论文部分内容阅读
针对中长期径流预报中存在许多不确定性因素,本文引入云理论构建径流预报的不确定性推理模型(UR).首先,该模型应用最大方差方法(MaxVar)对径流序列进行硬性分级,用级别概念表示径流分级区间,以期望(Ex)、熵(En)以及超熵(He)构成的云隶属函数描述径流级别概念的模糊性和随机性,实现分级区间软化,然后将径流量值进行属性转化,以此建立定性推理规则集,运用云算法进行径流不确定推理预报,成功实现径流序列不确定性传递;其次,对径流分级过程中超熵(He)参数确定进行了初探,对推理随机性输出结果进行统计分析,给出相应显著水平下的预报区间;最后,将该模型应用于南方某水库入库月径流预报中,并与广泛应用的最小二乘支持向量机(LSSVM)和ARMA模型进行比较分析,本文模型不仅具有较高的预报精度,而且能够进行区间预报,实例验证说明了模型的有效性和实用性.
In view of the many uncertainties in mid-long term runoff forecasting, this paper introduces cloud theory to construct a Uncertainty Inference Model (UR) for runoff forecasting.Firstly, this model uses the maximum variance method (MaxVar) to classify runoff sequences by level, The concept represents the classification interval of runoff. The cloud membership function composed of expectation (Ex), entropy (En) and super-entropy (He) is used to describe the fuzziness and randomness of the concept of runoff level and to realize the softening of the classification interval. And set up a set of qualitative reasoning rules. The cloud algorithm is used to forecast the uncertainty of runoff forecasting, and the uncertainty of the runoff sequence is successfully achieved. Secondly, the determination of the super-entropy (He) parameter in the runoff grading process is discussed. Finally, this model is applied to the monthly runoff forecasting of a certain reservoir in the South, and is compared with the widely used Least Squares Support Vector Machine (LSSVM) and the ARMA model For the comparative analysis, this model not only has a high forecast accuracy, but also can make interval forecast. The example verification shows that the model is effective and practical .