Design and analysis of a DC motor speed drive with generalized regression neural network (GRNN) and invasive weed optimization (IWO) algorithms

This advanced research focuses on designing and analyzing a DC motor speed drive with Generalized Regression Neural Network (GRNN) and Invasive Weed Optimization (IWO) Algorithms using MATLAB/SIMULINK as a simulation aid. The DC motor speed drive is designed for fast dynamic speed and current response in all four quadrants of the motor’s torque-speed plane. The DC motor’s mathematical model is used for characterizing the system; IWO controllers are designed and tuned with methods including (MATLAB tuning, particle swarm optimization (PSO), and Internal Model Control). Two control strategies, single loop IWO and cascaded PI loops, were studied. The cascaded DC motor speed control was used for developing the DC motor’s speed drive, which was tuned for a current loop bandwidth of 2π.600 rads/s, and a current limiting logic was implanted in the current loop of the controller. The PMDC machine’s speed was controlled using the voltage control method with a Full bridge DC-DC power converter. Metal–oxide–semiconductor field-effect transistor (MOSFET) was used as the switch. The switching was done using the Unipolar Pulse Width Modulation technique due to its positive effect on the motor’s current ripples. The use of active damping and active resistance in the speed and current loop was done to improve the drive performance. After tuning the controllers, the IWO-tuned single-loop GRNN controller had a better response than the MATLAB-tuned single-loop GRNN controller. GRNN-IWO gave a well-damped response with minimal overshoot compared to the MATLAB tuned IWO with an accuracy of 98.85%. Likewise, the cascade PI controller gains were obtained, and the controller yielded a well-damped response with negligible overshoot. The drive current limiting mechanism also ensured the rated continuous current of the motor was not exceeded during continuous operation. The cascade GRNN-IWO controller is shown to have excellent load disturbance rejection capacity with zero steady-state error. With the full bridge converter, the motor operated in both forward and reversed directions.