摘要
文章针对区域性地面沉降问题,从传统灰色模型出发,对其背景值进行改进,建立动态定权灰色模型,并结合BP神经网络建立组合预测模型。该组合模型克服了单一灰色模型或BP网络模型不能很好地处理原始沉降数列波动较大的问题。对某地一水准点的连续多年沉降量建立组合模型,对后期沉降进行预测,将预测结果与二等水准测量结果进行比较,主要采用后验差检验法和相对误差比较对模型模拟和预测结果进行了检验,结果表明该组合预测模型可以大大提高预测精度和鲁棒性。
In view of the regional land subsidence,considering the traditional grey model,the background value of the model is improved.Then a prediction model for land subsidence is established based on the combination of the dynamic grey model with fixed weight and BP network.The combination model overcomes the disadvantages of the single grey model or BP network model that can not handle the fluctuant original subsidence series well.The combination model is used to predict the yearly subsidence of a standard point at late stage.The prediction results are compared with those of second-order leveling measurement.Posterior difference test method and relative error comparison are used to test the model simulation and prediction results.The results show that the combination model can improve the prediction accuracy and robustness greatly.
出处
《合肥工业大学学报（自然科学版）》
CAS
CSCD
北大核心
2013年第3期361-364,共4页
Journal of Hefei University of Technology：Natural Science
基金
国家自然科学基金资助项目(41071273)
江苏高校优势学科建设工程资助项目(SZBF2011-6-B35)
关键词
灰色模型
BP网络
组合预测
地面沉降
grey model
BP network
combination prediction
land subsidence
