人民长江
人民長江
인민장강
YANGTZE RIVER
2014年
19期
41-46
,共6页
杨哲峰%罗林%贾东彦%唐霞
楊哲峰%囉林%賈東彥%唐霞
양철봉%라림%가동언%당하
基坑变形%小波去噪%R/S分析%BP神经网络
基坑變形%小波去譟%R/S分析%BP神經網絡
기갱변형%소파거조%R/S분석%BP신경망락
foundation pit deformation%wavelet de-noising%R/S analysis%BP neural network
基于小波去噪原理,对基坑变形数据小波去噪过程中的相关参数进行优化,对于去噪中的影响因素和作用规律研究有积极意义。利用最优小波去噪将原始监测数据分为趋势项序列和误差项序列,再利用BP神经网络对两序列加以预测,并与传统BP神经网络预测结果进行对比分析。结果表明:采用硬阈值取值和10层小波分解时的去噪效果最好,且通过去噪分离了原始数据的长期性和游离性,增加了数据的可预测性;并且由后期预测结果可知,小波神经网络预测精度要优于传统的BP神经网络预测,具有更高的可信度。
基于小波去譟原理,對基坑變形數據小波去譟過程中的相關參數進行優化,對于去譟中的影響因素和作用規律研究有積極意義。利用最優小波去譟將原始鑑測數據分為趨勢項序列和誤差項序列,再利用BP神經網絡對兩序列加以預測,併與傳統BP神經網絡預測結果進行對比分析。結果錶明:採用硬閾值取值和10層小波分解時的去譟效果最好,且通過去譟分離瞭原始數據的長期性和遊離性,增加瞭數據的可預測性;併且由後期預測結果可知,小波神經網絡預測精度要優于傳統的BP神經網絡預測,具有更高的可信度。
기우소파거조원리,대기갱변형수거소파거조과정중적상관삼수진행우화,대우거조중적영향인소화작용규률연구유적겁의의。이용최우소파거조장원시감측수거분위추세항서렬화오차항서렬,재이용BP신경망락대량서렬가이예측,병여전통BP신경망락예측결과진행대비분석。결과표명:채용경역치취치화10층소파분해시적거조효과최호,차통과거조분리료원시수거적장기성화유리성,증가료수거적가예측성;병차유후기예측결과가지,소파신경망락예측정도요우우전통적BP신경망락예측,구유경고적가신도。
Based on the theory of wavelet de-noising, relevant parameters in the wavelet de-nosing process of foundation pit deformation were studied, which is helpful for the study of influential factors and laws in de-noising. The original monitoring da-ta were classified as trend sequence and error sequence, and the two sequence data were predicted by BP neural network. The prediction results were also compared with the analysis results by traditional BP neural network. The results show that using hard threshold value and 10-layer wavelet decomposition can obtain the best results in de-noising process. The data predictability is increased by data de-noising that separates the long-term and free features of the original data. The subsequent forecast result shows that the forecasting precision of wavelet neural network method in this paper is superior to the traditional BP neural network method.