Fault diagnosis based on Motor Current Signature Analysis for a predictive maintenance

Predictive maintenance is becoming an important tool for companies, increasingly used to prevent equipment failures and avoid unplanned downtimes that reduce productivity and efficiency. This strategy relies on condition monitoring techniques to assess the component's state, enabling the early detection of potential failure by analyzing the collected signals such as vibration or current. By combining real-time data with historical data, it is possible to predict the remaining useful life of the equipment, allowing to plan cost-effective scheduled maintenance activities. The thesis focuses on fault detection techniques, particularly those based on Motor Current Signature Analysis (MCSA), which is a non-invasive method that relies only on the motor's current and does not require any additional expensive sensors to gather information on the equipment's health. The major defects that can occur in a Permanent Magnet Synchronous Motor, which can be detected with the MCSA, such as the eccentricity faults, the demagnetization and the mechanical unbalance faults, are investigated and explained. The latter issue is of particular interest in motor control applications for HVAC (Heating, Ventilation, and Air Conditioning) systems, where problems related to unbalanced conditions can arise in rotating components. In fact, over time, fan blades may deteriorate or can be affected by the accumulation of material, such as dust or ice, leading to a radial imbalance in mass distribution. The goal is to detect when this situation occurs and this has been achieved by modeling and simulating the system in the MATLAB/Simulink environment, followed by the analysis of the data collected from the experiments to validate the theory.