Classification of Inter-Turn Short-Circuit Fault Severity in Permanent Magnet Synchronous Motors Using Decision Tree and Bayesian Neural Network
Subject Areas : مهندسی برق و کامپیوترAbbas Darvishi 1 , Seyed Mohsen Seyed Moosavi 2 , Behzad Moshiri 3
1 - Faculty of Elec. Eng., Ahvaz Branch, Islamic Azad University, Ahvaz, Iran
2 - Faculty of Elec. Eng., Ahvaz Branch, Islamic Azad University, Ahvaz, Iran
3 - School of Elec. and Comp. Eng., University of Tehran, Tehran, Iran
Keywords: Inter-turn short-circuit, permanent magnet synchronous motor, feature extraction, feature selection, decision tree, Bayesian neural network.,
Abstract :
This paper investigates the identification of inter-turn short-circuit fault severity in a 3-kilowatt permanent magnet synchronous motor using a decision tree and a deep Bayesian neural network. The primary dataset includes three-phase current signals under both healthy and faulty conditions, covering six fault severity levels. A preprocessing stage is conducted to analyze the data in time and frequency domains using discrete wavelet transform and power spectral density analysis. To reduce the dimensionality of the feature space, statistical indicators such as mean, standard deviation, kurtosis, and skewness are initially extracted. Kernel principal component analysis is then employed to identify the most salient features. A decision tree algorithm is trained to detect motor fault conditions. Finally, a deep Bayesian neural network is applied to classify the severity of the inter-turn short-circuit fault. The proposed algorithm’s performance is evaluated in terms of accuracy, precision, recall, and F1-score, considering varying numbers of selected dominant features.