前馈人工神经网络法在大坝安全监控中的应用

应用预测模型来监控大坝复杂的工作性态是一条有效途径。但因大坝的工作条件复杂、影响因素众多，给应用精确的数学模型监控大坝工作性态带来了困难。为此，应用人工神经网络模型隐式的数学表达形式，提出并建立了基于交替学习迭代算法的人工神经网络模型，并结合清江隔河岩水电站的实际，研究了这种模型在大坝基础渗流量及进水闸顶位移预测中的实际应用，其误差收敛快，预报精度较高。通过进一步的研究后，这种模型可望为大坝安全性态的实时在线监控提供有力的技术支持。

Application of the feed-forward artificial neural network approach to dam safety monitoring

Using the forecasting model to monitor the com plex working behaviors of dam is an effective way. Because of its complex working conditions and many affecting factors, it is difficult to monitor dam behaviors by using precise mathematical model. Based on the implicit mathematical expression and the information processing of nonlinear, self-adaptation, self-learning, etc., of artificial neural network (ANN) approach, this paper presents the ANN model based on the training method of learning into groups. Combined the practice of the Geheyan hydropower station in the Qingjiang River, the application of the ANN model to the prediction of the seepage quantities and the displacement of the top of the inlet sluice of the dam foundation is studied in this paper. It is high accuracy in the prediction result through the ANN method. The research results demonstrate that this approach can supply some powerful technical assistance for on-line dam safety monitoring.