Application of Radial Basis Neural Network on Damage Assessment of Structures

Abstract

Methods to assess damage in a structure and to evaluate their life are very important to ensure the structural integrity of operating plants and structures. The difficulties faced in implementing traditional procedures and the need to develop computer based automated evaluation process motivates the application of soft-computing tools like artificial neural network. The present works focus on development of a computer code to assess structural damages from curvature damage factor using radial basis neural network (RBNN). Comparative study on damage assessment of structures has been carried out between RBNN and the well established back propagation neural network. The structural damage has been introduced by incorporating a stiffness reduction factor. The inverse problem in the damage assessment technique is formulated as optimization problem. The basic idea applied in case of neural network is to train a suitable network to recognize the behavior of the structure with various possible damage scenarios. The steepest descent gradient algorithm is used to train the simple neural network. The developed code has been implemented on a cantilever beam. The results obtained from both the neural network techniques showed the efficiency of the developed code using RBNN. The outcomes of the results are quite encouraging.