Adaptive Population Learning EJaya Algorithm for Real-World Optimization Problems

EJaya enhances the global exploration capability of the Jaya algorithm but still suffers from drawbacks such as local stagnation and slow convergence due to its single learning strategy and weak population maintenance. To address these issues, this paper proposes an adaptive population-learning based improved variant, termed “PLEJaya.” In the PLEJaya algorithm, first an adaptive learning method improves the EJaya performance by refining the initial population and then a linear reduction method diminishes the worthless members from the population and thus improves the overall algorithm’s performance. The performance of PLEJaya has been examined on 53 benchmark functions, including 23 standard and 30 CEC 2017 test suite’s benchmark functions, and compared with eight established meta-heuristic algorithms such as TLBO, EJAYA, JAYA, PSO, GA, AOA, GWO, and WOA. Additionally, the practicality of PLEJaya has also been confirmed on 4 constrained engineering design applications. The experimental results confirm that the proposed PLEJaya solution significantly outperforms its competitors in terms of accuracy as well as convergence rate and thus provides a viable alternative to current optimization techniques.