The Use of Simulated Annealing with Fuzzy Objective Function to Optimal Frequency Selection for Analog Circuit Diagnosis

Presented work focuses on detecting a single hard fault in analog electronic circuit in AC domain. The combined approach to analog circuit fault test frequency selection is proposed. Reduction of a fault dictionary size is the main goal of the method. It is achieved by the use of heuristic algorithm and modified fitness function. Simulated annealing as a search engine and fuzzy system as a fitness function is applied. During the search process the fault detection rate of a CUT must remain at the same level (obligatory condition) while localization should stay unchanged (both with respect to the use of all frequencies). The input sources frequencies should separate all possible (assumed) circuit states. Proposed approach is verified on the example and compare with pure simulated annealing algorithm (weighted function).