In order to overcome the problems such as the complex interplay and dynamic evolution of multiple parameters (temperature, atmosphere, pressure, etc.) in the growth of graphene nanostructures that cannot be observed in real time, as well as the sluggishness of traditional empirical adjustment methods, the idea of optimizing the parameters during the growth process based on model predictive control is proposed. Taking chemical vapor deposition as an example, this paper establishes a computational control system with a framework of state variable representation, data modeling, rolling prediction, and closed-loop feedback, and incorporates temperature, CH₄ flow rate, H₂ flow rate, tank pressure, and surface growth state into the optimization model. The method is verified in the experimental device and simulation environment to achieve accurate tracking of the growth process, with the response lag time shortened to 0.74 s. The convergence time of the tracking reference signal is reduced to 2.31 s, ensuring a small steady-state error of 1.9% and good anti-disturbance recovery ability and stability. This study demonstrates that the online computer control strategy based on model predictive control can improve the parameter regulation efficiency and uniformity of the graphene nanostructure preparation process.
Povzetek: Predlagana je MPC-zasnovana metoda optimizacije rasti grafenskih nanostruktur.Z računalniškim modeliranjem stanj, drsnim napovedovanjem in zaprtozančnim uravnavanjem izboljša odzivnost, konvergenco in stabilnost procesa ter ohrani robustno delovanje ob motnjah dosledno v simulacijah in laboratorijskih poskusih.
