优化神经网络模型在建筑物变形监测预测分析中的应用  

Application of Optimized Neural Network Model in Building Deformation Monitoring and Prediction Analysis

作　　者：朱波 孙曼曼 ZHU Bo;SUN Manman(Zhejiang Yuanwang Land Survey and Planning Design Co.,Ltd.,Hangzhou 311203,China;Zhejiang Mingkun Survey and Planning Design Co.,Ltd.,Hangzhou 310012,China)

机构地区：[1]浙江远望土地勘测规划设计有限公司,浙江杭州311203 [2]浙江明坤勘测规划设计有限公司,浙江杭州310012

出　　处：《测绘与空间地理信息》2023年第9期179-182,共4页Geomatics & Spatial Information Technology

摘　　要：工程施工过程中建筑结构变形大小是施工安全信息的直接体现,对建筑变形趋势进行准确预测,能够提前规避施工风险,保障施工安全。本文将小波去噪和BP神经网络模型充分结合,以某工程连续47期沉降监测数据为原始数据序列,构建去噪前原始序列BP模型;同时对原始序列进行小波去噪处理,构建去噪后BP模型。将去噪前后BP神经网络模型预测结果进行对比分析,发现小波去噪后预测模型精度相对更高,预测结果与实测数据更为匹配,能够对工程施工过程中建筑结构变形进行更为精准的预测分析。The magnitude of structural deformation during construction is a direct reflection of construction safety information.Accu-rately predicting the trend of building deformation can avoid construction risks in advance and ensure construction safety.In this pa-per,Wavelet denoising and BP neural network model are fully combined,and 47 consecutive periods of settlement monitoring data of a project are taken as the original data sequence to build the original sequence BP model before denoising;It simultaneously performs wavelet denoising on the original sequence and constructs a denoised BP model.By comparing and analyzing the prediction results of the BP neural network model before and after denoising,it was found that the prediction model after Wavelet denoising has relatively higher accuracy,and the prediction results are better matched with the measured data,which can provide more accurate prediction and analysis of building structure deformation during construction.

关 键 词：神经网络模型 小波去噪 变形预测 数据分析 

分 类 号：P25[天文地球—测绘科学与技术] TB22[天文地球—大地测量学与测量工程]