基于RS-SVM的建筑施工项目安全预警模型

针对目前建筑施工项目安全风险管理的现状,应用粗集理论对建筑施工项目的安全因素进行预处理,将处理后的信息结构作为支持向量机的输入数据进行训练和预测,构建建筑施工项目安全风险预警模型,并在小样本条件下,与BP神经网络进行对比分析。结果表明,RS-SVM预测模型的最小均方根误差为0.011 5,BP神经网络的均方根误差为0.070 7,RS-SVM预警模型的预测精度、泛化能力明显优越于BP神经网络学习方法。

Safety early-warning model of construction project based on rough set and support vector machine

In this paper,the finite difference strength reduction criteria was used to determine slope failure analysis.Taking the high rock slope in the head of dam for the example,Flac3D was used to analysis the high slope stability before and after water based on the basis of Survey.The safety coefficient was solved by using the strength reducing method.The results showed that the high slope is stability and finite difference strength reduction can resolve quantitative rock slope stability problems.