Learning Machine Based on Optimized Dimensionality Reduction Algorithm for Fault Diagnosis of Rotor Broken Bars in Induction Machine

Induction machine health monitoring is considered a developing technology for the online detection of faults that occur even at initial stage. The objective this study to present an artificial intelligence (AI) technique and localization adjacent distant broken bar in induction machine, through multi-winding model simulation these cases. In work, it was found application Artificial Neural Networks (ANN) based on Mean Squared Error (MSE) Random Forest (decision tree) plays important role detecting locating defaults. stator current signal Ias motor dynamic state acquired from healthy faulty with rotor fault. 9 statistical features 8 wavelet packet parameters are extracted signal. These were employed as input vector train test ANN random fores29t determine whether running under normal conditions or defective. For optimizing defect classification procedure, feature selection algorithms adopted, such BBAT BPSO. reduction, we used principal component analysis (PCA) algorithm, reduce number features. results showed forest classifier followed by PCA can detect defective high accuracy (98.3333%) compared other methods.