A novel approach of an adaptive neuro-PI vector controller fed induction-motor servo drives
Essamudin Ali Ebrahim · 2003
This paper presents a novel approach for a very simple architecture of an induction-motor (IM) servo drive using only one single neuron (SN) as an online self-tuning artificial neural network (ANN). The action of this controller is similar to an adaptive PI-controller. The adaptation of the proposed controller is achieved by self-tuning for both weight and bias of the SN. Also, using an adaptive learning rate insures adaptation to overcome uncertainty and nonlinearity of the plant. Based on an indirect field-oriented vector control algorithm, the ANN speed tracking controller is developed and integrated with the adaptive hysteresis current-controlled pulse-width modulation (PWM) inverter to offer a high performance IM-drive. The complete drive system is implemented in a real time using a digital signal processor (DSP) controller board DS1102 on a laboratory 1-hp IM. Using the experimental rig, the performance of the proposed drive is evaluated under various speed trajectories. The test results validate the efficacy of the proposed simple controller for precise speed and position tracking of IM drive. Furthermore, the use of SN makes the drive system robust, accurate, and insensitive to parameter variations. Also, the controller contributes towards time consumption reduction through the real-time DSP control algorithm.