An efficient enhanced exponential distribution optimizer: applications in global, engineering, and combinatorial optimization problems

Abstract This paper introduces an enhanced exponential distribution optimizer (EDO), termed the exponential distribution optimizer with Levy flight orthogonal learning (EDO-LFOL). EDO-LFOL enhances EDO by integrating the Levy flight (LF) strategy during the intensification phase and utilizing the orthogonal learning (OL) approach after the end of the optimization cycle. This adjustment aims to mitigate the risk of local optima entrapment and improve the quality of the solutions obtained. This paper presents EDO-LFOL as a viable global and practical optimization problem solution. To evaluate the effectiveness of EDO-LFOL, we compare it against seven robust optimizers across 12 unconstrained test functions associated with the IEEE Congress on Evolutionary Computation 2022 (CEC 2022). The optimizers include the improved multi-operator differential evolution algorithm (IMODE), adaptive guided differential evolution algorithm (AGDE), whale optimization algorithm (WOA), grey wolf optimizer (GWO), sinh cosh optimizer (SCHO), RIME optimization algorithm (RIME), and the original EDO. Additionally, EDO-LFOL is tested on three combinatorial optimization challenges: the job shop scheduling problem (JSSP), quadratic assignment problem (QAP), and bin packing problem (BPP), to evaluate its applicability. Furthermore, EDO-LFOL addresses four distinct design engineering challenges: the pressure vessel design, three-bar truss, welded beam, and speed reducer. The results, supported by significance tests, demonstrate that EDO-LFOL significantly outperforms the standard EDO and its competitors.