each set of data FSV requests to specify the input format of the datasets by the GUI of Fig 3 5 At the present time they are three X Y ASCII and Y only ASCII see Chapter 6 1D FSV 4 f 1D FSV 4 f Select format for set 1 Select format for set 2 X Y ASCII X Y ASC Y only ASCII Y only ASCII Fig 3 5 GUIs select input format for set 1 lef
40. t and set 2 right See Chapter 4 for detailed data loading procedure to follow for each analysis mode 3 4 1D FSV Computation 3 4 1 Non combined data The FSV Tool performs its calculations in the background 53055 1D FSV 4 0 2L Elaborating data 1 of 1 SU 0 02 40 G sgs s i a _ Started at 13 41 57 cpu time used 0 00 07 Fig 3 6 1D FSV progress bar Once this has been completed the user is prompted to use or decline the GDM weighting procedure outlined in point 20 of the Section 1 1 or not Fig 3 7 Window for selecting the GDM weighting 17 At the end of the calculation a new question 1s issued as in the next figure 1D FSV Analysis Ended Sid Do vou want Fig 3 8 Window at the end of a complete 1D FSV analysis of a pair of data sets If it is answered Load New Data then the 1D FSV Tool back to Initial Window Fig 3 1 and 1s prepared for a new analysis If it is answered View Results the figures of the results are shown see Section 5 2 3 4 2 Combined data The FSV Tool performs its calculations in background _ 35 1D FS 4 0 21 SHE Elaborating data 1 of 2 35 0 01 31 a rcl Started at 13 45 43 cpu time used 0 00 02 Fig 3 9 Progress bar of combined analysis In the case of combined analysis for each component comparison FSV calculates ADM FDM and GDM separately and for each one is issued to use the GDM weighting procedure Compute K ADM amp k
41. t the second file 1 e chart2b_new txt containing the second set as in Fig 4 7 24 Select Amplitude Data 2 amp Y ASCII Mame file ichart2b_new tet Tipo file txt Annulla Fig 4 7 Window for data files selection 1 4 5 Frequency domain analysis Again in the following examples the input data files are in X Y ASCII format 4 5 1 Magnitude In the FREQUENCY DOMAIN Magnitude FSV Tool requests two input datasets to compare Basically you can follow the same procedure as seen in TIME DOMAIN Amplitude analysis 1 Select the first file 1 e chart2a_new txt containing the first magnitude set 2 Select the second file 1 e chart2b_new txt containing the second magnitude set 4 5 2 Phase In the FREQUENCY DOMAIN FSV Tool requests two input datasets to compare 3 Select the first file 1 e chart2a_new txt containing the first phase set 4 Select the second file 1 e chart2b_new txt containing the second phase set 4 5 3 Combined In the FREQUENCY DOMAIN Combined analysis FSV Tool requests two combined input sets to compare 1 Select the directory in which your data are located For test purposes the data examples combined directory that is located in same folder of the main program fsv D exe can be used Select the first file 1 e magnitudel txt containing the data of component 1 of first combined set 29 Select Combined Magnitude Data 1 X Y ASCII Cerca In C3 combine
42. ta Fig 6 3 Structure of the of the Data file The supported delimiter between two columns are tabular character space and comma Also the exponential format 1s supported If the data isn t available in this format then is needed a conversion how is described in the section 6 2 1 6 2 1 Conversion to Y only ASCII Format The aim of this section is to describe the Y only ASCII format of the input data for the use of FSV 1D from ver 4 0 0 on and to show how to generate this format by using MATLAB Two cases are considered 44 e Case 1 the data y is computed directly in MATLAB e Case 2 the data yis generated computed measured etc outside MATLAB In the case 1 the data are ready to be stored in Y only ASCII format In the case 2 it s possible to recast the data in this way Y yy 6 11 6 2 1 1 Building the proper data format for FSV 1D Assume that the vector Y has been defined as In the previous sections Now they should be cast for proper FSV 1D input FSV 1D requires for each one of the two set that should be compared one input file This file contains the y values The structure of this file is described in the following figure Fig 6 4 Structure of the of the Data file TheY values are stored in ASCII format This file Is created by the following MATLAB command that writes data into an ASCII file Matlab Code dlmwrite inputdatal txt y delimiter o 6 12 WET pre
