摘要
为了降低建筑施工安全管理事故率,减少事故损失,构建PCA与BP神经网络相结合的施工安全预测模型;在分析建筑施工特点,提炼建筑施工风险“4M”因素体系的基础上,本文以近十年的30组样本数据为例,利用SPSS主成分分析法对数据样本降维,将多个变量转化为少数几个能反映原始信息的互不相关的综合变量(成分),利用Matlab软件编写BP神经网络程序,对数据样本进行训练和检验。研究表明,PCA-BP预测模型的预测相对误差为3.728%,而直接运用BP神经网络预测的相对误差为4.618%,说明使用PCA-BP预测模型进行预测是可行的。
In order to reduce the accident rate of construction safety management and reduce the accident loss,the construction safety prediction model combined with PCA and BP neural network was built to analyze the characteristics of construction and extract the 4M construction risk factor system.Taking 30 groups of sample data in recent ten years as an example,SPSS principal component analysis was used to reduce the dimension of the data samples,and multiple variables were transformed into a few independent comprehensive variables(principal components)that could reflect the original information.Matlab software was used to write BP neural network program for training and testing the data samples.The results show that the relative error of PCA-BP prediction model is 3.728%,while the relative error of BP neural network is 4.618%,indicating that the PCA-BP prediction model is feasible for prediction.
作者
笪可宁
刘廷轩
Da Kening;Liu Tingxuan(School of Economics and Management Shenyang Chemical Engineering University,Shenyang Liaoning 110142,China)
出处
《城市建筑》
2021年第27期196-198,共3页
Urbanism and Architecture
