MPC-Based Process Control of Deep Drawing: An Industry 4.0 Case Study in Automotive

Deep drawing is a metalworking procedure aimed at getting cold metal sheet plastically deformed in accordance with pre-defined mould. Although this well-established industry, it still susceptible to several issues affecting the quality of stamped products. In order reduce defects workpieces, process control approaches can be performed. Typically, employs simple proportional-integral-derivative (PID) regulators that steer blank holder force (BHF) based on error punch force. However, single PID only single-input single-output systems and cannot handle constraints variables. Differently from state art, paper we propose architecture Model Predictive Control (MPC), which considers multi-variable system model. particular, represent deep multiple-output Hammerstein-Wiener model relates BHF draw-in $n$ different critical points around die. This allows avoidance workpiece are due abnormal sliding during forming phase. The effectiveness proposed controller shown real case study digital twin framework, where performance achieved by MPC-based analyzed detail compared against results obtained through an ad-hoc defined multiple PID-based architecture. Note Practitioners —This work motivated emerging need for effective implementation zero-defect manufacturing paradigm Industry 4.0 framework. Especially process, various parts lead significant product waste inefficiencies. turns into considerable economic losses companies, particularly automotive sector, one most used techniques. applications, sample inspections performed batches finished-product, subsequent time resources. For sake improving quality, innovative strategies real-time viable promising solution. context, approach correct shaping stroke, thanks monitoring locations indeed variables provide stroke. A useful easy-to-implement non-linear provided perform strategy. comprehensive methodology applied study, ranging modeling (model identification validation experimental data acquisition) MPC (controller tuning testing software-in-the-loop validation). presented method easily implemented any press, providing multivariable constrained suitable able make well formed.