Data-Driven Control Algorithms And Simulator Development

Model-based control theory has a solid theoretical foundation and rigorous mathematical proof. It is the direction of the research in the mainstream in control field. However, with the development of industrial technology, the structure and properties of plants are becoming more complex; modeling is becoming increasingly difficult. These all force the scientists to search for a new control method. In this case, data-driven control theory emerged.Today, data-driven control theory is one of the main studies in automatic control field. In recent years, it is becoming more and more influent in the industrial process control. The emergence of data-driven control algorithm and the lack of a unified open platform for simulation and experiment will inevitably bring about a "merits" issue.The paper's main task is to design and develop a data-driven simulator. This simulator will provide a unified and open "competitive arena" for a variety of data-driven control algorithms; so that the comparison among them will become possible. It will be useful for further research of data-driven control theory to provide such powerful simulator.With the unified and open data-driven simulator developed on the basis of three-tank system, the paper carried out a series of simulation and experimental studies on three typical data-driven control algorithms - Model-Free Adaptive Control, Iterative Feedback Tuning and Virtual Reference Feedback Tuning—to analyze the merits of three data-driven control algorithms and validate the reliability of simulator.The main contents of this paper are as follows:First, the paper introduced the background and significance of data-driven control theory.Second, the paper introduced three typical data-driven control algorithms.Third, the paper introduced the development of a data-driven simulator.Fourth, the paper analyzed the merits of three data-driven control algorithms and validated the reliability of simulator with simulation and experiments.Finally, the paper gave the conclusion of the paper and its prospect.