An Improved Differential Evolution for Optimization of Chemical Process*

Differential evolution (DE) is an evolutionary optimization method, which has been successfully used in many practical cases. However, DE involves large computation time, especially, when used to optimize the computationally expensive objective function. To overcome this difficulty, the concept of immunity based on vaccination is used to help proliferate excellent schemata and to restrain the degenerate phenomenon. To improve the effectiveness of vaccines, a new vaccine autonomous obtaining method, and a method of deciding the probability of vaccination are proposed. In addition, a method for modifying the search space dynamically is proposed to enhance the possibility of converging to the true global optimum. Experiments showed that the improved DE performs better than the classical DE significantly.