Parallel Test Sheets Generation Using Differential Evolution Algorithm with Constraint Effective Encoding and Either-Or Mutation

Abstract

Parallel test-sheet generation (PTSG) is a NP-hard constrained combinatorial optimization problem. For the large scale PTSG problem in real-world applications, evolutionary algorithm is an attractive way to find high quality solutions. For its reliability and high performance, differential evolution algorithm (DE) has been a promising optimizer in evolutionary computing. In this paper, DE algorithm with the state-of-the-art rand/1/Either-Or mutation scheme is designed to solve PTSG problem. A simple truncating encoding method and an elaborately designed constraint effective encoding method for DE are developed. To evaluate the performance of the proposed DE algorithm, simulation experiment was conducted on a series of item banks with different scales. Superiority of the proposed constraint effective encoding method is demonstrated by comparing it with truncating encoding strategy.