粗糙集-RBF神经网络的青岛地铁施工风险评价模型研究

Research on risk evaluation model of Qingdao subway construction based on rough set and RBF neural network

考虑地铁施工风险因素的复杂性和不确定性,从人员、环境、技术和管理4个方面建立地铁建设项目施工风险原始评价指标体系.结合粗糙集理论的属性约简功能和RBF神经网络方法的优势,建立基于粗糙集-RBF神经网络的地铁建设项目施工风险评价模型.对获取的青岛地区22个地铁施工样本数据采用等距离划分算法进行离散化处理,利用粗糙集理论将原始指标体系的27个指标因素约简至17个核心指标因素,作为RBF神经网络的输入变量.构造RBF神经网络并训练样本数据,通过测试样本来验证评价结果.通过实例数据对比表明,模型输出结果与期望值误差在6%以内,同传统的WBS-RBS-AHP评估方法等级一致,证明该模型在地铁施工风险评价中具有可行性和适用性.

Considering the complexity and uncertainty of subway construction risk factors,the original risk evaluation index system of subway construction project is established from four aspects:personnel,environment,technology and management.Combining the attribute reduction function of rough set theory and the advantages of RBF neural network method,the risk evaluation model of subway construction project based on rough set-RBF neural network is established.The data of 22 samples in Qingdao area are discretized by the equidistant partition algorithm.Using rough set theory,27 index factors of the original index system are reduced to 17 core index factors,which are used as input variables of RBF neural network.The RBF neural network is constructed and sample data are trained,and the evaluation results are verified by testing the samples.The comparison of example data shows that the error between the model output and the expected value is less than 6%,and the result is consistent with the traditional WBS-RBS-AHP evaluation method.It proves the feasibility and applicability of the model in risk assessment of subway construction.