A Global Harmony Search Algorithm Based on Tent Chaos Map and Elite Reverse Learning

Abstract

To improve the performance of the harmony search algorithm and enable the processing of increasingly complicated optimization problems, a global harmony search algorithm based on tent chaos map and elite reverse learning (HS-TE) has been proposed. The algorithm uses the tent chaos map to initialize the population and adopts the elite reverse learning strategy to optimize the iterative process. The method reduces the algorithm’s dependence on the initial solution, improves the search optimization ability, enhances the diversity of the population, and establishes adaptive parameters to control the development and exploration of the iterative process, which is beneficial to improving the algorithm’s search ability. Create test experiments: Various HS algorithms perform classic benchmark function tests. The experimental test data shows that the algorithm is better than the current five improved harmony search algorithms and has better convergence and accuracy. The algorithm is used to improve the penalty parameters and kernel function parameters of SVR, and then use the optimized SVR to perform regression prediction on the daily opening number of the Shanghai Stock Exchange. According to the experimental results, the upgraded SVR provides better prediction performance. It works both in theory and in real life and can be used to predict the Shanghai Securities Composite Index.