Diagnosing faults in electric motors using B-splines and Bayesian Mixture Model

Diagnosing faults in electric motors is crucial for various applications, from everyday devices to industrial machinery. Authors propose a method for identifying motor faults using acoustic signals, which are easy to capture with microphones. Proposed approach involves analyzing these signals using Functional Data Analysis (FDA), representing frequency patterns with B-splines and Bayesian Mixture Model as classifier. In this paper, there was developed a classifier to categorize five motor fault types based on these transformed signals. By focusing on frequencies up to 2500 Hz relevant to motor issues, authors aim to detect faults without needing complex equipment and greatly shorten computation time. This approach yields promising results.