基于BP神经网络的火灾受损混凝土构件质量评估

对火灾后混凝土构件受损程度进行鉴定和评估时,影响因素多,各因素间的主次关系和对构件质量的影响难以确定,且评估结果易受人为因素的影响,因此有必要建立一套人工智能基础上的评估系统.将BP人工神经网络技术引入到火灾后混凝土构件受损等级评估研究中,利用BP人工神经网络的自学习、鲁棒性和容错性强的优势,结合专家评估的样本数据对其进行训练,并应用到实际工程中.工程实例结果表明利用BP神经网络进行火灾后混凝土构件质量评估实用可行,且能简化评估计算分析过程,实现评价标准的统一化.

Quality Evaluation Method of Reinforced Concrete Members Damaged by Fire Based on Back-Propagation Neural Network

In this paper,a quality evaluation model of reinforced concrete member damaged by a fire is set up based on back error propagation（BP）neural network technique.By means of such advantages as self-learning,robustness and strong fault tolerant property of BP,the model can be used to assess and determine damaged grades of reinforced concrete members automatically after been trained properly by a series of samples data.At last,the simplicity,feasibility and reliability of the method are verified by the testing of an engineering example.