Orthogonal Learning-Based Improved Slime Mould Algorithm for Global Optimization

Abstract

The meta-heuristic algorithms with many excellent optimization characteristics have been widely adopted in practical engineering projects, such as production scheduling, robot path optimization, and mechanical design. The slime mould algorithm (SMA) is a newly swarm-based meta-heuristic algorithm by simulates the foraging and movement behavior of slime mould. When SMA deals with global optimization problems, it can get a more promising solution most of the time compared with other algorithms. But it still has some shortcomings like other meta-heuristic algorithms, for example, when facing complex problems, the convergence speed and the capability of avoiding local minima are insufficient, and the capabilities of global and local search are not balanced. Therefore, this paper proposed an improved SMA with orthogonal learning strategy (OLS) and boundary restart strategy (BRS), called OBSMA. Specifically, without major changes in the structure of SMA, the OLS is used to discover more useful information to update the best position, enhance the local exploitation capability, and increase the convergence of the SMA. Simultaneously, the BRS can adjust the individual position which outside the search boundary and help the algorithm to avoid local minima, maintain the diversity of the population, and enhance the exploration capability. The performance of OBSMA was validated and the effectiveness of the two strategies were compared on the 30 IEEE CEC2014 functions. Experimental results and statistical analysis prove the effectiveness of these two strategies and the combined use of the OLS and BRS enables SMA to obtain better performance.