Electromagnetic wave propagation algorithm: A novel electromagnetic wave propagation-inspired optimizer for engineering applications

This study introduces a new physics-based meta-heuristic optimization technique, termed Electromagnetic Wave Propagation Algorithm (EMWPA), which models the propagation behavior of electromagnetic waves for global search diversity to improve exploration. The algorithm employs two cosine waveforms with different amplitudes to create a dynamic balance between global exploration and local exploitation. To improve the conventional wave-based search approach, EMWPA integrates a directional update mechanism that enhances convergence accuracy and a phase modulation strategy that prevents premature convergence. The contributions of this work are threefold: (i) the formulation of a physics-driven optimization model based on electromagnetic wave principles; (ii) enhancement of the basic propagation framework through directional learning and phase modulation to prevent premature convergence; (iii) performance validation through extensive experiments on 30 benchmark test functions including, unimodal and multimodal that evaluate convergence behavior, robustness, and optimization quality; and demonstration of EMWPA's applicability through four real-world engineering applications encompassing the design of pressure vessels, optimizing welded beam structures, developing gear train mechanisms, and harmonic minimization in multilevel inverters. Results demonstrate that EMWPA achieved the best average solution in 27 out of 30 benchmark functions compared to existing methods such as ABC, CSA, MVO, SSA, GJO, and PO. In constrained engineering applications like pressure vessel and welded beam design, EMWPA obtained the best or second-best cost solutions while maintaining low computational time. In the harmonic minimization case, the EMWPA and PO algorithms yield the most effective reductions in Total Harmonic Distortion (THD). Notably, EMWPA achieves convergence 38.91% faster than PO, highlighting its computational efficiency and harmonic suppression capability. Simulation results reveal that EMWPA consistently outperforms several well-known algorithms (ABC, CSA, MVO, SSA, GJO, and PO) in terms of solution quality, stability (low standard deviation), and computational efficiency.