Hybrid Algorithm of Locust Swarm and Spider Colony for Global Optimization Problems

Nature-inspired algorithms demonstrate great potential for solving complex search engine optimization problems in engineering calculations and design of large and small energy facilities, including renewable energy sources. In this paper, we propose a novel hybrid algorithm that models the behavior patterns of a locust swarm and a spider colony to efficiently solve global optimization problems of multivariate multi-extremal functions. The modified locust swarm algorithm avoids concentration of individuals at the current best positions, reduces the probability of premature convergence, and maintains the intensification/diversification balance. The modified spider colony algorithm models two unique behavior patterns, allows diversification of the search space for the optimal solution, and acts as a filter to reduce the influence of very good or bad solutions on the search process. Hybridization of the modified algorithms is done by combining them sequentially (preprocessor/post-processor). We experimentally tested the algorithm on seven known multivariate functions. We compared the results with competing algorithms for particle swarm, differential evolution, and colony of bees. The proposed algorithm provides better results for all considered functions. Checking the statistical significance of the results obtained using the Wilcoxon rank sum test for independent samples showed that the algorithm results are statistically significant.