Optimizing sequential diagnostic strategy for large-scale engineering systems using a quantum-inspired genetic algorithm: A comparative study

Abstract Sequential diagnostic strategy (SDS) is widely used in engineering systems for fault isolation. In order to find source faults efficiently, the optimized SDS selects the most useful tests and schedules them in an optimized sequence. In this paper, a multiple-objective mathematical model for SDS optimization problem in large-scale engineering system is established, and correspondingly, a quantum-inspired genetic algorithm (QGA) specially targeted at this SDS optimization problem is developed. This QGA algorithm uses the form of probability amplitude of quantum bit to encode each possible diagnostic strategy extracted from fault-test dependency matrix, and then goes through evolutionary process to find the optimal strategy considering dual objectives of the expected testing cost and the number of contributing tests. Crossover and mutation operations are combined with quantum encoding in this algorithm to expand the diversity of population within a small population size and to increase the possibility of obtaining the global optimum. A case of control moment gyro system from real practice is used to verify the effectiveness of this algorithm, and a comparative study with two conventional intelligent optimization algorithms proposed for this problem, PSO and genetic algorithm, are presented to reveal its advantages.