针对一类目标函数非光滑的l1模极小化问题,提出了一种改进的和声搜索算法。结合差分进化算法的变异策略,用差分向量算子取代和声搜索算法的音调微调。实验结果表明,改进后的和声搜索算法能够获得原问题的全体解。
Concerning the di?culty in solving nonlinear l1 norm minimization problems whose objective func-tion is nonsmooth, a new method, called improved harmony search (IHSDE) algorithm, was proposed in this paper. Combining with mutation strategy in differential evolution (DE) algorithms, we replaced the pitch-adjusted operation in classical harmony search (CHS) with different vector operators borrowed from the DE algorithms. The numerical results, compared with the CHS, show that the IHSDE method can get all the solu-tions to the original problem.