ANN-Based Observer for Controlling a SynRM

Abstract

The rotor position for electric motors, especially synchronous motor type, is an important information in speed / motion control applications. The position information of the rotating machines can be measured via incremental/absolute encoder, resolver or hall-sensors. However, these measuring devices have many disadvantages such as increased hardware complexity, high cost, increased volume, cable addition, decreased noise immunity, decreased reliability and increased maintenance. All these drawbacks can be overcome by using a sensorless algorithm. In this study, a Synchronous Reluctance Motor (SynRM) is controlled by using sensorless algorithm based on Artificial Neural Network (ANN). The algorithm presents not only position/speed estimation but also torque and flux observation which are critical to obtain maximum torque per ampere from SynRM. In this study, the motor is powered via solar photovoltaic panels. The environmental conditions such as irradiation and temperature affect the motor control performance. Therefore, the estimation performance of the ANN-based observation is analyzed under various environmental conditions. The results show that the sensorless system has satisfactory performance in all conditions.