Researches On Comprehensive Evaluation Method Based On Fuzzy Neural Network

With its applications in various fields, fuzzy comprehensive evaluation (FCE) is becoming more and more significant. However, how to determine the weights and give the single factor evaluation matrix and avoid unreasonable results are needed more researches. Though many people have put forward some new methods to improve comprehensive evaluation model, the results are not always satisfied. With the advances in artificial intelligence (AI), people begin to use the neural network (NN) technology and fuzzy techniques in comprehensive evaluation.In the paper, we discuss the unreasonable or ineffective phenomenon and two problems in FCE: (1) How to determine the weight; and (2) how to give the single factor evaluation matrix. Then, an improved method is first proposed to obtain correct results when the weight information is partly known or unknown completely, and to effectively avoid unreasonable phenomenon at the same time. And then, we combine FNN with this improved method and give a new method to determine the weights and evaluation matrix. Finally, an example is given to illustrate the feasibility and effectiveness of our new method.