Diagnostic information based test points selection for analog circuit diagnosis

Test point selection is a critical step for analog circuit fault diagnosis. The conventional conception of test point is replaced by a more extensive one which means any measurable variable carrying fault information of the analog circuit. The prior and posterior uncertainty of the circuit fault state is described by Shannon entropy. Based on this description the conception of diagnostic information of test point set is proposed and the method for its estimation is given. The problem of optimal test point selection is modeled as searching for the test point subset with the maximal diagnostic information in the set of all accessible test points of the circuit. Genetic algorithm is used for the optimal subset searching and the strategy for chromosome coding, the choice of fitness function and genetic operator is discussed. The proposed method is used for test point selection of a linear analog circuit and satisfying results have been achieved.