A Multi-objective Co-evolutionary Algorithm Based On Uniform Design And Its Application

Many real-world problems involve simultaneous optimization of several competing objectives that reflect various design specifications and constraints. The goal of multi-objective optimization algorithms is to find a set of the most reasonable and most reliable solutions from all possible solutions. There are two main challenges for multi-objective optimization algorithms:How to find out more nondominated solutions in the sparse regions of nondominated frontier in order to obtain uniformly distributed solutions along the nondominated frontier. How to find out nondominated solutions which are as near to the true Pareto frontier as possible in order to obtain high quality nondominated solutions. In this thesis, a new multi-objective co-evolutionary algorithm based on uniform design is proposed. The main works include:Firstly, the initial population is composed of two parts. One part is generated by uniform design method and the other part is randomly generated, Then the genetic operators are executed on the population to look for a nondominated frontier, thereafter, some new designed co-evolutionary operators, such as symbiotic operator, absorption operator and discrete operator, are used to produce a set of improved nondominated solutions. These proposed operators are used to two sub-populations to make their information exchange mutually. Secondly, in order to address the limitations of weighted sum method, we design a new discrete operator, which is expected to find the solutions in the non-convex part. Thirdly, the proposed co-evolutionary process, elitism preservation strategy and the designed crowding distance are expected to make the optimal solutions uniformly distributed along the Pareto frontier.Based on the above, a new evolutionary algorithm:a multi-objective co-evolutionary algorithm based on uniform design is proposed. The computer simulations on four test problems and the comparisons of the proposed algorithm with the well known algorithm NSGA-Ⅱ are made, the experimental results indicate that the proposed algorithm is more efficient and effective. Finally, this proposed algorithm is applied to the base stations allocation model of the flyers measurement and the results indicate the effectiveness and efficiency of the proposed algorithm.