An Online Neural Network Based Model-Free Control Method for High-Order DC–DC Converter

Among the various control strategies for complex dc–dc converters, those employing neural networks have risen to prominence. They excel in their ability to approximate functions without the need for precise mathematical models, which is especially advantageous for complex, nonlinear, and uncertain control systems. This article presents a novel, online, model-free control strategy for high-order dc–dc converters that harnesses the power of neural networks. By leveraging real-time operational data to train the neural network, this approach facilitates the development of a converter controller devoid of the requirement for an intricate model. The process gradient is extracted using an online estimation technique. This article delves into the principles underlying model-free systems and provides a detailed analysis of the stability of the proposed control method. To validate the practicality, robustness, and responsiveness of this control framework, an extensive series of experiments has been conducted on a high-step-up dc–dc converter, which is high-order and difficult to model. The selection of this converter poses a significant challenge in terms of achieving an accurate model, thereby serving as a rigorous test case for the efficacy of the proposed neural network-based control strategy.