摘要
在混凝土公路路基施工过程中,为了控制施工进度,同时保证路基的稳定与适用,需要对路基沉降进行监测,并通过已有沉降数据推测后续沉降数值。本文提出了一种新型预测路基沉降的方法,将模糊神经网络和小波包分解相结合,针对位移沉降序列的时序性和偶然性,利用模糊神经网络强大的记忆性,建立沉降预测模型。通过小波包对沉降数值序列进行分解,得到不同属性和频率的沉降序列,分别通过模糊神经网络对各个子序列进行预测,将子序列预测结果叠加得到预测结果。
In order to control the construction progress and ensure the stability and applicability of the roadbed of concrete pavement, the roadbed settlement needs to be monitored, and the subsequent settlement data needs to be predicted through the existing settlement data. A new prediction method for subgrade settlement is put forward, which combines fuzzy neural network and wavelet packet decomposition. Aiming at the timeness and contingency of displacement settlement sequence, the fuzzy neural network’s strong memory is used to establish settlement prediction model. To decompose the settlement numerical sequence by wavelet packet, settlement sequence of different attributes and frequency, respectively through the prediction of each sub sequence of fuzzy neural network, the sequence prediction results are superimposed to obtain the forecast result.
出处
《混凝土与水泥制品》
北大核心
2018年第6期81-84,共4页
China Concrete and Cement Products
基金
地质灾害防治与地质环境保护国家重点实验室项目(SKLGP2014K011)
中国建筑第五工程局有限公司项目(030017D119)
关键词
小波包分解
路基沉降
神经网络
时间序列
Wavelet packet decomposition
Subgrade settlement
Neural network
Time series
