A Preliminary Study On Intelligent Diagnosis System Of Diseases Of Small And Medium Span Concrete Beam Bridges

The purpose of this paper is to study the intelligent diagnosis system for the disease of medium and small span concrete girder bridges,so as to improve the disease diagnosis abilities of bridge maintenance workers and bridge disease prevention and control.Based on the analysis of common disease diagnosis intelligent systems,the expert system and BP neural network are combined.Artificial intelligence theory is used as a support to study the knowledge of disease diagnosis of small and medium span concrete girder bridges and a BP neural network simulation diagnosis model is established.The main research content of this paper is as follows:1.Design of intelligent diagnosis system for disease of medium and small span concrete girder bridges.The intelligent systems commonly used in the field of disease diagnosis include expert systems,BP neural networks,and their derivation and integration systems.Based on the analysis of their system characteristics and problems,they combine the user types,disease characteristics of medium and small span bridges,and system design principles.The intelligent diagnosis system structure of medium and small span concrete girder bridges was constructed using the combination of expert system and neural network to provide the basis for the formation of system diagnostic parameters and rule tables for subsequent knowledge analysis of diseases.2 Determine the diagnostic parameters,establish a knowledge base of the disease diagnosis rules and data tables.By studying the knowledge of the field of diagnosis of medium and small span concrete disease diagnosis,and based on the knowledge representation rule generation method of the knowledge base of the intelligent diagnosis system,10 diagnostic input parameters such as disease types,component types,material types,environment types,and vehicle conditions of the small and medium span concrete girder bridges were determined.Disease and disease treatment and disease treatment of two kinds of disease output parameters to form a diagnosis rule table for case analysis;inference rules for bp neural network inference;input parameters in the disease rule table to normalize and output parameters “n take 1” binary code pre-processing of data to form a data table,to provide data for subsequent learning and testing of bp network diagnostic model.3 Design BP neural network GUI graphical interface and establish BP neural network diagnosis model.According to the design principle of the BP neural network,the diagnostic parameters are input variables,the disease causes and the disease treatment are output variables,and the number of network layers,the number of nodes at each layer,and the training parameters are determined.For the problems such as slow convergence rate of standard BP network and difficulty in determining the number of hidden layer nodes,using MATLAB neural network toolbox to design the BP network GUI graphical interface that can change the network structure parameters,and determine the optimal network structure by comparing the test results.