铁道科学与工程学报
鐵道科學與工程學報
철도과학여공정학보
JOURNAL OF RAILWAY SCIENCE AND ENGINEERING
2015年
4期
839-844
,共6页
桥梁耐久性预测%GM(1,1)模型%BP 神经网络
橋樑耐久性預測%GM(1,1)模型%BP 神經網絡
교량내구성예측%GM(1,1)모형%BP 신경망락
bridge durability prediction%GM(1,1)model%BP Neural network
针对混凝土桥梁耐久性历史评估数据的特点,提出一种基于 BP 神经网络与 GM(1,1)模型的桥梁耐久性组合预测方法。通过 GM(1,1)模型,以部分数据作为样本进行预测,在此基础之上,引入 BP 神经网络对预测的残差序列进行处理,旨在克服单一预测模型的不足,取得更高的预测精度。算例表明,本文算法精度明显高于传统 GM(1,1)模型,与类似算法相比,精度上也有所提高。
針對混凝土橋樑耐久性歷史評估數據的特點,提齣一種基于 BP 神經網絡與 GM(1,1)模型的橋樑耐久性組閤預測方法。通過 GM(1,1)模型,以部分數據作為樣本進行預測,在此基礎之上,引入 BP 神經網絡對預測的殘差序列進行處理,旨在剋服單一預測模型的不足,取得更高的預測精度。算例錶明,本文算法精度明顯高于傳統 GM(1,1)模型,與類似算法相比,精度上也有所提高。
침대혼응토교량내구성역사평고수거적특점,제출일충기우 BP 신경망락여 GM(1,1)모형적교량내구성조합예측방법。통과 GM(1,1)모형,이부분수거작위양본진행예측,재차기출지상,인입 BP 신경망락대예측적잔차서렬진행처리,지재극복단일예측모형적불족,취득경고적예측정도。산례표명,본문산법정도명현고우전통 GM(1,1)모형,여유사산법상비,정도상야유소제고。
Predicting the bridge durability accurately is of great significance for saving maintenance costs and en-suring traffic safty.For this reason,this paper has proposed a combined algorithm based on BP neural network and GM(1,1)model on the basis of the data character of bridge durability.By taking some of the evaluation as a sam-ple,the GM(1,1)model and BP network were used to deal and correct the residual,which aims at improvs the accuracy of the alqorithm and overcome the defect of the single model.Finally,an example shows that the accura-cy of combined algorithm is obviously better than the single GM(1,1)model and also superior to similar method.