论文部分内容阅读
径流预测为流域水资源的合理开发利用与统筹配置提供依据。运用多元线性回归、主成分回归、BP神经网络及主成分分析和BP神经网络相结合的方法,对新疆呼图壁河流域石门水文站2009—2011年各月径流量进行预测,并采用相关系数、确定性系数及均方根误差对各模型预测精度进行比较。结果表明:(1)神经网络等智能算法具有高速寻优的能力,对短时间尺度的月径流量的预测结果较好;(2)主成分回归等常规算法能充分反映出某地区径流的年际的稳定性,对全年径流总量的模拟精度较高;(3)主成分分析和BP神经网络相结合的方法,提高了神经网络的收敛速度,同时降低了局部极值的影响,优于简单的BP神经网络,适用于呼图壁河月径流量预测。
Runoff prediction provides the basis for the rational development and utilization of water resources in the basin as well as overall allocation. By using multiple linear regression, principal component regression, BP neural network, principal component analysis and BP neural network, the monthly runoff of Shihmen Hydrological Station in Hutubitai Basin of Xinjiang from 2009 to 2011 was predicted and the correlation coefficient , Deterministic coefficient and root mean square error to compare the prediction accuracy of each model. The results show that: (1) Intelligent algorithms such as neural networks have the ability of high-speed optimization, and the prediction of monthly runoff of short-time scale is better; (2) The conventional algorithms such as principal component regression can fully reflect the year (3) The combination of principal component analysis and BP neural network improves the convergence speed of neural network and reduces the influence of local extreme value For simple BP neural network, it is suitable for forecasting the monthly runoff of Hutubi River.