多目标优化Pareto支配性预测方法研究-计算机科学与技术专业论文.docxVIP

I II Abstract Evolutionary algorithms have been successfully applied to solve multi-objective optimization problems (MOPs), however, for expensive multi-objective optimization problems (EMOPs), evaluation of objective functions or constraint functions is a huge time-consuming procedure taking several hours or even days, which referred as curse of computation cost. Although surrogate models have lessen the evaluations of objective functions at some extent, lots of prior knowledge is needed to approximation techniques. In addition, different objective function requires different model, the type and parameter precision of model influence the accuracy of estimation directly, the construction of surrogate models are itself challenging. This paper engages in the study of Pareto dominance classification method , applying pattern recognition techniques to overcome the curse of computational cost, and presents a basic framework of pattern classifers to predict Pareto dominance between candidate solutions with unknown objective vector value through learning the sample Pareto dominance of candidate solutions. On the assumption that class condition probability density function subjects to normal distribution, this paper preliminarily implements a Bayesian classifier based on statistical learning theory, laying a foundation for further research of Pareto dominance prediction. To improve the accurracy of predicting Pareto dominance, this paper conducts a study of MOPs’ own characteristics. For MOPs whose domains differ in orders of magnitude, two similarity measurements of the weighed sum of binary bit string and the sum of ranked dimensional sequential number are put forward, which bring about obvious improvements of prediction accuracy and robustness compared with the nearest neighbor classification using Euclidian distance in decision space. For MOPs whose decision vector has equivalent components, the equivalent components of decision vector are defined by analyzing the contr