Wiener Method Fault Detection in Rotating Machines: A Comparative Analysis

Accurate fault detection in rotating machines ensures optimal performance and prevents costly downtime in industrial applications. This research utilizes the power of machine learning to assess the identification of faults, in rotary machines. A specialized fault setup collected the vibration data that ranged from healthy to faulty conditions. The data include a total of 96 recorded data, of which 24 data are for the healthy condition and 72 data for the faulty condition. After collecting data, our analysis with machine learning involved using features and classification techniques such as Support Vector Machines (SVM) k Nearest Neighbors (KNN) and Multi-Layer Perceptron (MLP). Furthermore, we employed correlation analysis (CA) and the Wiener method to enhance fault detection accuracy. The results show that the best performance is achieved using features extracted by the Wiener method which improved the accuracy of test data for all classifiers, reaching up to 96.42% for SVM, and 93.33% for both MLP and KNN models. These findings highlight the potential of machine learning, in improving fault detection and categorization in rotary machines, particularly through the effectiveness of Wiener methods.