基于神经网络的模糊综合评判在建筑结构可靠性评估中的应用

Study on Application of Multi-grade Fuzzy Comprehensive Evaluation Based on Artificial Neural Network to Reliability Appraisal of Existing Buildings

提出了一种对结构可靠性进行鉴定的新方法,该方法采用了人工智能技术来减少鉴定的主观性.由于对结构可靠性的认识存在一定的模糊性,采用了多级模糊综合评判进行可靠性鉴定.根据现行鉴定规程,将结构系统分为构件、子单元、鉴定单元三个层次.首先,利用训练后的神经网络求出构件对于不同可靠性等级的隶属度,然后通过多级模糊综合评判求得子单元及鉴定单元对于不同可靠性等级的隶属度.同时,对于检测参数的选取也进行了讨论.图2,表2,参8.

A new method to appraise the reliability of existing buildings rationally was provided. Artificial inteUigenee was adopted to reduce subjective factors in reliability appraisal. Multi-grade fuzzy comprehensive evaluation was applied beeanse of the fuzziness in reliability appraisal. Based on the current national standards, the way to appraise the reliability of an existing building through three levels was explored, which were structural members, subsidiary units and units. Firstly, artificial neural network was adopted in order to get the membership grades of structural members by the stone standard. Secondly, after getting the membership grades of structural members, the membership grades of subsidiary units could be figured out through fuzzy comprehensive evaluation. Thirdly, the membership grades of the whole unit could be worked out through two-stage fuzzy comprehensive evaluation. Furthermore, how to select the parameters that rejecting the state of a target structure was discussed.