Data-Driven Force Control of an Automated Scratch Test

The development of chemicals, like coatings, is increasingly automated in high-throughput plants. The assessment of the quality of the coating formulation is also automated to achieve precise and reproducible results. Usually, a cross-cut test is performed to characterize the adhesion of coatings. To obtain comparable conditions and comply with standards, the force during scratching has to be controlled. In this paper, the force control of an automated testing equipment is optimized. There are different substrates, coating formulations and process parameters influencing the quality of the adhesion of the coatings. Because of this, the force controller has to cover a variety of dynamics. The different coatings and nonlinear behavior of the scratching makes modeling the system extremely difficult. Therefore, a data-driven approach is used. Two data-driven methods, the intelligent PID (iPID) and Virtual Reference Feedback Tuning (VRFT), are applied to the field of force control. The results for both control methods are compared. Additionally, both methods are combined with an Iterative Learning Control (ILC) to optimize the control performance over multiple iterations. The qualities of the data-driven controllers are evaluated. All results are compared to the currently implemented controller.