基于小波分解的径向基神经网络径流预测研究

摘要

Kashgar River is located in the southwestern margin of the Tarim Basin in Xinjiang ,The water shortages di-rectly affect agricultural productions ,so it is necessary to make relatively accurate predictions of the river runoff .Here , the measured annual runoff data of Kashgar River from 1956 to 2000 was chosen in analysing the trends and different cy-cles of the runoff by using wavelet decomposition .The results provided input data for the runoff forecasting ,based on which an annual runoff model of RBF neural network was established .The study proved that RBF neural network predic-tion model had higher accuracy than that of conventional models ,and it could more accurately predict the annual runoff of Kashgar River .%噶尔河流域位于新疆塔里木盆地西南缘，流域严重缺水，直接影响到流域的农牧业生产，因此需要对其径流量做出相对准确的预测。选用喀什噶尔河流域1956年-2000年实测年径流量，运用小波分解方法揭示径流的趋势性和周期性，并利用周期性结果为径流预测提供数据输入依据，建立径向基神经网络预测模型，与传统径向基神经网络预测模型相比，精度更高，能够更加准确的预测喀什噶尔河流域年径流量。