Enhancing feature selective validation (FSV) interpretation of EMC/SI results with grade-spread

The comparison of high volumes of potentially visually complex data requires an automated method to support the engineers involved. The Feature Selective Validation (FSV) method is becoming increasingly popular as a solution to this. This paper reports on an enhancement to the method which uses the confidence in the FSV's component measures to provide a means of relative weighting of them and thus improving the ability of the method to satisfy one of the key criteria: mimicking human response. This is a good analogue to how the data would be considered by individuals. The trend components of the data are compared to provide an Amplitude Difference Measure (ADM) and the feature components are compared to provide a Feature Difference Measure (FDM). These two components can be viewed as point-by-point results (allowing direct comparison with the original data) in order to identify the precise locations of the elements leading to poor comparisons or as single summary numbers allowing an overall level of agreement to be obtained. The point-by-point values and the summary values can be combined (treating them as independent) to give a Global Difference Measure, a 'headline' measure of the quality of agreement.