Implementing the Feature Selective Validation (FSV) technique

Whether the problems to be validated are generated from the previous sets or from data generated by the modeler for a specific problem, the level of agreement needs to be determined between those results and reference results (such as from measurements or other models).  The Feature Selective Validation (FSV) method is the accepted technique for doing this.  This section contains additional information to support the implementation of the FSV method and to provide assistance in the interpretation of the results produced by the method.  No ‘standard code’ is provided, users are free to generate their own using their preferred symbolic solver, high or low level language or to embed within a spreadsheet package as suits their purposes and requirements.


This section should be read in conjunction with clause 7.4 of IEEE P1597.1™/D4.2, from which this section quotes freely and annotates with examples and notes.


As the main purpose of the FSV will be to quantify changes and variations, small variations caused by differences in implementation should not prove problematic and users are encouraged to look at relative differences and the detailed information to provide intelligence about the comparisons that can then be used to provide a measure of acceptability or to answer whether any changes have been improved sufficiently

 

Questions concerning this site should be addressed to the Joint EMC Model Validation Committee chairman, Bruce Archambeault at barch@us.ibm.com