An Improved Hybrid Evolutionary Algorithm

Conventional genetic algorithm is prone to many problems, such as premature convergence, poor performance of partial search, inefficient in the final stage, difficulty in keeping balance between population diversity and selective pressure. In order to resolve these problems, the amount of information from parents was measured with correlation coefficient. Then an alternation strategy based on hereditary information was presented, which not only guaranteed the population diversity, but provided support for searching the optimum solution. Adaptive probabilistic crossover and mutation that can vary according to the change of the population fitness is applied to the evolution. Finally, an improved genetic simplex algorithm was put forward, which not only increased the population diversity, but also improved the solution quality according to simulation results.