Vector Design Optimizations Using an Improved Cross-Entropy Method

An improved vector cross-entropy method is proposed to solve multi-objective design optimizations. To balance exploitation and exploration searches, the entire iterative process is recast into two phases: 1) diversification and 2) intensification phases. Different parameter updating mechanisms of the probability density function (pdf) are consequently designed. To enhance the diversity of the population, several different pdfs are evolved synchronously; to guarantee uniform distribution of the searched Pareto solutions, an elite projection mechanism is proposed. Mathematical functions and two inverse problems are used to testify the effectiveness and efficiency of the proposed method. The numerical results show that the proposed method can obtain well-distributed Pareto solutions using a small iteration number.