Classification of the Supply Voltage conditions of a Three-Phase Induction Motor with Machine Learning techniques.
Keywords: Fault Diagnostic, Machine Learning, Pattern Recognition, Three-Phase Induction Motor, Unbalanced Supply Voltages
AbstractThree-Phase Induction Motors (TPIM) is a fundamental part, as they are the main responsible for carrying out the mechanical work process in the industry. It is estimated that they are responsible for consuming more than half of all energy destined for the industrial sector. Thus, any failure of operation in motors of this type is reflected in energy, economic and environmental losses. Among the most common failures is the unbalance of the supply voltages, which can cause total loss of the machine depending on the magnitude of the unbalance. This article addresses a comparative analysis between the Machine Learning K-Nearest Neighbors (KNN), Random Forest (RF), Suport Vector Machine (SVM), Principal Component Analysis (PCA) and Multilayer Perceptron Neural Network (MLP) techniques applied to the classification of unbalanced supply voltages of a three-phase induction motor. For this, a database was used with mechanical and electrical variables related to the balanced and unbalanced operation of the motor, divided into classes of different levels of unbalance according to the National Electrical Manufactores Association (NEMA).