C-ANFIS based fault diagnosis for voltage-fed PWM motor drive systems

Since most of the induction motors are operated by the inverter, an unexpected fault of the inverter can cause serious troubles such as downtime of equipment, heavy loss, and etc. Therefore, the studies on the robust drive system for induction motors to protect the system under the fault modes are gaining more interests. This paper investigates the fault diagnosis for open-switch damages in a voltage-fed PWM motor drive system. For diagnosing the conditions of a inverter, we transform the current signal to the d-q axis. And then, we obtain the features consisting of the trajectories of d-q phase currents for each fault mode. In the ideal cases, a set of fault modes can be classified by using the shape of these trajectories. There are, however, many variational elements such as load torque and the electrical/mechanical variable parameters. So, we propose a robust diagnosis method based on the neuro-fuzzy algorithm. For this, we adopted the Clustering Adaptive Neuro Fuzzy Inference System(C-ANFIS) to recognize the various and vague fault patterns. Finally, we implement the method for the diagnosis module of the inverter with MATLAB and show its usefulness.