An Improved MOEA/D Algorithm Based on Angle Trajectory and Self-Adaptive Weight
Multi-objective optimization algorithm based on decomposition(MOEA/D)is an effective method to solve multi-objective optimization problems.The main idea is to weight the target by a set of evenly distributed weight vectors to form different scalar subproblems,and get the Pareto optimal solution set by coevolution in its neighborhood.However,these evenly distributed weight vectors can not be well adapted to the multi-objective problem of all frontier shapes,resulting in a decrease in the diversity of solution sets.In order to make the algorithm efficiently judge the frontier type and effectively adjust the weight vector,this paper proposes an improved MOEA/D algorithm based on angle trajectory and adaptive weight.The algorithm reduces the time complexity of clustering by using the angle-based clustering method,and improves the accuracy of frontier division by using the clustering-mer-ging strategy.After judging the frontier type according to the clustering results,the algorithm adopts the allocation strategy of divid-ing regions according to the angle,or allocating according to the length and number of frontier segments.In this way,computing re-sources can be allocated effectively and the efficiency of solving multi-objective problems can be improved.Experiments have proved that the allocation strategy of the algorithm is effective and can improve the diversity of solution sets of peaks,long tails and discon-tinuous fronts,and it is superior to the comparison algorithm in convergence and diversity.
万方数据
维普
