Abstract: The conventional series system and parallel system are no longer able to adequately address the current system condition due to the rapid advancement of science. If faults are not identified and isolated, any system failure can lead to the inability to perform a single task or even multiple tasks. In order to promptly identify and isolate faults, the systems must have high testability. However, problems like omitting structural elements in testability influencing factors and ambiguous testability-related data make standard approaches useless when building and resolving testability allocation models. The grey optimization approach will lose a lot of grey information if a planning model is employed for solution, which will lead to large inaccuracies. Therefore, this paper proposes a testability allocation model for complex systems in the setting of grey information, and proposes the grey double-layer particle swarm optimization algorithm to solve the model. First, the particular factor that influence the testability allocation process is identified. Second, this paper proposes the TOPSIS method based on the improvement of grey entropy weight and determines the weights of the subsystems. Then, this paper proposes the grey nonlinear planning testability allocation model, and proposes the grey double-layer particle swarm optimization algorithm to solve the model. Finally, the viability and efficacy of the model are demonstrated by strength testing and comparison with other algorithms. 

Key words: Complex system, Testability allocation, System lifetime importance, Grey entropy weight, Grey double-layer particle swarm optimization algorithm