Abstract: This paper proposed an improved set pair analysis model to solve the problem of subtle differentiation between uniform. In view of the defects of using subjective analysis separately, or objective analysis alone, we put forward a new idea of comprehensively determine the weight coefficients. Weight coefficients are calculated by the aid of optimized analytic hierarchy process for subjective weights and entropic coefficients method for objective weights, and then the objective weights and subjective ones are combined according to the minimum discrimination information principle, which improves the veracity of the weight. This method was used to evaluate the water quality of the Dahuofang reservoir, and the result is compared with those from fuzzy set method and BP neural network method. The results show that the new method has good application prospect as it reflects the actual situation and water quality much better.

Key words: water quality assessment, set pair analysis, combinational weight, AHP, Dahuofang Reservoir