Improving photovoltaic water pumping system performance with ANN-based direct torque control using real-time simulation.

Photovoltaic Water Pumping Systems (PVWPS) have become increasingly important as a renewable energy solution in rural areas, providing energy independence, cost savings, and environmental friendliness. This system has two main controllers. The first controller is employed to maximize power extraction from the PV array by controlling the duty ratio of the DC-DC boost converter. The second controller is responsible for regulating the operation of the induction motor through the switching pulses of the Voltage Source Inverter (VSI). These two controllers play an essential role in the system, which increases efficiency and performance. Therefore, the innovative aspect of this work consists of introducing Artificial Neural Networks (ANNs) based on each PVWPS controller. On the one hand, ANN-based MPPT is implemented to ensure optimal performance of the PV array under varying irradiation levels. On the other hand, to overcome the defects and problems caused by Direct Torque Control (DTC), such as flux and torque ripples, high switching frequency, and challenges at low speeds, an ANN-based DTC is proposed in which each of the hysteresis comparators, switching table, and speed controller in the DTC are replaced by ANN controllers. The PVWPS based on the proposed controls is thoroughly modeled and simulated using MATLAB/Simulink software and validated using dSPACE DS1104 Board. The results demonstrate significant improvements, including a 75.51% reduction in flux ripples, a 77.5% reduction in torque ripples, a 44.79% improvement in response time, and an increase in the water quantity. Furthermore, the Real-Time simulation and visualization obtained are consistent with the simulation outcomes.