International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 12 Issue: 04 | Apr 2025
p-ISSN: 2395-0072
www.irjet.net
Fault Detection and Classification in Three-Phase Induction Motors Using Grey Wolf Optimized PRNN N. Sivaraj1, B. Rajagopal2 1Research Scholar, Annamalai university, Annamalai nagar, Tamil Nadu, India 2Associate Professor, Annamalai university, Annamalai nagar, Tamil Nadu, India
---------------------------------------------------------------------***--------------------------------------------------------------------contain noise and external disturbances, making it difficult to extract fault-related features effectively [6]. phase induction motors are essential for improving reliability and reducing maintenance costs. This research To address these challenges, this research proposes a presents a Grey Wolf Optimized Pattern Recognition Neural Hilbert Transform-based preprocessing technique to Network (GWO-PRNN) for diagnosing faults, with a focus on enhance signal quality and extract essential fault-related broken rotor faults, bearing faults and stator interturn features [7]. A 1.5 kW three-phase induction motor was fault. A 1.5 kW three-phase induction motor was tested tested under different load conditions to evaluate the under various load conditions to evaluate the proposed effectiveness of the proposed approach. Furthermore, a approach. The Hilbert Transform is applied as a Grey Wolf Optimized Pattern Recognition Neural Network preprocessing technique to extract meaningful features and (GWO-PRNN) is implemented for fault detection and remove noise from motor current and speed signals. The classification. The Grey Wolf Optimization (GWO) Grey Wolf Optimization (GWO) algorithm fine-tunes the algorithm is employed to fine-tune the neural network neural network parameters, enhancing classification parameters, enhancing classification accuracy and accuracy and convergence speed. The proposed method is convergence speed [8][9]. evaluated on multiple fault scenarios, including broken rotor faults, inter-turn stator faults, and bearing defects. This research aims to develop an intelligent and Experimental results show that GWO-PRNN achieves reliable fault diagnosis system capable of detecting and superior performance across different load conditions, with classifying induction motor faults with high precision. The an overall performance of 97.06%, considering accuracy, proposed approach contributes to predictive maintenance recall, precision, and F1-score. This approach contributes to strategies, improving the long-term operational stability of predictive maintenance strategies, ensuring improved induction motors. operational efficiency and extended motor lifespan.
Abstract - Fault detection and classification in three-
2. Literature Review
Key Words: (Fault Detection, Fault Classification, ThreePhase Induction Motor, Grey Wolf Optimization (GWO), Pattern Recognition Neural Network (PRNN), Hilbert Transform, Broken Rotor Fault, Predictive Maintenance, Motor Condition Monitoring.
Several studies have also investigated the impact of varying load conditions on the effectiveness of fault diagnosis techniques. Singh et al. [10] conducted a comparative study of induction motor fault diagnosis using vibration analysis and MCSA, concluding that currentbased techniques are more suitable for real-time applications. However, they emphasized that load variations introduce additional complexities, requiring adaptive preprocessing methods. Similarly, Wang et al. [11] explored the influence of load fluctuations on neural network-based fault classification, highlighting the importance of dynamic feature selection methods to maintain high accuracy across different operating conditions. Furthermore, hybrid techniques integrating multiple preprocessing and classification methods have shown promise in improving fault diagnosis accuracy. Riera-Guasp et al. [12] combined wavelet analysis with MCSA to enhance the detection of broken rotor faults, demonstrating improved performance compared to standalone MCSA. Li et al. [13] proposed a hybrid deep learning approach incorporating feature fusion techniques, achieving superior classification performance across multiple fault conditions. Although significant progress has
1.INTRODUCTION Three-phase induction motors are the backbone of modern industrial systems due to their efficiency, durability, and cost-effectiveness [1]. However, these motors are prone to various faults, including broken rotor faults, inter-turn stator faults, and bearing defects, which can lead to unexpected failures, increased maintenance costs, and reduced operational efficiency [2][3]. Early and accurate fault detection is crucial for preventing costly downtime and ensuring reliable motor performance. Traditional fault diagnosis techniques, such as vibration analysis and thermal imaging, require direct access to the motor, making them expensive and impractical for real-time monitoring [4]. In contrast, Motor Current Signature Analysis (MCSA) is a non-invasive and cost-effective approach that utilizes current signals for fault detection [5]. However, raw current signals often
© 2025, IRJET
|
Impact Factor value: 8.315
|
ISO 9001:2008 Certified Journal
|
Page 1825