基于TESAOC的ARIMA-BP组合模型在混凝土坝变形预测中的应用
|
摘要
针对基于最优加权法的ARIMA-BP组合预测模型在对混凝土坝位移变形进行预测时模型权重固定、时效性不强的缺点,引入时效函数自适应优化方法(TESAOC),挖掘监测样本的时效特征,从而实现单一模型权系数随时间的动态变化;利用移动算术平均法筛选出最佳计算样本,减小数据突变等因素带来的不利影响,同时对预测偏差落在不同范围内的时间序列权重进行模糊补偿处理,控制其对组合预测结果的影响,最终建立基于TESAOC的ARIMA-BP组合预测模型.实例分析表明,该组合模型的预测效果在精度和时效性方面均优于最优加权法,更符合工程实际情况,有更好的应用前景.
In light of the weakness of fixed weights and weak timeliness of the ARIMA-BP combined forecasting model based on the optimal weighting for concrete dam displacement,self-adapt optimization combination method integrating time effect function(TESAOC)is imported in this thesis to mine time characteristics of the monitoring samples,so as to realize the dynamic change of the weight coefficients of a single model over time.A method of moving arithmetic average is used to screen out the optimum calculation sample in order to reduce the adverse effects of data mutation and other factors.Meanwile,the weights of time series are dealt with fuzzy compensation according to the prediction deviation in different ranges;and the influence of them on the combined forecasting results is controlled.Finally,the ARIMA-BP combined forecasting model based on TESAOC is established.Experimental results indicate that the prediction effect of the combined model based on TESAOC is better than that based on the optimal weighting method in terms of accuracy and timeliness;and it is better appropriate to the engineering practical situation,which means that the forecasting model in this thesis has extensive applied foreground.
In light of the weakness of fixed weights and weak timeliness of the ARIMA-BP combined forecasting model based on the optimal weighting for concrete dam displacement,self-adapt optimization combination method integrating time effect function(TESAOC)is imported in this thesis to mine time characteristics of the monitoring samples,so as to realize the dynamic change of the weight coefficients of a single model over time.A method of moving arithmetic average is used to screen out the optimum calculation sample in order to reduce the adverse effects of data mutation and other factors.Meanwile,the weights of time series are dealt with fuzzy compensation according to the prediction deviation in different ranges;and the influence of them on the combined forecasting results is controlled.Finally,the ARIMA-BP combined forecasting model based on TESAOC is established.Experimental results indicate that the prediction effect of the combined model based on TESAOC is better than that based on the optimal weighting method in terms of accuracy and timeliness;and it is better appropriate to the engineering practical situation,which means that the forecasting model in this thesis has extensive applied foreground.
引文
[1]梅泽宇,许青,康飞.基于人工蜂群算法-逐步回归模型的大坝变形监测[J].防灾减灾工程学报,2013,33(6):651-656,670.
[2]齐银峰,谭荣建.基于改进粒子群优化算法的BP神经网络在大坝变形分析中的应用[J].水利水电技术,2017,48(2):118-124.
[3]Su Huaizhi,Chen Zhexin,Wen Zhiping.Performance Improvement Method of Support Vector Machine-Based Model Monitoring Dam Safety[J].Structural Control&Health Monitoring,2016,23(2):252-266.
[4]顾冲时,吴中如.大坝与坝基安全监控理论和方法及其应用[M].南京:河海大学出版社,2006.
[5]王成,胡添翼,顾艳玲,等.大坝安全监测的ARIMA-BP组合预测模型[J].三峡大学学报(自然科学版),2018,40(1):20-24.
[6]戴明远.基础信息论[M].上海:同济大学出版社,2003.
[7]李默涵.数据时效性的理论和算法研究[D].哈尔滨:哈尔滨工业大学,2016.
[8]殷建成,袁宗明,岑康.天然气需求自适应优化组合预测模型的改进[J].油气储运,2005(10):17-20,3,61.
[9]方向华,王文杰,汤兵勇.基于模糊补偿的组合预测方法研究[J].中国储运,2006(1):77-79.
[2]齐银峰,谭荣建.基于改进粒子群优化算法的BP神经网络在大坝变形分析中的应用[J].水利水电技术,2017,48(2):118-124.
[3]Su Huaizhi,Chen Zhexin,Wen Zhiping.Performance Improvement Method of Support Vector Machine-Based Model Monitoring Dam Safety[J].Structural Control&Health Monitoring,2016,23(2):252-266.
[4]顾冲时,吴中如.大坝与坝基安全监控理论和方法及其应用[M].南京:河海大学出版社,2006.
[5]王成,胡添翼,顾艳玲,等.大坝安全监测的ARIMA-BP组合预测模型[J].三峡大学学报(自然科学版),2018,40(1):20-24.
[6]戴明远.基础信息论[M].上海:同济大学出版社,2003.
[7]李默涵.数据时效性的理论和算法研究[D].哈尔滨:哈尔滨工业大学,2016.
[8]殷建成,袁宗明,岑康.天然气需求自适应优化组合预测模型的改进[J].油气储运,2005(10):17-20,3,61.
[9]方向华,王文杰,汤兵勇.基于模糊补偿的组合预测方法研究[J].中国储运,2006(1):77-79.