Heuristic based in situ efficiency estimation for induction motors

This thesis presents a method based on heuristic optimization algorithms to estimate the efficiency of three-phase induction motors in-situ, non-invasively, and with high accuracy. The method relies on calculating motor efficiency through the estimation of equivalent circuit parameters and is implemented with reference to the IEEE 112-A test standard. The study comprises a two-stage application process involving algorithm-based efficiency estimations conducted on motors with different power ratings and experimental tests performed under voltage unbalance conditions. In the first stage equivalent circuit parameters of six three-phase induction motors with different power ratings were estimated and efficiency predictions were made accordingly. Within this scope measurement data were obtained based on the IEEE 112-A test standard and the efficiency values of the same motors were estimated using eight different heuristic optimization algorithms (PSO, GA, GWO, HHO, TSA, DSA, WOA, RFA). The accuracy of each algorithm's prediction was evaluated through comparative error analysis at four different loading levels (25%, 50%, 75% and 100%). In the second stage the effects of voltage unbalance on motor efficiency were examined. For this purpose, an experimental setup was established on a 1.1 kW three-phase induction motor, which was tested under six voltage unbalance levels (0%, 1%, 2%, 3%, 4%, 5%) and four different loading conditions (25%, 50%, 75%, and 100%). In this part only one heuristic optimization algorithm was used to estimate efficiency based on equivalent circuit parameters and the results were compared with efficiency values measured in accordance with the IEEE 112-A standard. Additionally, the model's prediction performance under voltage unbalance conditions was analyzed graphically. The results indicate that the proposed method performs with high accuracy especially under full-load and low voltage unbalance conditions while its prediction accuracy decreases relatively under low load levels and higher voltage unbalance. Nonetheless, the method stands out with its advantages such as field applicability, non-intrusive computation capability and cost-effectiveness. Overall, the developed approach is considered to have significant potential in both energy efficiency optimization and industrial motor monitoring systems.