A non-intrusive efficiency estimation method for in-service induction motors using neural networks

摘要

Numerous methods exist for determining the efficiency of induction motors. Many of them require a no-load test, which is not possible for in-situ determination. On the other hand, the evaluation of motor efficiency based on the motor's nameplate or manufacturer's data, in many cases cannot ensure a fair assessment of induction motors employed in the plant. This paper describes the results of a very low cost and accurate method for estimating the motor efficiency without the need for removing the toad from the motor, and without the need for measuring the output power or torque based on the neural networks. The results are compared with direct efficiency determination method by simulation and experimental results. Test results indicate that RMS value of the error by the novel method is less than 0.7% in efficiency estimation.