43. ted folder Fig 5 9 Data Display Tool warning 5 3 1 Data Display In the following example a Time Domain analysis in Folder 1 and a Magnitude Frequency Domain analysis in Folder 2 are shown Fig 5 10 Input Dataset 1 Folder 1 File Edit view Insert Tools Desktop Window Help Dki S b S osx Js Amplitude Input Data Amplitude Input Dataset 1 Folder 7 Sele File Edit View Insert Tools Desktop Window Help Da E e 5 m 0 9 xD sm Magnitude Input Data Magnitude 38 1D FSY 4 0 2L Results Data display Grade Spread chart Data Display ADMI Excellent 0 0 1 Very Good 0 1 0 2 Good 0 2 0 4 Fair 0 4 0 8 Poor 0 8 1 6 Very Poor 1 6 inf o 100 150 200 X m Edit Figure Export Figure Time Domain Analysis ADMi amp DMc Load data O FDMi O FDMc Folder 1 C GDMi GDMc Hidelegend Hide synthetic figures Magnitude Folder 2 Excellent 0 0 1 Very Good 0 1 0 2 Good 0 2 0 4 Fair 0 4 0 8 Poor 0 8 1 5 Very Poor 1 6 inf 100 150 200 X m Edit Figure Export Figure Frequency Domain Analysis ADMI ADMc C FDMi FDMc C GDMi GDMc Hidelegend Hide synthetic figures Fig 5 10 Data Display Tool after selecting two folder 5 3 2 Grade Spread chart Selecting the Grade Spread chart button the user has the option to select which output data should be considered for the GRADE SPREAD chart from
44. the most recent version please frequently check the UAq EMC Laboratory FTP site at http ing univaq it uaqemc More information on the FSV project can be found at http www eng cse dmu ac uk FS Vweb 12 Acknowledgement 54 A special thanks to Ing Carlo Polisini Ing Carmine Ritota Ing Franco Campitelli and of the UAg EMC Laboratory for their substantial contribution to this project 55 13 Selected FSV Validation Bibliography 10 11 12 13 14 Johnson J and Picton P Concepts in Artificial Intelligence Vol II Butterworth Heinneman London Hilsenrath OA and Zeevi YY 1990 Feature Extraction and Sensitivity Matching in Visual Search in Man and Machine in Brogan D ed Visual Search Taylor and Francis London Koffa K 1935 Principles of Gestalt Psychology Kegan Paul London Cook WM 1931 Ability of Children in Colour Discrimination Child Development 2 pp 303 Duffy AP Woolfson MS and Benson TM 1994 The use of Correlation Functions to Assist the Experimental Validation of Numerical Modelling Techniques Microwave and Optical Technology Letters 7 8 pp 361 364 Woolfson MS Benson TM Christopoulos C and Duffy AP 1995 Quantitative Assessment of the Comparison of Electromagnetic Calculations with Experimental Data Applied Computational Electromagnetics Society Newsletter 1 1 pp 34 39 Zanazzi E and Jona F Reliability Factor for Surface Structure Deter
45. ues e FREQUENCY DOMAIN frequency domain analysis is performed In this case FSV requests the type of frequency analysis to perform by the following dialog window 1D FSV 4 0 X Frequency domain analysis Magnitude Fig 3 3 Frequency analysis menu o Magnitude the data input are magnitude values Phase the data input are phase values o Combined the 1D FSV combined analysis performs two FSV analysis magnitude and phase and combines them in a unique result O Note though in the TIME DOMAIN Amplitude analysis and in the FREQUENCY DOMAIN magnitude and phase analysis 1D FSV applies an identical algorithm the tool requests to specify the 15 nature of the input data to be specified in order to use an appropriate terminology and graphic presentation 3 2 Output data folder Push the button Run FSV Now FSV Tool requests the directory to select where the output data files will be saved Sfoglia per cartelle Select Folder For OLITPLIT Data Desktop zJ Documenti E 4 Risorse del computer w Disco locale C E3 a Unita DVD RAM D isco locale E Set Disco locale F E GA Unit DVD lt Documenti condivisi C Documenti carlo re s LES hl Crea nuova cartella Annulla Fig 3 4 Window for selecting the destination directory In the selected destination directory the FSV Tool will create a subdirectory named Time Domain Analy
