摘要

Because the basic Harmony Search algorithm has the following disadvantages such as robustness is not high, slow convergence and premature etc., an improved chaotic harmony search algorithm is proposed. Firstly, chaotic Logistic mapping was introduced in harmony search memory, and the periodicity and randomness of chaos is used to improve the quality of the initial solution. Secondly, the memory is divided into several sub-memories, and the worst solution of each sub-memory is updated to improve the convergence rate. Then the optimal solutions of every sub-memory are mutated by chaos to avoid premature, thus enhance the robustness and convergence speed of basic harmony search algorithm. The performance of the proposed algorithm is verified by simulations through a series of test functions. Simulation results show that compared with the basic Harmony Search algorithm and Genetic Algorithm, the improved algorithm can get better global optimal solution and a smaller variance. The proposed chaotic harmony search algorithm has been applied to spectrum sensing of Cognitive Radio Sensor Networks which are widely used in many engineering fields such as environmental monitoring, intelligent transportation.

全文