An Enhanced Equilibrium Optimizer for Solving Optimization Tasks

Abstract

The equilibrium optimizer (EO) represents a new, physics-inspired metaheuristic optimization approach that draws inspiration from the principles governing the control of volume-based mixing to achieve dynamic mass equilibrium. Despite its innovative foundation, the EO exhibits certain limitations, including imbalances between exploration and exploitation, the tendency to local optima, and the susceptibility to loss of population diversity. To alleviate these drawbacks, this paper introduces an improved EO that adopts three strategies: adaptive inertia weight, Cauchy mutation, and adaptive sine cosine mechanism, called SCEO. Firstly, a new update formula is conceived by incorporating an adaptive inertia weight to reach an appropriate balance between exploration and exploitation. Next, an adaptive sine cosine mechanism is embedded to boost the global exploratory capacity. Finally, the Cauchy mutation is utilized to prevent the loss of population diversity during searching. To validate the efficacy of the proposed SCEO, a comprehensive evaluation is conducted on 15 classical benchmark functions and the CEC2017 test suite. The outcomes are subsequently benchmarked against both the conventional EO, its variants, and other cutting-edge metaheuristic techniques. The comparisons reveal that the SCEO method provides significantly superior results against the standard EO and other competitors. In addition, the developed SCEO is implemented to deal with a mobile robot path planning (MRPP) task, and compared to some classical metaheuristic approaches. The analysis results demonstrate that the SCEO approach provides the best performance and is a prospective tool for MRPP.

A Hybrid Equilibrium Optimizer Based on Moth Flame Optimization Algorithm to Solve Global Optimization Problems

“…The proposed EO variant was tested on 96 benchmark functions and 6 engineering optimization problems, comparing favorably against various advanced metaheuristic techniques. Finally, in [55], adaptive mechanisms, Cauchy perturbation, and cosine search strategies were introduced into EO to enhance the overall search capabilities of the algorithm. Testing on 15 classic test functions and the CEC 2017 benchmark test set demonstrated that the algorithm outperformed the comparison algorithms.…”