Abstract:In recent years, artificial intelligence technology has been widely used in fault prediction and health management (PHM). The machine learning algorithm is widely used in the condition monitoring of rotating machines, and normal and fault data can be obtained through the data acquisition and monitoring system. After analyzing the data and establishing a model, the system can automatically learn the features from the input data to predict the failure of the maintenance and diagnosis equipment, which is important for motor maintenance. This research proposes a medium Gaussian support vector machine (SVM) method for the application of machine learning and constructs a feature space by extracting the characteristics of the vibration signal collected on the spot based on experience. Different methods were used to cluster and classify features to classify motor health. The influence of different Gaussian kernel functions, such as fine, medium, and coarse, on the performance of the SVM algorithm was analyzed. The experimental data verify the performance of various models through the data set released by the Case Western Reserve University Motor Bearing Data Center. As the motor often has noise interference in the actual application environment, a simulated Gaussian white noise was added to the original vibration data in order to verify the performance of the research method in a noisy environment. The results summarize the classification results of related motor data sets derived recently from the use of motor fault detection and diagnosis using different machine learning algorithms. The results show that the medium Gaussian SVM method improves the reliability and accuracy of motor bearing fault estimation, detection, and identification under variable crack-size and load conditions. This paper also provides a detailed discussion of the predictive analytical capabilities of machine learning algorithms, which can be used as a reference for the future motor predictive maintenance analysis of electric vehicles.

Majority Voting Machine Learning Approach for Fault Diagnosis of Mechanical Components

Fault Diagnosis Method Based on Improved Deep Boltzmann Machines

Digital twin-driven machine learning: ball bearings fault severity classification

Machine Learning-Based Fault Diagnosis for Rotating Machinery in Industrial Settings

Development of Intelligent Fault Diagnosis Technique of Rotary Machine Element Bearing: A Machine Learning Approach

Machine Learning-based Predictive Maintenance for Fault Detection in Rotating Machinery: A Case Study

A Closer Look at Bearing Fault Classification Approaches

A multi fault classification in a rotor-bearing system using machine learning approach

Machine Learning Based Mechanical Fault Diagnosis and Detection Methods: A Systematic Review

Application of Machine Learning to a Medium Gaussian Support Vector Machine in the Diagnosis of Motor Bearing Faults

Fault Detection of Bearing using Signal Processing Technique and Machine Learning Approach

Application of PCA and SVM in Fault Detection and Diagnosis of Bearings with Varying Speed

A Model for Prediction of Outer Race Defects of Rolling Contact Bearing based on Vibration Data Using Machine Learning Algorithms

Identification and Fault Diagnosis of Rolling Element Bearings Using Dimension Theory and Machine Learning Techniques

Recent advances in the application of deep learning for fault diagnosis of rotating machinery using vibration signals

Fault Classification in Vehicle Power Transmission using Machine Learning

Fault Diagnosis for Rotating Machinery Using Vibration Measurement Deep Statistical Feature Learning

Performance evaluation of deep learning approaches for fault diagnosis of rotational mechanical systems using vibration, sound, and acoustic emission signals

Fault Diagnosis of Machines Using Deep Convolutional Beta-Variational Autoencoder

Fault Detection of Bearing: An Unsupervised Machine Learning Approach Exploiting Feature Extraction and Dimensionality Reduction

Accurate Detection of Bearing Faults Using Difference Visibility Graph and Bi-directional Long Short-term Memory Network Classifier