Search Mechanism Of Harmony Search Algorithm And Its Application

In scientific research and industrial design,a large number of practical problems are often translated into optimization problems.For problems with high dimensional and multi-parameter properties,it is very difficult to solve t by the traditional optimization methods.Therefore,the scholars have put forward adaptive meta-heuristic intelligent algorithm.Harmony Search(HS)is a newly developed meta-heuristic algorithm.The algorithm is an underivative real-time parameter optimization algorithm,which has been inspired by the process of music improvisation to find a perfect harmony state.The harmony search algorithm has fewer requirements for mathematics and can solve various engineering optimization problems easily.However,the algorithm is easy to fall into the local optimal solution,poor convergence and insensitive parameter adjustment.In order to improve the convergence of the harmony search algorithm and avoid premature convergence phenomenon,the improved harmony search algorithm is constantly proposed.In this paper,the research background of harmony search algorithm(HS)is introduced.And,the basic principle of harmony search is introduced in detail in the second chapter.At the same time,inspired by the self-adaptive parameters,the particle swarm optimization(PSO)and differential evolution algorithm,this paper presents two different strategies for improving the harmony search algorithm,namely semi-self-adaptive harmony search algorithm(SSaHS)and adaptive harmony search with differential evolution based on linear population size reduction(aHSDE).In addition,compared with other advanced improvement harmony search algorithms on CEC 2014 benchmark.Finally,the results show that the improved harmony search algorithms proposed in this paper have better results and better performance.Firstly,a semi-self-adaptive harmony search algorithm(SSaHS)is proposed in the third chapter,which adopts the adaptive adjustment of bandwidth and the optimization strategy of particle swarm optimization.SSaHS uses the difference between the maximum and minimum values in the harmony memory as the bandwidth of adaptive adjustment.This algorithm can dynamically adjust the bandwidth of specific problems,enhance local development capability and improve the accuracy of optimization results.In order to verify the effectiveness and learning strategy of the proposed algorithm,a set of well known benchmark functions is employed and compared with some competitive HS variants.Experimental results indicate that the semi-self-adaptive harmony search algorithm can find better solutions and have better performance than the basic harmony search algorithm and the improved harmony search algorithm,such as IHS,GHS,NGHS.In addition,based on differential evolution strategy and the Lehmann mean,difference operator is introduced and the concept of adaptive adjustment parameters,this paper proposes an adaptive harmony search with differential evolution based on linear population size reduction(aHSDE)in the fourth chapter.The bandwidth is adjusted using the differential mutation strategy,so that aHSDE provides an effective bandwidth adjustment method and enhances the performance of the harmony search algorithm.In order to increase the diversity and convergence of the harmony memory,the size of harmony memory is adjusted linearly.At the same time,using the peroid learning and Lehmann mean strategy,adaptive adjustment is made to the fine-tuning probability PAR and the mutation rate F.Finally,aHSDE is compared with the basic HS algorithm and HS variant algorithms by the CEC 2014 benchmark functions.Experiment results indicate that aHSDE has stronger competitiveness and better convergence.