Direct Preference Optimization-Based Adaptive Control for Minimizing Total Harmonic Distortion in Photovoltaic-Powered Electric Drives

In order to reduce Total Harmonic Distortion (THD) in electric drives that are driven by photovoltaic (PV) systems, this study suggests a unique Direct Preference Optimization-based Photovoltaic Voltage Control (DPO-PVC) framework. In order to simulate realistic operating circumstances, framework combines dynamic PV modelling, inverter management, and electric motor simulation with real-world solar irradiance data from NREL PVDAQ dataset. DPO approach, in contrast to conventional scalar loss-based techniques, adapts robustly to nonlinear, noisy, and time-varying environments by learning from preference rankings across controller configurations. efficiency of suggested controller is shown by extensive simulations conducted under various irradiance profiles, including clear, cloudy, and transient weather situations, as well as dynamic load scenarios. According to quantitative studies, DPO-PVC outperforms PID, fuzzy-PID, and genetic algorithm-optimized PID (GA-PID) controllers by up to 60%, achieving a voltage THD as low as 2.88%. Likewise, current THD drops to 2.57%, guaranteeing better motor current quality. Additionally, controller has a smaller speed regulation error (1.49%) and a faster settling time (180 ms), which enhances system stability and dynamic response. Achieving 94.6% energy efficiency highlights best use of solar electricity that is available. Further tests demonstrate DPO-PVC controller's practical usefulness by demonstrating that it retains performance robustness against irradiance changes, abrupt load surges, and sensor noise. efficacy of choice-based training is confirmed by learning convergence metrics, which show that preference accuracy exceeds 95% after 50 epochs. All things considered, DPO-PVC framework offers a noteworthy development in PV-powered electric drive control by successfully lowering harmonic distortion while boosting efficiency and stability. This method has a lot of promise for practical renewable energy systems that need high-fidelity, adaptive control in a variety of load and environmental scenarios.