优化BP神经网络的位移预测模型

A displacement prediction model based on improved particle swarm-BP neural network algorithm

针对大坝位移预测常规方法存在的问题,基于改进粒子群算法的BP神经网络(IPSO-BP)的大坝位移预测方法,通过IPSO对常规BP神经网络的权值和阈值进行优化,弥补了BP网络的不足,保证了预测精度。以2011-12-21—2013-06-27观测得到的某混凝土重力坝某一典型坝段坝顶的顺河向位移值为研究对象,建立基于IPSO-BP的大坝预测模型并进行仿真分析研究。同时,为了验证该模型的拟合及预测效果,建立PSO-BP模型、利用最小二乘法求解参数的统计模型进行对比分析。上述研究结果表明,此模型预测精度优于常规模型且拟合效果好、预测结果的平均相对误差小,说明此方法有效可行。

According to the existing problems of the conventional method of dam displacement prediction, BP neural network based on improved particle swarm optimization （IPSO-BP） is put forward to predict the dam displacement. The weights and threshold of the conventional BP neural network are optimized by IPSO, thus making up the shortage of BP network and improving the prediction accuracy. The observed longitudinal displacement of the typical section of a concrete gravity dam crest from January 10, 2012 to July 31,2012 is taken as the research object. And based on the IPSO-BP prediction model the simulation analysis is carried out. At the same time, in order to verify the effect of fitting and prediction of the model, a statistical model using the least squares method to make parameter analysis and PSO-BP model are developed. It is found from the prediction accuracy given by this model that the model is superior to the conventional model; the fitting effect is the best; and the average relative error of the prediction results is the minimum. Therefore, it shows that this method is effective and feasible in prediction and analysis of dam engineering.