External Fault Classification Experienced by Three-Phase Induction Motor Based on Multi-Class ELM

Abstract

Fault diagnosis and condition assessment (FDCA) of rotating machines becomes important due to the age of machine in service. Proper FDCA enhance the machine's operational life, efficiency and reducing catastrophic failure. This paper describes a realistic FDCA method for three phase induction motors (IMs) using readily available data. External faults experienced by IM are monitored by Multi-class Extreme Learning Machine (ELM) and compared its performance with multilayer perceptron (MLP) neural network which revealed that ELM algorithm is quite faster in investigations leading to reduction in computational load. RMS value of 3-phase voltages and currents are utilized as input variable in ELM model to identify six types of external faults experienced by IM and normal operating (NF) condition. Testing analysis of 160 instances has been performed to represent the robustness of the investigated seven status conditions of IM for wide changes in operating and loading condition perturbation.