Neural Network-Based Diagnostic Tool for Detecting Stator Inter-Turn Faults in Line Start Permanent Magnet Synchronous Motors

Abstract

Three-phase line-start permanent magnet synchronous motors are considered among the most promising types of motors in industrial applications. However, these motors experience several faults, which may cause significant financial losses. This paper proposed a feed-forward neural network-based diagnostic tool for accurate and fast detection of the location and severity of stator inter-turn faults. The input to the neural network is a group of representative statistical and frequency-based features extracted from the steady-state three-phase stator current signals. The current signals with different numbers of shorted turns and loading conditions are captured using the developed finite element JMAG™ model for interior mount LSPMSM. In addition, an experimental set-up was built to validate the finite element model and the proposed diagnostics tool. The simulation and experimental test results showed an overall accuracy of 93.125% in detecting the location and the size of inter-turn, whereas, the accuracy in detecting the location of the fault is 100%.

Keywords

• Electric motors
• fault currents
• fault detection
• finite element analysis