Research On NPS-CSS Algorithm Based On The Decision-makers’ Preference Information

In real life, many problems are composed of multiple targets, which are conflicting and influencing. Such problems turn to the optimization problem that at the same time making these goals best in a given area, that is known as a multi-objective optimization problem. Normally, Multi-objective optimization are divided into two steps. The first step is to obtain the Pareto frontier, or to get the set of non inferior solution, which is close to the Pareto frontier, this is the compromising process; The second step is to extract the final decision-making information, and then to determine a solution as the final "optimal solution" of an optimization problem from the set of non inferior solution, this is the decision process.This research is on the basis of NPS-CSS algorithm. We take into the decision-makers’ preference information, and then design the effective solving mechanism, so as to narrow my search space, which making the search only in the areas of decision makers interested in, specific content is as follows:Researching the construction method of solution set and comparing them with the standard test functions through simulation experiment. Results show that construction method of dominating sets based on reducing dominations relation matrix is superior to many of the common construction methods, including Deb. Construction method of dominating sets based on reducing dominations relation matrix improves the efficiency, and it’s important to improve the multi-objective optimization efficiency.Give the multi-attribute fuzzy evaluation of various purposes, for the second step which is the decision process of multi-objective decision-making problem. Combining comprehensive opinions from the experts and the decision-makers’ preference information, we take decision-makers’discrete opinions of the target attribute to comprehensive views of each target. We determine the decision-makers’ preference weight of each target by defining a fuzzy order index. According to the target weight we build the fitness function, and then solve the problem with NPS-CSS algorithm, so we can get the solution which gives decision-makers a high degree of satisfaction, and it also provides a feasible scheme for solving multi-objective problem. The algorithm fully considers the preference information of decision-makers and expert opinion, and it makes the whole solving process more Conform to the actual situation. The simulation results show the effectiveness of the algorithm.