GA-BP遗传神经网络在地下厂房变形预测中的应用  

Application of BP Neural Network Based GA Optimization in Prediction of Underground Powerhouse Deformation

作　　者：张倩[1] 陈新[1] 周武松[1] 魏成勇[1] 

机构地区：[1]四川大学水利水电学院,四川成都610065

出　　处：《水电能源科学》2016年第6期150-152,164,共4页Water Resources and Power

摘　　要：大型水电站地下厂房多采用分层分区开挖方法,施工期围岩变形受施工程序的影响呈强烈的非线性特点。针对传统回归模型对此类地下厂房围岩变形预测精度较低的问题,考虑了影响围岩变形的主要因素,采用遗传算法优化BP神经网络,结合动态分析法建立了施工期围岩变形预测的GA-BP模型。GA-BP模型在向家坝地下厂房运用结果表明,与回归模型相比,GA-BP预测模型提高了预测结果的精度与稳定性,适合施工现场的监测分析与预测。Underground powerhouse of large hydropower station usually adopts layered and blocked excavation method.The deformation of surrounding rock which is influenced greatly by the construction program shows strong nonlinear characteristics during the construction period.Aiming at the low prediction accuracy of the traditional regression model,some main factors affecting the deformation of surrounding rock are taken into consideration.Genetic algorithm is used to optimize BP neural network.Combined with dynamic analysis,the GA-BP model for predicting the deformation of surrounding rock during the construction period was established and it was applied to underground powerhouse of Xiangjiaba hydropower station.Compared with regression model,the results show that the GA-BP prediction model can improve prediction precision and stability,and is suitable for the monitoring and forecasting in construction spots.

关 键 词：地下厂房 施工期 围岩变形 遗传算法 BP神经网络 

分 类 号：TV731.6[水利工程—水利水电工程]