Simplified harmony search: novel algorithm design and its applications in engineering design optimization problems

Abstract The harmony search (HS) algorithm solves combinatorial optimization problems by simulating multiple musicians iteratively improvising their musical notes (i.e., decision variables for the concerned problem) to search the best harmony (i.e., the optimal solution). Each iteration of the HS algorithm employs two random values in two stages to choose one of the three operations consisting of harmony memory consideration, pitch adjustment, and random generation to generate a new note. In practice, however, the HS algorithm consumes an enormous number of iterations to find the best harmony, so that the two-stage generation of each new note consumes large computing resources when solving complex optimization problems with a considerable number of decision variables. Therefore, this work devises a simplified harmony search (SHS) algorithm that simplifies the classical HS algorithm through employing only one random value to choose one of the three new note generation operations. Our proposed SHS algorithm that adopts one-stage judgment can find optimal solutions much more efficiently than the classical HS algorithm that adopts two-stage judgment. The experimental performance of the SHS algorithm on eight benchmark function optimization problems as well as five practical engineering design optimization problems is evaluated. Experimental results show that this algorithm obtains better solutions more efficiently and stably than some representative HS variants and classical metaheuristic algorithms in all the experimental problems.