基于LM-BP和SVR的倾倒变形体变形预测

为了深入了解黄登水电站1号倾倒变形体的变形趋势，采用LM BP神经网络和SVR进行变形预测研究。基于倾倒变形体的实际变形监测资料，对位移、降雨、库水位、温度等资料进行分析，以库水位、降雨量、温度、时间作为输入参数，以位移变形作为输出参数，构建LM BP神经网络模型和SVR模型，对部分监测数据进行（先行学习）训练，对后续的监测数据进行验证预测，预测预报了研究测点的变形情况。分析结果表明，2个模型精度都比较高，LM BP神经网络模型的最大误差为2.53%，SVR模型的最大误差为4.35%，预测方法有效。

Deformation prediction of toppling deformed slope based on LMBP and SVR

To gain a deeper understanding of the deformation trend of No. 1 toppling deformed slope in the Huangdeng Hydropower Station, the LMBP neural network and the SVR were used to conduct the deformation prediction. Based on the practical deformation monitoring data of the toppling deformed slope in the Huangdeng Hydropower Station, this study analyzed the data of displacement, rainfall, reservoir water level, temperature, etc. Then the reservoir water level, rainfall, temperature, and time were taken as input parameters and the displacement was used as the output parameter to construct the LMBP neural network model and the SVR model. Two models were trained by a part of the monitoring data, and the subsequent monitoring data was used for verification and forecasting, which predicted the deformation of the measuring point in advance. The results show that the accuracy of the two models is higher, the maximum error of the LM BP neural network model is 2.53% and the maximum error of the SVR model is 4.35%, which demonstrate that two prediction methods are both effective.