Performance Evaluation of the Plant Growth Optimization Algorithm for Constrained Nonlinear Optimization

Constrained nonlinear optimization plays a fundamental role in engineering design due to the presence of irregular feasible regions and interacting nonlinear restrictions. This study evaluates the performance of the Plant Growth Optimization (PGO) algorithm in a constrained nonlinear benchmark problem. The algorithm was implemented in MATLAB® and assessed using a fixed external penalty formulation for constraint handling. Performance was analyzed through convergence dynamics, constraint evolution, dispersion across 20 independent runs, and computational efficiency. A comparative study was conducted against Particle Swarm Optimization (PSO), Genetic Algorithms (GA), and Differential Evolution (DE) under identical experimental conditions. Results show that PGO achieves stable convergence within 87 generations, consistently attaining a feasible solution near the constraint boundary with low dispersion across runs. Statistical validation using the Friedman test (χ2=32.45, p<0.001) confirmed significant performance differences among algorithms, while post-hoc Wilcoxon tests indicated comparable performance between PGO and DE and significant differences relative to PSO and GA. These findings demonstrate that PGO provides a balanced compromise between robustness, convergence stability, and computational efficiency, supporting its suitability for constrained nonlinear engineering optimization tasks.