枯水径流预报的最优模糊划分自激励门限自回归模型

Optimal Fuzzy Partitioned Self excited Threshold Autoregressive Model for Low Flow Forecast

介绍了建立最优模糊划分自激励门限自回归模型的观点、理论和方法。该建模方法克服了传统方法确定门限自回归模型门限值的困难，为用线性化方法对非线性时间序列的建模和预报提供了一种新的数学方法，具有广泛的实用性。建立了６种数据处理方式下的泾河、北洛河、渭河及大通河主要水文站的候、旬、月枯水径流预报模型。模型评定与检验显示，将时段平均流量序列转换为消除年周期的模比系数序列建立的模型优于其他类型的模型，可作为作业预报模型。

The importance of low flow forecast was briefly described in this paper.The standpoint,theory and methodology to develop optimal fuzzy partitioned self excited threshold autoregressive (OFPSETAR) model were introduced,and the OFPSETAR models for five day,ten day and monthly low flow forecast were developed for the main hydrologic stations on the Jinghe River,Beiluo River,Weihe River and Datong River,respectively.The model identification and verification showed that the model developed by using the series of ratio of measured flow to its seasonal average had an advantage over others,and might be applied to operational forecast.