Multi-objective evolutionary algorithm using decomposition method and polynomial mutation operator

The algorithm ofMOEA/D-M2Mcan transform multi-objective optimization problem into a number of multi-objective optimization sub-problems and obtains the Pareto solutions of these sub-problems separately, and finally obtains the Pareto solution of the multi-objective optimization problem, which ensures the diversity of population and has better algorithm performance than MOEA/D. Polynomialmutation operator has the function of strengthening local search and convergence, but there is less work on the application of themutation operator in the multi-objective evolutionary algorithm.A new multi-objective evolution algorithm using decomposition methods and polynomial mutation operators (MOEA/PmD) is proposedby combining the polynomial mutation operator and the new decomposition method of multi-objective optimization problem proposed in MOEA/D-M2M. Non-uniform mutation operator obtains the adaptability by dynamically adjusting the step size, an attempt is done to replace the polynomial mutation operator in MOEA/PmD with the non-uniform mutation operator, and multi-objectiveevolutionary algorithm using non-uniform mutation operator and decomposition method (MOEA/NumD) is devised. Experiments show MOEA/PmD algorithmhas better performance than MOEA/NumD and MOEA/D-M2M.