Energy Efficiency Estimation of Induction Motors with Artificial Neural Networks

Abstract

Induction motors make up 90% of today's motors in the industry. For this reason, the contribution of energy efficiency analyses to induction motors is very important. There are many techniques for measuring the efficiency of electric motors. These are the generally experimental ones as specified in certain standards. Experimental methods can also be divided into direct (IEEE 112-B, CSA-390) or indirect (IEC 34-2, JEC 37) methods. The use of experimental methods is not common due to the cost of installing and operating test laboratories worldwide. Therefore, energy efficiency estimation methods are used in worldwide. In this study, efficiency estimations are made with artificial neural network (ANN), which is an optimization-based estimation method with using data of 307 induction motors' (from small to large) from three different companies (AEG-TECO-GAMAK). The results are very close to the efficiency values given in catalog values. However, another noteworthy issue is that the estimation errors of the efficiency change from company to company. The errors of one company are higher than the others.