Handling Constrained Many-Objective Optimization Problems via Problem Transformation

Objectives optimization and constraints satisfaction are two equally important goals to solve constrained many-objective problems (CMaOPs). However, most existing studies for CMaOPs can be classified as feasibility-driven-constrained evolutionary algorithms (CMaOEAs), they always give priority satisfy constraints, while ignoring the maintenance of population diversity dealing with conflicting objectives. Consequently, may pushed toward some locally feasible optimal or infeasible areas in high-dimensional objective space. To alleviate this issue, article presents a problem transformation technique, which transforms CMaOP into dynamic (DCMaOP) handling optimizing objectives simultaneously, help cross large discrete regions. The well-known reference-point-based NSGA-III is tailored under model CMaOPs, namely, DCNSGA-III. In article, $\boldsymbol {\varepsilon }$ -feasible solutions play an role proposed algorithm. end, DCNSGA-III, mating selection mechanism environmental operator designed generate choose high-quality offspring solutions, respectively. algorithm evaluated on series benchmark three, five, eight, ten, 15 compared against six state-of-the-art CMaOEAs. experimental results indicate that highly competitive solving CMaOPs.