Data-driven modeling and control of perovskite deposition for solar cells

Precise control of thin-film deposition is essential in perovskite solar cell manufacturing. This work presents data-driven strategies based on Model Predictive Control (MPC) and Predictive Reference Governor (PRG) to regulate temperature and deposition rate during the perovskite thin-film deposition process while explicitly handling operational constraints. First, a grey-box model of the process is identified from experimental data and integrated into an MPC controller. In parallel, a PRG strategy enhanced with a Kalman filter is proposed to retain the built-in Proportional–Integral–Derivative (PID) controller while enforcing operational constraints and ensuring accurate setpoint tracking. Simulation results show that the proposed MPC and PRG approaches improve tracking performance compared with conventional PID control, thanks to their predictive capability. Finally, a hardware-in-the-loop implementation of the PRG in a Raspberry Pi confirms suitability for embedded deployment.