A hybridized multi-algorithm strategy for engineering optimization problems

Abstract

This paper presents a new multi-hybrid algorithm (MHA) based on the hybridization concepts incorporated in the naked mole-rat algorithm (NMRA). The new algorithm uses the exploratory concepts of whale optimization algorithm (WOA), moth flame optimization (MFO) and marine predator algorithm (MPA) to enhance the worker phase or the exploration operation in NMRA. In order to overcome the problem of local optima stagnation, a new stagnation phase inspired from the concepts of GWO and cuckoo search (CS) algorithm is also added. Apart from these modifications, self-adaptivity has been incorporated by adapting all the parameters of the proposed algorithm. Here five new mutation strategies including simulated annealing (sa), exponentially decreasing (exp), chaos based (chaotic), linearly decreasing (linear) and oscillatory inertia weight (oscillating) have been exploited for different parameters of all the algorithms under study so that no user based parametric adjustments are required. For performance evaluation, the MHA is subjected to CEC 2005, CEC 2014 benchmark problems and image thresholding problem. The comparison on variable population sizes and dimension size has been done to prove the significance of the proposed algorithm over variable function evaluations and higher dimensional problems. From the experimental results, it can be seen that proposed MHA performs better than adaptive differential evolution with optional external archive (JADE), self adaptive differential evolution (SaDE), opposition and exponential whale optimization algorithm (OEWOA), sine cosine crow search algorithm (SCCSA), fractional-order calculus-based flower pollination algorithm (FA-FPO), success-history based parameter adaption differential evolution (SHADE), SHADE with linear population size reduction hybrid with semi-parameter adaptation of CMA-ES (LSHADE-SPACMA), Laplacian biogeography-based optimization (LX-BBO), random walk grey wolf optimizer (RW-GWO), improved symbiotic organisms search (ISOS), variable neighbourhood bat algorithm (VNBA), improved elephant herding optimization (IMEHO) and others. Statistical results in terms of Wilcoxon’s rank-sum test, Freidman’s test and convergence profiles further validate the superior performance of the proposed MHA algorithm. The source code can be downloaded from https://github.com/rohitsalgotra.