科技广场
科技廣場
과기엄장
SCIENCE TECHNOLOGY PLAZA
2015年
3期
34-39
,共6页
时间序列%GM%SVM%大坝变形%影响因子
時間序列%GM%SVM%大壩變形%影響因子
시간서렬%GM%SVM%대패변형%영향인자
Time Series%GM(1,1)%SVM%Dam Deformation%Impact Factors
针对单一预测模型都存在各自优缺点的问题,本文提出时序回归GM-SVM模型,以达到最优的变形预测效果。首先对灰色模型中的灰参数导致的时间序列残差进行研究,形成时间序列模型,根据时间序列模型对其残差进行最优化设计,获取时间序列估计模型,并将该模型与支持向量机进行无缝融合以建立新的预测模型,然后根据该预测模型对观测的大坝变形影响因子进行训练和预测,并将预测结果与实际的变形值进行对比分析,经过实例分析确定该模型的预测结果更加接近实际观测值,说明该模型更加适用于基于大坝变形影响因子的变形分析。
針對單一預測模型都存在各自優缺點的問題,本文提齣時序迴歸GM-SVM模型,以達到最優的變形預測效果。首先對灰色模型中的灰參數導緻的時間序列殘差進行研究,形成時間序列模型,根據時間序列模型對其殘差進行最優化設計,穫取時間序列估計模型,併將該模型與支持嚮量機進行無縫融閤以建立新的預測模型,然後根據該預測模型對觀測的大壩變形影響因子進行訓練和預測,併將預測結果與實際的變形值進行對比分析,經過實例分析確定該模型的預測結果更加接近實際觀測值,說明該模型更加適用于基于大壩變形影響因子的變形分析。
침대단일예측모형도존재각자우결점적문제,본문제출시서회귀GM-SVM모형,이체도최우적변형예측효과。수선대회색모형중적회삼수도치적시간서렬잔차진행연구,형성시간서렬모형,근거시간서렬모형대기잔차진행최우화설계,획취시간서렬고계모형,병장해모형여지지향량궤진행무봉융합이건립신적예측모형,연후근거해예측모형대관측적대패변형영향인자진행훈련화예측,병장예측결과여실제적변형치진행대비분석,경과실례분석학정해모형적예측결과경가접근실제관측치,설명해모형경가괄용우기우대패변형영향인자적변형분석。
The research of time series regression model is proposed to improve the deformation forecast aim-ing at the disadvantages of single forecasting model. The paper researches the residual of time series caused by the grey parameters of grey model and gives the time series model. The time series estimation model is acquired ac-cording to the optimal design of time series residual. This model is integrated with the support vector machine (SVM) to construct the new prediction model. The dam deformation influencing factors are trained and predicted according to this model. The proposed model is close to real observation values according to the comparative anal-ysis between the predicted results and the actual deformation values. The example shows that this model is more suitable for the dam deformation analysis based on its influencing factors.