Improvement of Fuzzy Comprehensive Evaluation Model for Water Quality

Dimensionless method will make information distortion, and can not be controlled by information condition of water quality classification. Thus, theoretically, fuzzy recognition which takes ¿distance¿ as similarity measurement is not suitable for evaluating water quality, the best choice for which is to use membership degree as similarity measurement between sampling points and classes. However, fuzzy comprehensive evaluation based on membership degree exists the following problems:1) It is faulty to use ¿over standard weight¿ to realize the conversion of member ship degree from index value to sample point; 2) Fuzzy membership functions is not normal. Both of the problems cause undesirable results, and our improved methods be:1) Ruled fuzzy membership function and its properties, such as topological space structure, algebraic properties and normalized method; 2) Point out that the conversion of membership degree from index value to sample point is a pure mathematical problem and index distinguish weight be determined only by the entropy of index value membership degree. Finally, we take an example to analyze the improved method, and comparing with the results of same cases, we can observe the difference between the original method and the improved method in the paper.