Detection and diagnosis of bearing faults using Artificial Intelligence techniques on vibration signals
fault detection and diagnosis, artificial intelligence techniques, bearing fault diagnosis, machine learning.
With the development, increase in the complexity and costs of industrial systems, management measures that aim to prevent or mitigate the loss of reliability, decrease in productivity and safety risks, caused by process abnormalities and component faults, become increasingly more important. In this context, Artificial Intelligence (AI) has been consolidating itself as an effective and challenging means in the process of monitoring, detecting and diagnosing faults in industrial equipment and systems. Among the equipment, which are frequently the object of studies, bearings stand out, which are critical mechanical components of rotating machines. Vibration monitoring is the most widely used technique for detecting, locating and distinguishing bearing faults. Motivated by the efficient and growing performance of AI techniques and the importance of bearings in industrial processes, this work proposes the development of an intelligent method for detecting and diagnosing faults in bearings, applying AI techniques. For the development of the proposed method, it will be using the Case Western Reserve University (CWRU) bearing test database. The raw vibration signals will go through signal processing approaches such as statistical analysis, Fourier Transform and Wavelet Transform to extract the representative characteristics of the noisy and non-stationary signal. To improve the performance of the diagnostic system, methods of dimension reduction to the resources obtained in the extraction stage will be used, such as Principal Component Analysis (PCA) and Sequential Resource Selection (SRS). In the last step, the features extracted from the vibration signal will be used to categorize the bearing conditions, using machine learning algorithms such as Artificial Neural Network (ANN), support vector machine (SVM), k-nearest neighbor (k -NN) and/or Deep Neural Network (DNN), to classify bearing faults.