Diagnosis of an Inverter by Clark Transform Technique Based on Neural Network

Inverters are widely used as electrical drive systems in many industrial applications. The inverter IGBT switch open-circuit fault; object of our study; is one type of the inverter faults that attracts the attention of many researchers. In this paper an open-circuit fault detection technique is proposed using the artificial neural network (ANN) aided by the Clarke transform. The technique is presented according to two approaches. The first approach calculates the Clarke stator currents and other quantities such as the intervals fault angles a and the trajectory surface S. Once all these quantities are obtained, they are then introduced to the ANN structure for the learning process; the output obtained from the ANN corresponds to the exact angle related to the inverter IGBT switch open-circuit fault. The second approach follows exactly the same calculation steps as the first approach except that in this case all the computation is performed within the ANN architecture. To illustrate the effectiveness of the technique and validate the results, some experimental works are conducted using a three-phase voltage inverter fed induction motor.