46. unt named threshold and set at 85 by default of the total samples of the data sets to be compared GRADE value ranges from 1 best quality to 6 worst quality The GRADE range 1s given by considering the range of the classes included in the GRADE value computation 18 Determine the SPREAD value and range The SPREAD is computed by taking the number of classes starting from the most populated to the lowest one which includes a user defined amount named threshold and set at 85 by default of the total samples of the data sets to be compared SPREAD ranges from 1 best quality to 6 worst quality The SPREAD range is given by considering the range of the classes included in the SPREAD value computation 19 Plot the GRADE SPREAD chart Each figure of merit ADM FDM GDM has associated with it a GRADE SPREAD pair This pair is plotted on the GRADE SPREAD plane named GRADE SPREAD chart Fig 1 2 The chart has colored regions different colors indicate different quality of the comparisons Grade Spread Chart Grade Spread Chart 1 thresholid 85 Kadm 1 Kfdm 1 Points to plot ADM FDM Grade GDM Grade Spread thereshold Threshold 1 2 3 4 5 B Spread Export Figure Fig 1 2 The GRADE SPREAD coloured chart 20 Compute kApy cm and krpm cm 10 In forming GDM in 1 11 the relative weight of ADM and FDM depends on their level of confidence or reliabilit
47. ures Fig 8 3 An example of ADMc and FDMc 51 The values of xDMc are used to compute the GRADE and SPREAD values and displayed in form of the GRADE SPREAD chart 1D FSV 4 0 2L Results Data display Grade Spread chart Grade Spread Chart Grade Spread Chart 1 threshold 85 Kadm 1 Kfdm 1 Points to plot ADM FDM GDM Grade Spread thereshold Threshold 6 Fig 8 4 Grade Spread chart Spread 9 Examples In the directory data examples there are sets of data two files for each set that can be used for the FSV algorithm by means of the FSV Tool 10 License This software is distributed with a trial period of 30 days at the end of this the user can request a license with a time limited or unlimited validity to continue the use of the software By clicking on License button in Main menu or when the trial license period is expired is displayed the GUI of Fig 10 1 52 FSV license Foy License To obtain an FSV license please send Tour Hame Organization The Authorization Code that are displayed below to orlandig ing univag it Authorization Code Remaining trial days 27 Fig 10 1 License GUI dialog In this GUI is shown a remaining trial days and the instructions to obtain an valid license When the trial time is expired is displayed this error message at FSV start FSV Message TRIAL PERIOD Ee
48. users must install the MCR Matlab Component Runtime library ver 7 3 if it is not already installed on the user machine 1 Get the package MCRInstaller exe from http ing univaq it uaqgemc 1D FSV 2 0 0 that is the MCR ver 7 3 bundled with MATLAB 7 1 0 246 R14 SP3 2 Run MCRInstaller exe once on the target machine that is the machine where you want to run the application The MCRInstaller opens a command window and begins preparation for the installation Then the MCR Installer wizard appears click Next to begin the installation and follow the instructions on the GUI 3 Unzip the content of the zip file 1D FSVxxx zip in the installation directory 1 e 1D FSV and double click on fsv D exe 4 In the zip package it 1s also provided an icon file 1co to create a desktop shortcut for the fsvI D exe file according to the standard Windows procedure Note for Windows Users You must have administrative privileges to install the MCR on a target machine since it modifies both the system registry and the system path Running the 1D FSV after the MCR has been set up on the target machine requires only user level privileges 3 Run FSV Tool 3 1 Initial Window and analysis Run the FSV Tool from the exe in the installation directory or double click on the desktop icon When FSV Tool is started an initial window appears see Fig 3 1 The user may choose any of the major functions desired 13 1D FSV 4 0 2L FS
49. utdataz txt E inpubdakavanlv ExE Mome file linputdatayonly tet Tipo file txt Annulla Fig 4 4 Window for data files selection The same sequence of windows is proposed for the loading of the second set of data 4 4 Time domain analysis 23 1D FSV 4 0 2L FSV PROCEDURE UAg EMC Laboratory DMU Applied Electromagnetic Group FREQUENCY DOMAIN UMR EMC Laboratory NOTE TO THE USER Run FSv All the figures fig and ASCII files txt are generated in the working directory The figures are not displayed on the screen ASCII files should be read by spreadsheet Plot Results Read Me programs Le WordPad to not altering the tabulated format License Fig 4 5 Selecting the Time Domain analysis In the TIME DOMAIN Amplitude analysis after the Output data folder is set FSV Tool requests the input data In this example the input data files are in MATLAB FSV ASCII format l Fig 4 6 Select the directory in which your data are located For test purposes it can be used the data_examples survey_data 2 directory that 1s located in same folder of the main program fsvID exe Select the file 1 e chart2a_new txt containing the data and domain set as in Fig 4 6 Select Amplitude Data 1 amp Y ASCII Cerca irr 3 2 charbza new Ext B chartzb new Exk Mome file IcharlZa new txt Tipo file txt Annulla Fig 4 6 Window for domain files selection 1 2 Selec
