模型 of the main circuit of APF in d-q rotating coordinate system, the inverse system method is adopted to decouple it into a pseudo-linear system. Then, a controller is designed based on Linear Quadratic Regulator (LQR) to satisfy specific performance indices. On this foundation, the energy conversion in parallel APF is discussed, and a direct current control method is proposed based on power balance theory. This approach bypasses the harmonic current detection stage by calculating the reference grid current via a PI regulator using the difference between the capacitor voltage on the current side and its reference value.
The thesis further delves into the design of a parallel active power filter controlled by TMS320F2812 DSP, detailing both the hardware and software configurations. Experimental studies are initially conducted on the parallel active power filter, validating the effectiveness of the proposed methodologies.
In conclusion, the research presented in "人工智能-深度学习-有源电力滤波器控制系统研究.pdf" primarily focuses on enhancing the control systems of Active Power Filters (APF) for power quality improvement. The author investigates various aspects of APF technology, including harmonic current detection methods, current tracking control strategies, and control techniques leveraging inverse system theory and LQR. The proposed direct current control based on power balance theory simplifies the system architecture while maintaining effective compensation. The work also showcases the practical implementation using a DSP controller, demonstrating its potential in real-world applications.
This study contributes to the field of power electronics by offering innovative control solutions for active power filters, addressing the challenges posed by nonlinearity in power systems and harmonics generated by electronic devices. By integrating principles from artificial intelligence and deep learning, future advancements could potentially lead to even more efficient and intelligent APF systems capable of adapting to dynamic power grid conditions.
