Research on monthly flow uncertain reasoning model based on cloud theory

Abstract

In view of the mid and long term runoff forecasting containing many uncertain factors, this paper constructs a uncertain reasoning model (UR) based on the cloud theory to solve the problem of uncertain reasoning. Firstly, in the proposed model, a classification method, i.e., attribute oriented induction maximum variance (MaxVar), is used to divide the runoff series into different intervals, which are softened and described by the cloud membership with expected value (Ex), entropy (En) and hyper-entropy (He), then an uncertain reasoning rule set is constructed by means of the runoff value generalization and applied to monthly flow for uncertain prediction. Next, a new modification formula is used to calculate He in runoff forecasting, and a confident level probability prediction interval is obtained by statistical method. Finally, this paper takes the monthly flow of Manwan station in China as an example and uses UR model, LSSVM model, and ARMA model to calculate the monthly flow, respectively. The results show that the UR model has the highest prediction accuracy compared to other models, and that it not only provides random output but also supports probability interval prediction.