论文部分内容阅读
为了解决当前我国地铁施工过程中所面临的安全风险评估问题,提出了一种基于粗糙集和RBF神经网络的地铁施工安全风险评估模型。在分析地铁施工安全风险评估指标的基础上,从地质条件、施工技术、施工管理等方面选择了33个变量指标。针对传统神经网络收敛速度慢、容错性差、结果并不唯一的缺点,利用粗糙集理论的属性约简方法使RBF神经网络的输入数据减少且不相关,并利用长沙、武汉、杭州、昆明、北京、上海、广州、重庆的28个地铁工程项目的问卷调查数据实现模型的训练及检测。结果表明,采用粗糙集方法约简属性能使RBF网络的输入数据从33个减少至15个,用经过粗糙集约简后的样本集作为神经网络的训练样本集可以有效地简化神经网络的结构,减少训练步数与训练时间,并提高网络的学习速度和判断准确率。同时,经过粗糙集约简后的神经网络的收敛速度和计算精度能够满足地铁施工安全风险评估的需要。通过粗糙集与RBF神经网络相结合所构建的耦合模型可以识别地铁施工过程的安全状态,进而有针对性地完善地铁施工的相关安全技术。
In order to solve the problem of safety risk assessment in subway construction in our country, a risk assessment model of subway construction safety based on rough sets and RBF neural network is proposed. Based on the analysis of the subway construction safety risk assessment index, 33 variable indicators were selected from the aspects of geological conditions, construction technology and construction management. Aiming at the shortcomings of traditional neural network, such as slow convergence rate, poor fault tolerance and not the only result, the input data of RBF neural network is reduced and irrelevant by using the attribute reduction method of rough set theory. And by using Changsha, Wuhan, Hangzhou, Kunming and Beijing , Shanghai, Guangzhou, Chongqing, 28 subway project survey data to achieve the model training and testing. The results show that using rough set method to reduce attributes can reduce the input data of RBF network from 33 to 15, and using rough set reduced sample set as neural network training sample set can effectively simplify the structure of neural network, Reduce the number of training steps and training time, and improve the learning speed of the network and to determine the accuracy. At the same time, the convergence rate and accuracy of the neural network after rough set reduction can meet the needs of subway construction safety risk assessment. The coupling model constructed by combining rough set and RBF neural network can identify the safety status of subway construction process, and then improve the related safety technology of subway construction.