Diagnosis of fault and condition monitoring of dynamic structures using the multiple adaptive-neuro-fuzzy inference system technique

Abstract

Fault diagnosis and condition monitoring of dynamic structures are analysed and discussed in the current article. To diagnose the crack in the structure, the multiple adaptive-neuro-fuzzy inference system (MANFIS) methodology has been applied. The adaptive neuro-fuzzy controller has an input layer, hidden layers, and an output layer. The input layer is the fuzzy layer. The other layers are neural layers. The inputs to the fuzzy layer are relative deviation of the first three natural frequencies and relative values of percentage deviation for the first three mode shapes. The interim outputs of the fuzzy layer are inputs to the neural layers of the neuro-fuzzy controller. The final outputs of the MANFIS controller are relative crack depth and relative crack location. Several hundreds of fuzzy rules and neural network training patterns are derived using natural frequencies, mode shapes, crack depths, and crack locations. Experimental results have been obtained using real mobile robots during navigation. Comparison of the simulation and experimental results showed that there is good agreement between them. This methodology can be effectively used for condition monitoring of dynamic structures.