Dynamic Analysis and Fault Detection of Multi Cracked Structure Under Moving Mass Using Intelligent Methods

The present thesis explores an inclusive research in the era of moving load dynamic problems. The responses of vibrating structures due to the moving object and different methodologies for damaged identification process have been investigated in this analogy. The theoretical-numerical solutions of the multi-cracked structure with different end conditions subjected to transit mass have been formulated. The Runge-Kutta fourth order integration approach has been applied to determine the response of the structures numerically. The effects of parameters like mass and speed of the traversing object, crack locations, and depth on the response of the structures are investigated. The proposed numerical method has been verified using FEA and experimental investigations. The novel damage prediction processes are developed on the knowledge-based concepts of recurrent neural networks (RNNs) and statistical process control (SPC) methods as inverse approaches. The Jordan’s recurrent neural networks (JRNNs), Elman’s recurrent neural network (ERNNs), the integrated approach of the JRNNs, and ERNNs, the autoregressive (AR) process in the domain of SPC and the combined hybrid neuro-autoregressive process have been developed to identify and quantify the faults in the structure. The accuracy and exactness of each approach has been verified with experiments and FEA. The proposed methods can be useful for the online condition monitoring of faulty cracks in structures.