隔河岩电站进水闸位移的前馈网络预测模型

通过对清江隔河岩水电站进水闸的位移监测资料的分析，发现进水闸顶的位称过程线呈多峰型曲线，应用常规统计模型所建立的预测模型其精度不够理想，在对大坝安全监测资料进行物理推断分析的基础上，针对水电站进水闸顶位移建立了基于前馈人工神经网络模型的预测模型，对闸顶视准线的位移量的前馈人工神经网络模型预测研究表明，其精度较高，通过进一步的研究后可望推广到大坝安全监测实际中去。

Study on feed-forward ANN displacement prediction model for water intake tower of Geheyan hydropower station

Through the analysis on the displacement monitoring data of water intake tower of Geheyan hydroproject on the Qingjiang river, it is found that the crest displacement graph of the tower is in the form of multi-peaks and the predicted accuracy by prediction model established on conventional statistical model is not very satisfactory. Based on physical inference analysis on safety monitoring data of the dam, the feed forward Artificial Neural Network prediction model was established for the crest displacement of water intake tower of hydropower project. The research on the displacement of the crest collimating line of the tower by feed forward Artificial Network model demonstrates that the prediction accuracy is high and can be spread in practical dam safety monitoring after further research.