A test point selection method based on recognition of typical topology structure of complex networks

System modeling and Test Point Selection (TPS) are among the key issues in Prognostics and Health Management (PHM) for electronic systems. In this paper, we propose a novel methodology for system-level testability modeling and test point selection, which is based on recognition of three typical topology structures. A network modeling method is firstly designed based on transformation property of fault signal. Genetic Algorithm (GA) is then used as searching and recognition method for typical struc­tures, for which TPS strategy is designed. Seven typical similarity calculation criteria and one cosine based criterion are compared by their performance in distinguishing structure feature matrixes, thus improves the accuracy and robustness of the algorithm. Numerous feasible solutions are formed based on different combination of extracted structures. The most feasible solution is picked by the ratio of Fault Detection Rate (FDR) per Test points Number (FDRTN) satisfying stated FDR requirement. The proposed method is validated by a case study of a typical avionic power circuit. The recognition performance and TPS result are evaluated and discussed.