Research On Multiple-criteria Decision Making Based On Differential Evolution Algorithm

As a kind of new method, Evolutionary algorithms have been used in engineering, management science, financial system, mathematical research, and many other fields. Differential evolution algorithm as an important member of the family of evolutionary algorithms, it is clear thinking, easy and quick to operate, and the numerical calculation effect is very good. We will introduce the origin, development of differential evolution algorithm, and give a research and discussion about its application of multiple criteria decision problems.First, based on the multi-criteria optimization strategy, we give an improved differential evolution algorithm and check its effectiveness with some test functions. It has the advantage of dynamic adjustment variation factor and cross rate, through the individual fitness and the distances information to search the best. This algorithm can get the global optimal solution with good result and fast convergence rate.Secondly, we have a research about the multi-criteria decision making problem. As an important component part of the decision theory, it generally contain multi-attribute decision making problem and multi-objective decision problem. On the one hand, research work of multiple attribute decision making problems need to discuss the consistency between attributes(solutions). We analyzed Saaty consistency index through statistical methods, then we talk about the ordinal consistency and give an interactive method to detection and correction the consistency of judgment matrix method, and combined with examples verified this method. On the other hand, in the process of the study of multiobjective decision making problems, the research emphasis is placed on the evolution algorithm for sovling these problems, based on the existing evaluation methods, we provided an multi-objective optimization method based on differential evolution algorithm, this algorithm was tested in the end too.We introduce the differential evolution algorithm and some multi-criteria decision making problems. However, due to the limited level, the problems and methods in the paper are inevitable, I will correction them in the future positively.