Prognostics and health management of induction motor by supervised learning classifiers

Current occasions are tied in with being modern and bringing down the expense of working the framework. Albeit the underlying interest in the framework can’t be changed, which is fixed, the methodology is to bring down the expense of activity of the machine with the goal that the support cost is limited since the chance of disappointment will be lower. The machine might be in a distant area where physically checking the status or performing upkeep won’t be effective. Accordingly, to improve execution and increment plant profitability, predictive support can be planned in such a way that numerous machines can be overhauled simultaneously. What’s more, this will expand the general uptime of the plant all in all. For most machines, condition monitoring is a custom-fitted way to deal with predictive support. Also, late patterns show that AI is the quickest and most proficient approach to accomplish the ideal answer for this issue. In this work, predictive maintenance is done for two major occurring faults broker rotor bar and bearing faults. Time-domain analysis has been used to develop hundred feature vectors, which are utilized for training, testing, and validation. The best suitable classifier model to predict faults is developed after comparison of Artificial Neural Network, Ensemble, and Decision Tree smart classifiers based on model accuracy. Out of these, the maximum accuracy is obtained by Artificial Neural Network.