Zernike radial basis neural network control of DC–DC power converter driven permanent magnet DC motor: design and experimental validation

This article presents a novel control architecture for an enhanced closed-loop speed tracking of a DC–DC buck power converter fed Permanent Magnet DC motor (PMDC) motor in face of large exogenous load torque uncertainty. The proposed architecture combines a new self learning Zernike radial polynomial neural network (ZRNN) estimator with the backstepping controller. The design involves a computationally simple online learning based ZRNN to rapidly and accurately estimate the unknown large load torque uncertainties. The proposed control solution concurrently guarantees stability and excellent dynamic performance through an effective neural network based estimation and subsequent compensation of unanticipated load torque perturbations over a wide range. The closed loop stability of the DC–DC buck power converter driven PMDC motor and asymptotic speed tracking with the proposed neuro-adaptive controller is proved using the stability theory for non-autonomous systems. The effectiveness of the proposed controller has been investigated through experimentation on an indigenously developed laboratory prototype of 200 W under closed loop operation using digital signal processors. The tests conducted around different operating conditions include the motor start-up response, step variations in the load torque, and step changes in the reference speed. Experimental results demonstrate a significant improvement in the speed tracking performance achieving 48.13% reduction in the settling time and no-change in speed during start-up and load torque perturbations upto 600%, respectively. Experimental validations and extensive tests spanning over a large operating region, substantiate the theoretical claims and real-time suitability of the proposed controller for sensitive applications demanding high performance.