Social harmony search algorithm for continuous optimization

This paper presents a social harmony search algorithm to solve optimization problems with continuous design variables. Although the Harmony Search (HS) algorithm (HSA) has proven its ability in finding near global regions within a reasonable time, it is rather inefficient in performing local search. The proposed method applies the harmony search optimizer for global optimization and normal distribution is employed to update the position of each design variable of a new harmony found by the first rule of the HS (memory consideration) in every stage to rapidly attain the feasible solution space. Normal distribution works as a global search in early iterations and as a local search in final iterations to improve HS in order to quickly converge and find better solutions. Various benchmark optimization problems are used to illustrate the effectiveness and robustness of the proposed method. Finally, the experimental results reveal the superiority of the proposed method in quick convergence and finding better solutions compared to the classic HS, its recently developed variants, and some other optimization algorithms.