水文预报模型模糊综合评价方法研究及应用

Research and Application on Fuzzy Comprehensive Evaluation Method for Hydrologic Forecast Model

摘要: 本文根据水文情报预报规范准确度、精密度和可信度的要求，以应用于三峡梯调中心的年径流预报自回归小波模型、最近邻回归小波模型、均生函数模型为研究对象，采用基于模糊集理论的模糊综合评价法对水文预报模型在特定流域条件下的预报效果进行定量评价。为此，首先建立水文预报模型模糊综合评价多指标评价体系，然后依据最大隶属度原则及贴近原则选取抛物型隶属函数对模型评价因素集进行规范化处理，最后对规范化得到的评价集采用基于决策者偏好的信息熵方法进行主观和客观相结合的综合评价，从而得到符合工程实际需求的水文预报优选模型。研究结果表明，模糊综合评价法在水文预报领域有很强的适用性，采用基于决策者偏好的信息熵多指标模糊综合评价法可以引入决策者主观经验对水文预报模型进行主观与客观相结合的定量评价，并有效降低预报模型评价指标体系中的信息冗余，从而实现对预报模型科学精确的定量评价。

Abstract: According to accuracy, precision and reliability requirement of hydrological forecasting specification, we use the fuzzy comprehensive evaluation method based on fuzzy set theory to quantitatively evaluate the performance hydrological forecasting models. The study selected auto-regressive wavelet model, nearest neighbor regressive wavelet model and mean generating function model of three gorges annual runoff fore- cast as the study object, and built a multi-index evaluation system for hydrologic forecasting model compre- hensive evaluation. Set parabolic type membership function as the standardized method, on the basis of maximum degree principle and close to principle. Using the information entropy method based on decision makers’ preference in standardized value set to get a weight combining subjective and objective aspects. The best model is the one which has the highest result value most close to the theoretical optimal value. The research results show that the fuzzy comprehensive evaluation method is applicable for hydrological forecasting field. Fuzzy comprehensive evaluation method based on decision makers’ preference information entropy method can effectively reduce redundant information and introduce decision maker’s subjective experience to quantitatively evaluation making with a combination of subjective and objective to get a scientific and exact result.