基于灰色系统与RBF神经网络的中长期水文预报

及时、准确的中长期水文预报能有效促进水库管理优化。以非汛期各月径流量为预报因子，通过计算所需预报年份与已有径流资料历史年份的预报因子之间的灰色关联度，遴选出与该年灰色关联度较大的年份作为代表年份。采用MATLAB数学软件构建RBF神经网络预报模型，利用选定的代表年份径流量对目标年份汛期径流量进行预报。以清河水库为例,用该模型预报汛期径流量。结果表明，模型简单可操作、运行速度快、预报效果好。

Application of gray system and RBF neural network in the mid-long term hydrological forecasting

The timely and accurate long-term hydrological forecasting can promote the optimization of reservoir management effectively. We regard the monthly runoff of non-flood period as the predictor, calculate the gray correlations of runoff between the future years and the past years, and then select the years with larger gray correlations as example years. The MATLAB is applied to build RBF neural network forecasting model, and the runoff data of the selected example years is used to forecast the flood season runoff of the future forecasting years. Taking Qinghe Reservoir as an example, we forecast the flood season runoff by this forecasting model. The forecasting results show that the model is feasible with quick forecasting speed and satisfying results.