Centroid fault detection method for three-phase inverter-fed induction motors
Description
Three phase electrical machines are critical with respect to the power and manufacturing industry. Being able to keep these machines and the systems running at an optimal efficiency with as few of maintenance outages as possible has become one of the highest priority. In this dissertation a new, simpler, and non-invasive fault detection and data reduction method is developed. Using a centroid determination method we are able to show the location and type of the fault in a three-phase system, specifically in the electrical machine and the inverter that feeds it. This research reports the literature on the background needed to develop the new fault detection method. The computer simulations and the experiments to prove the effectiveness of the fault detection method is discussed in detail. The detection method makes use of the Concordia Transform to be able to view the AC quantities (currents and/or voltages) of the system in a two-dimensional frame. The proposed research efforts are focused on the fault monitoring methods rather than system control and its monitoring. The central aspect to the development of the fault monitoring is the use of pattern symmetry across the positive and negative Beta axis after the three-phase quantities (currents) are transformed to the two-dimensional AC quantities (currents). The currents are sensed using Hall Effect sensors, which are well known for accuracy and fast response time. The detection system looks at the system on a cycle-by-cycle single point summary of the current spectrum. The ability to simplify the computational methods for fault detection leads to development of a real-time condition monitoring techniques which is low cost, simpler, and free from excessive hardware needed in traditional diagnosis methods. This dissertation reports experimental verification of simulated results covering samples of the results on an inverter-fed induction motor. The laboratory prototype of test system employs DSpace DSP system to implement various MATLAB based real-time condition monitoring and fault detection algorithms