50. v PROCEDURE ANALYSIS UAg EMC Laboratory C TIME DOMAIN DMU Applied Electromagnetic Group UMR EMC Laboratory FREQUENCY DOMAIN HOTE TO THE USER All the figures 11g and ASCII files txt are generated in the working directory The figures are not displayed on the screen ASCII files should be read by spreadsheet Plot Results Read Me programs i e WordPad to not altering the tabulated format License Fig 3 1 Initial window Main menu The major functions are Select the analysis mode in the ANALYSIS selection box Run FSV run an 1D FSV analysis see Chapter 3 Quit exit the program Plot Results display results of previous analysis without running a new one see Section 5 3 Read Me display this document Se s To run the first analysis the user has to select an analysis mode by using the radio button in the ANALYSIS selection box 14 1D FSV 4 0 2L FSV PROCEDURE UAg EMC Laboratory DMU Applied Electromagnetic Group UMR EMC Laboratory FREQUENCY DOMAIN NOTE TO THE USER Run FSv All the figures fi and ASCII files itxt are generated in the working directory The figures are not displayed on the screen ASCII files should be read by spreadsheet Plot Results Read Me programs i e WordPad to not altering the tabulated format License Fig 3 2 Selecting the Time Domain analysis e TIME DOMAIN time domain analysis is performed on two set of data that are generic Amplitude val
51. x DMc contains the confidence levels of the FSV output Analysis mode expands in Amplitude Magnitude Phase Combined x expands in A F G 46 e Analysis mode _ x DM_average contains the A F G DMtot A F G Dmconf A F G DMpw conv values and when available the GDM weighting factors Kadm Kfdm e Analysis mode _inputpaths contains the complete path of the input files used in the FSV run The fig files are MATLAB figure files e Analysis mode _inputdata figure of the input data after synchronization e Analysis mode DC inputdata figure of the DC Data used in the calculation of ADM FDM and GDM e Analysis mode _Low_inputdata figure of the Low Data used in the calculation of ADM FDM and GDM e Analysis mode High inputdata figure of the High Data used in the calculation of ADM FDM and GDM e Analysis mode x DMc confidence histogram e Analysis mode xjDMi figure of the point by point amplitude differences e Analysis mode ODMi figure of the point by point offset amplitude differences The mat file is MATLAB data file containing all the relevant FSV variables It is the input file for the Data Display Tool e Analysis mode Analysis contains the analysis data in MAT format The GSthrs x subdirectories have this structure x is the threshold value rs Amplitude GradeSpread 85 mat E Amplitude GradeSpreadChart_ 85 Fig GradeSpread 85 txt Fig 7 2 Grade Spread folder for Tim
52. y Higher is the reliability that 1s to say smaller is the value of SPREAD greater is the relative weight of one of them in GDM The following algorithm is used to compute the weighting factor kapy and krpy in 1 11 lt Spread lt Spread then Issa 1 u Spread Spread Else 1f Spread gt Spread Then Korn ES 1 K Spread Spread Else I n 1 K J 122 The Combined Analysis for complex values In the FSV 1D analysis the results of the comparisons of Magnitude and Phase can be combined and weighted This is the so called combined analysis FSV treats magnitude and phase parts of the data to be compared separately throughout and recombines them at the end This way 1s as close a similarity as possible to the manner in which engineers would approach the analysis of the magnitude phase data In a similar manner to the way visual decomposition into amplitude and feature comparisons are combined into an overall conclusion the magnitude and phase parts are considered separately and then weighted in the process of forming an overall opinion The FSV analysis is performed as in points 1 to 11 of the previous section on the magnitude and phase parts separately and recombined on a point by point x y base by using ADM and FDM functions through the K weighting factors according to ADM combined X E V Susa ADM ina X E IS hase j ADM pase X 1 i 12 FDM bina det M K mnd i FDM ina X g K iae FDM
Download Pdf Manuals
Related Search