基于高增益微分器的优化神经自适应三阶滑模控制提升光伏系统性能：仿真与实验验证

his paper addresses the challenge of optimizing photovoltaic (PV) system performance under rapidly changing environmental conditions, such as fluctuations in solar irradiance and temperature. To enhance energy extraction and regulate power delivery, a non-inverting DC-DC buck-boost converter is employed. Motivated by the nonlinear behavior and uncertainty inherent in PV systems, we propose a novel control strategy—Neuro-Adaptive Third-Order Sliding Mode Control (NATOSMC)—that ensures accurate and robust reference tracking. The control scheme integrates a Radial Basis Function Neural Network (RBFNN) for generating the reference signal, and a High-Gain Differentiator (HGD) to estimate system states using differential flatness theory. To ensure optimal performance, adaptive laws adjust the control gains in real-time, with parameter optimization performed via the Grey Wolf Optimizer (GWO). Stability is rigorously guaranteed through Lyapunov analysis. Both simulation and experimental results validate the proposed method, demonstrating superior performance in terms of tracking precision, dynamic response, and robustness compared to conventional controllers. Key results include a settling time of 0.014 s, rise time of 0.012 s, overshoot of 0.05%, and tracking efficiency of 98.71%. The controller also achieves improved performance indices: Integral Absolute Error (IAE) of 111.04, Integral Square Error (ISE) ofhis paper addresses the challenge of optimizing photovoltaic (PV) system performance under rapidly changing environmental conditions, such as fluctuations in solar irradiance and temperature. To enhance energy extraction and regulate power delivery, a non-inverting DC-DC buck-boost converter is employed. Motivated by the nonlinear behavior and uncertainty inherent in PV systems, we propose a novel control strategy—Neuro-Adaptive Third-Order Sliding Mode Control (NATOSMC)—that ensures accurate and robust reference tracking. The control scheme integrates a Radial Basis Function Neural Network (RBFNN) for generating the reference signal, and a High-Gain Differentiator (HGD) to estimate system states using differential flatness theory. To ensure optimal performance, adaptive laws adjust the control gains in real-time, with parameter optimization performed via the Grey Wolf Optimizer (GWO). Stability is rigorously guaranteed through Lyapunov analysis. Both simulation and experimental results validate the proposed method, demonstrating superior performance in terms of tracking precision, dynamic response, and robustness compared to conventional controllers. Key results include a settling time of 0.014 s, rise time of 0.012 s, overshoot of 0.05%, and tracking efficiency of 98.71%. The controller also achieves improved performance indices: Integral Absolute Error (IAE) of 111.04, Integral Square Error (ISE) ofT 11.42×10>5, Integral Time Absolute Error (ITAE) of 3.19, and Integral Time Square Error (ITSE) of 1.21×104. These results highlight the efficacy of NATOSMC in enhancing the control and reliability of PV energy systems.