大坝安全监测的ARIMA-BP组合预测模型

摘要

研究大坝安全监测中用于预测坝体变形问题的模型,给出了一种组合模型来预测坝顶位移.单一的ARIMA模型和神经网络预测模型在预测位移问题时精度不高,分析了两种模型的原理和建立方法,利用两种模型单独对时间序列进行拟合与预测,再通过赋予适当的权重,得到新的组合预测模型.通过某水电站工程实例分析表明,该组合预测模型结合了两种模型的优势,提高了模型的预测精度,有广泛的应用前景.%For the issue of dam deformation prediction in dam safety monitoring,a new combined forecasting model is presented.Because of the low accuracy of ARIMA model and neural network(NN) prediction model in deformation prediction,the principles and designs of ARIMA model and BP neural network model are discussed.Based on the discussion,using ARIMA model and BP neural network to fit and predict a single model,a new combined forecasting model is obtained by giving appropriate weight coefficient to combine two methods.The results show that this combined forecasting model has higher prediction accuracy and brings the advantages of single model into play;so it has extensive applied prospects.