Proactive Identification-Based Self-Learning Control Method for Distributed Parameter System

Distributed parameter system (DPS) is a class of systems with distributed characteristics, whose states change with time and space. Achieving precise state control for DPS is of great significance. However, the traditional lumped parameter method is difficult to describe the distribution characteristics effectively. On the other hand, limited by the data quality of DPS, such as sampling density and data information, it is difficult to analyze its complex spatial–temporal characteristics, which further restricts its control effect. In view of the above challenges, this article proposes a proactive identification-based self-learning control method for DPS. Specifically, a parameter proactive identification method is innovatively proposed to realize the accurate identification of the dynamic mechanism of DPS. Then, a reduced-order predictive model (ROPM) is proposed to reduce the high computational complexity of the mechanism model. Taking into account the difficulty in guaranteeing the stability of DPS during the control process, a stable predictive control method is proposed, which takes ROPM as the predictive model and introduces the smooth reference trajectory constraint, ensuring smooth control of DPS. To verify the effectiveness of the proposed method, this article takes the first-order hyperbolic DPS case and the heat transfer process of volatilization kiln as the research object to carry out in-depth experiments. Compared with some state-of-the-art methods, the proposed method can greatly improve the model prediction accuracy and the control accuracy.