A multi-objective constrained optimization algorithm based on infeasible individual stochastic binary-modification
Huantong Geng, Qing-Xi Song, Tingting Wu, Jingfa Liu · 2009
During solving the constrained multi-objective optimization problems with evolutionary algorithms, constraint handling is a principal problem. Analyzing the existing constraint handling methods, a novel constraint handling strategy based on infeasible individual stochastic binary-modification is proposed in the paper. Its key point lies in modifying randomly infeasible individual into feasible one according to predefined modification rate (Rm) during evolutionary optimization. Finally, the proposed strategy is applied to the constrained multi-objective optimization evolutionary algorithm, and then the algorithm is tested on 7 benchmark problems and the comparison between our strategy and Deb's constrained-domination principle demonstrates that our strategy optimizes 30% faster than Deb's in the circumstances to preserve equivalent distribution and convergence of the solutions found.