Nonlinear Multi-criteria Classification Methods In A Class Of Preference For Depolymerization

Multi-criteria classification decision making is an important research branch of multi-criteria decision making, it's also a kind of common questions in the administrative decision and social life, in brief, what it dealt with is the problem of classifying alternatives in well-defined categories under multiple criteria. In actual decision, the global preference of an alternative might not be a linear association with every criterion, that is to say, criteria is not independent but relative, and mean while, decision maker is very difficult to define all of the preference functions and the wights objectively before decision. Therefore, this paper has carried on systematic research on these questions and the nonlinear multi-criteria decision making problem with incomplete information.At first, this paper discusses the current domestic and international situation of multi-criteria decision making research and major classification methods on the basis of consulting a large number of documents, and has analyzed the limitation or the deficiency of these methods, such as MAUT, ELECTRE/PROMETHEE, UTA/UTADIS etc. Secondly, this paper develops an improved multi-criteria optimizing model with nonlinear preference aggregation function, and further more, provides a hierarchical discrimination approach for multi-criteria classification based on preference aggregation approach. This method has combined the advantage of current PROMETHEE, ELECTRE and MHDIS method together, and overcome the insufficiency among them.The next focal point and difficulty of study is to solve the multi-criteria optimizing model. In the case of common methods loses efficiency, the author proposes an effective hybrid algorithm which combines Evolution Strategies with dynamic penalty function and simulated annealing penalty function. It has successfully solved the multi-criteria decision model by computer programming and has greatly reduced the calculating difficulty in the past. Then from the view of a developer, the author explained the design and implementation of theinteractive multi-criteria classification decision making system that is based on the nonlinear classification optimization model and hybrid Evolution Strategies algorithm metioned before. This system not only has a high reliability and calculating speed, but also an interactive feedback mechanism, which makes the decision maker able to control the middle process, revise the result, and reduce people's subjectivity influence progressively, thus obtain the effective decision result rapidly.Finally, an instance has been indicated and verifed the feasibility, effectiveness and scientific of this method. It offers one for similar decision question in other relevant fields and disciplines to consult helpfully.