The Study Of RBF Neural Network's Application In The Plane Truss Damage Detection

Modern engineering structures are proceeding at the direction of maximization, complication, automation and continuousness. The complicated environments of service threat functional safety of the structure. To guarantee the safety and avoid disaster, the scholars, engineers and technicians are paying attention to the research of identification for the damage of structure real-timely, on-line and accurately. On the basis of a great deal of relevant structural damage detections, this text discusses on the theories of the structural damage diagnosis according to the modal parameter, put forward using theory of flexibility difference curvature to analyze the damage detection. ANSYS is used to make the modal analysis of a plane truss structure model, then APDL is used to compile procedures to calculate flexibility difference curvature .By simulative damage detection to the plane truss, it proves the feasibility of the methods of damage detection based on the flexibility difference curvature. According to some current outcomes of damage detection using the artificial network as well as the analysis of BP network and RBF network, carrying forward the study of the plane truss damage detection by RBF network and together with the steps of specific methods. And then to make use of MATLAB software to make the RBF network of the plane truss, inputting the first flexibility difference curvature, and outputting the corresponding extent of damage (damage location and damage degree), to train the network by many study samples, therefore, setting out the performance of network by the selected samples, the results show that the RBF network does a good job, which can accomplish the detection and judgment of the damaged location of the plane truss as well as the degree of damage. And only the first flexibility value is need ,so the actual measuring workload id greatly decreased ,it shows this method is simple and feasible in project. From that time on, a set of technology of damage detection to the plane truss based on the RBF network methods are formed. However, the study of the application of artificial neural network to structural damage detection is still in the period of theory analysis and numerical simulation and experimentation. So the further study is needed to find the better artificial neural network and more available parameter of damage detection, then the network can be applied to actual engineering step by step in order to accelerate this method's development.