An improved vector cross-entropy method is presented to provide a potential candidate for solving multi-objective inverse problems. To balance the exploitation and exploration searches, the whole evolutionary process is divided into two phases: diversification and intensification phases. Different parameter evolution mechanisms of probability density function (pdf) are proposed in each phase. To enhance the diversity of the population, several different pdfs arc evolved synchronously. To guarantee uniform sampling of the Pareto optimal solutions, an elite projection mechanism is proposed. MOP test functions and a high frequency inverse problem arc used to testify the effectiveness and efficiency of the proposed method.
展开▼