Abstract:Fault monitoring and diagnostics are important to ensure reliability of electric motors. Efficient algorithms for fault detection improve reliability, yet development of cost-effective and reliable classifiers for diagnostics of equipment is challenging, in particular due to unavailability of well-balanced datasets, with signals from properly functioning equipment and those from faulty equipment. Thus, we propose to use a Bayesian neural network to detect and classify faults in electric motors, given its efficacy with imbalanced training data. The performance of the proposed network is demonstrated on real life signals, and a robustness analysis of the proposed solution is provided.

What problem does this paper attempt to address?

The main problems that this paper attempts to solve are the challenges in motor fault detection, especially how to effectively detect and classify motor faults in the case of unbalanced datasets. Specifically: 1. **Unbalanced Dataset Problem**: Collecting a balanced dataset containing signals from normally operating devices and faulty devices is a significant challenge. Since fault data is difficult to obtain in actual operations, the training data is unbalanced, which affects the performance of traditional classifiers. 2. **Efficient and Reliable Fault Detection Algorithm**: Developing a cost - effective and reliable fault - detection classifier remains an unsolved problem. Existing methods are not effective in dealing with unbalanced data, and a more robust method is required to improve the accuracy of fault detection. To solve these problems, the author proposes a method based on Bayesian Neural Networks (BNNs), taking advantage of its superiority in handling unbalanced data. The specific contributions are as follows: - **Frequency - Domain Signal Representation**: A frequency - domain - based acoustic signal representation method for damage detection is proposed. - **BNN Structure Design and Analysis**: A Bayesian neural network structure for damage detection is designed and analyzed. Through these methods, the author aims to improve the accuracy and reliability of motor fault detection, especially in the case of unbalanced data. The experimental results show that the proposed BNN method achieves 100% accuracy in detecting faulty devices and approximately 70% accuracy in identifying normal devices. ### Formula Display The formulas involved in the paper include: 1. **Loss Function of Traditional Neural Networks**: \[ L(D, W)=\sum_{i}(y_{i}-f(x_{i}, w))^{2}+\lambda\sum_{d}w_{d}^{2} \] 2. **Log - Likelihood Function of Bayesian Neural Networks**: \[ \log p(D, w)=\sum_{i}\log N(y_{i}|f(x_{i}, w), I)+\sum_{d}\log N(w_{d}|0,\frac{1}{\lambda}) \] 3. **Posterior Distribution Calculation**: \[ p(w|D)=\frac{p(D|w)p(w)}{\int p(D|w')p(w')dw'} \] 4. **Posterior Predictive Distribution**: \[ p(\hat{y}(x)|D)=\int p(\hat{y}(x)|w)p(w|D)dw = \mathbb{E}_{p(w|D)}[p(\hat{y}(x)|w)] \] These formulas show how Bayesian neural networks handle weights through probability distributions to better deal with uncertainty problems.

Detection of Electric Motor Damage Through Analysis of Sound Signals Using Bayesian Neural Networks

Induction Motor Failure Identification Based on Multiscale Acoustic Entropy Feature Selection and Hierarchical Adaptive Neuro-Fuzzy Inference System with Localized Recurrent Input

Patterns Based on Clarke and Park Transforms of Wavelet Coefficients for Classification of Electrical Machine Faults

Enhancing the Efficiency of Failure Recognition in Induction Machines through the Application of Deep Neural Networks

Neural Classification of Rotor Faults in Three-Phase Induction Motors using Electric Current Signals in the Frequency Domain

Analysis of Inter-Turn Short-Circuit Faults in Brushless DC Motors Based on Magnetic Leakage Flux and Back Propagation Neural Network

Multiple Discriminant Analysis and Neural-Network-Based Monolith and Partition Fault-Detection Schemes for Broken Rotor Bar in Induction Motors

Detection and Classification of Rolling Bearing Defects Using Direct Signal Processing with Deep Convolutional Neural Network

A new metric for reliable diagnosis of rotating machines applied to a multi-fault rotor using Bayesian neural networks

Detection of common motor bearing faults using frequency-domain vibration signals and a neural network based approach

Deep learning-based mechanical fault detection and diagnosis of electric motors using directional characteristics of acoustic signals

Diagnosis of Stator Winding and Permanent Magnet Faults of PMSM Drive Using Shallow Neural Networks

Cutting Through the Noise: An Empirical Comparison of Psychoacoustic and Envelope-based Features for Machinery Fault Detection

Sound Visualization and Convolutional Neural Network in Fault Diagnosis of Electric Motorbike

Convolutional Neural Network and Motor Current Signature Analysis during the Transient State for Detection of Broken Rotor Bars in Induction Motors

Convolutional Neural Networks for Automated Rolling Bearing Diagnostics in Induction Motors Based on Electromagnetic Signals

Deep learning based mechanical fault detection and diagnosis of electric motors using directional characteristics of acoustic signals

Statistical Neural Networks for Induction Machine Fault Diagnosis and Features Processing Based on Principal Component Analysis

Bearing fault diagnosis using weakly supervised long short-term memory

Enhancing Machine Learning Multi-Class Fault Detection in Electric Motors through-Entropy-Based Analysis

Monitoring and Diagnosing Faults in Induction Motors' Three-Phase Systems Using NARX Neural Network