Sensorless Vector Control of Three Phase Induction Motor Based on Full Neural Estimator and Controller

Abstract

Conventional vector control of A.C drives are widely used in industry andmany other applications, where high dynamic performance is required, this type ofcontroller usually needs costly speed sensor. Sensorless vector control is stronglyrecommended in the dangerous sites and hostile environment, also to reduce costand increase reliability. In this way the rotor speed can be estimated from theterminal voltage and current by means of DSP microprocessor. The DSP-estimatoris very complex hardware, has many operation problems, and very sensitive to theelectromagnetic interference (EMI). This paper proposes using the ArtificialNeural Network (ANN) to estimate the rotor speed, flux vector, torque, and unitvectors instead of DSP-estimator. Also, the neural-based controller is proposedtoo. The ordinary vector control with speed sensor and sensorless vector controlbased on DSP-estimator PI-controller are represented in this work as point ofcomparison. Also, the mathematical representation and simulation of the threephase induction motor is illustrated in this paper. The proposed method, neuralbasedsensorless vector controller and estimator, gives superior performance indifferent speed with respect to DSP-estimator PI-controller.