A new method for objective weights computing in multi-objective optimization

For the problem that how to compute objective weight based on personal preference during multi-objective optimization process, this paper proposes a method which decides the objective weights by solving a constrained optimization problem. At first, it transforms the objective weight computation problem into a synthetical fitness optimization problem according to statistics theory, then it transforms the personal preference into constrain condition of the synthetical fitness optimization problem. At last, it uses gradient projection method to solve the constrained synthetical fitness optimization problem to get optimum objective weights. The proposed method is used to compute objective weight of economical efficiency, safety and rapidity during the electric locomotive repair strategy decision process, and the test result proves that the proposed method gets optimum objective weight under constraint of personal preference.