论文部分内容阅读
针对大坝安全监控中单一预测模型预测精度较低的问题,基于多元线性回归预测模型和BP神经网络模型,应用熵原理提出一种新的线性组合预测模型,并结合某大坝渗透压力实际观测资料对该组合预测模型进行实用性检验。结果表明,短期内三种模型均具有较高的预测精度,但预测长度增加后,该组合预测模型预测精度更高。
Aiming at the low prediction accuracy of single prediction model in dam safety monitoring, a new linear combination prediction model based on multivariate linear regression prediction model and BP neural network model is proposed based on entropy principle. Combined with the actual observation of osmotic pressure of a dam, Data for the combined forecasting model for practical testing. The results show that all the three models have higher prediction accuracy in the short term, but the predictive accuracy of the combined prediction model is higher after the prediction length is increased.