A novel hybrid optimization algorithm: Dynamic hybrid optimization algorithm

Nowadays, many algorithms are invented with different strengths and weaknesses, none of which is the best for all cases. Herein, a hybrid optimization algorithm entitled the dynamic hybrid optimization algorithm (DHOA) is presented. We cover the weaknesses of one algorithm with the strengths of another algorithm using a new method of combination. There are two methods for combining algorithms: parallel and sequential. We adopted the parallel method and optimized the algorithm’s performance. In this method, unlike other parallel methods, the population size of the better algorithm is enhanced. Three algorithms were selected due to their relatively different performance in the optimization, so that the results could be more accurately examined. We aimed to achieve better and more accurate results in a shorter time by using the exploitation ability of PSO, HHO, and the crossover of GA. Twenty-three well-known examples were provided to determine the fitness of the proposed method and to compare it with these three algorithms. A group of 10 modern benchmark test functions of Congress on Evolutionary Computation (CEC) was used as an extra evaluation for DHOA. Three well-known engineering examples (10-bar truss, welded beam, and pressure vessel designs) were also examined to evaluate the performance of the proposed method. The three algorithms were the Genetic Algorithm (GA), particle swarm optimization (PSO), and Harris Hawks algorithm (HHO). According to the findings, the proposed method has a faster convergence and better performance than the other algorithms. It also yields better results than its basic algorithms. The Friedman mean rank of the proposed dynamic hybrid optimization was one of the top three algorithms among 23 well-known functions and CEC2019 examples. As for the three famous engineering examples (10-bar truss, welded beam, and pressure vessel designs), it was one of the top three algorithms.