Transformation of a rolling mill aggregate to a Cyber Physical Production System: from sensor retrofitting to machine learning

Abstract

This paper describes the transformation of a rolling mill aggregate from a stand-alone solution to a fully integrated cyber physical production system. Within this process, already existing load cells were substituted and additional inductive and magnetic displacement sensors were applied. After calibration, those were fully integrated into a six-layer digitalization architecture at the Smart Forming Lab at the Chair of Metal Forming (Montanuniversitaet Leoben). Within this framework, two front end human machine interfaces were designed, where the first one serves as a condition monitoring system during the rolling process. The second user interface visualizes the result of a resilient machine learning algorithm, which was designed using Python and is not just able to predict and adapt the resulting rolling schedule of a defined metal sheet, but also to learn from additional rolling mill schedules carried out. This algorithm was created on the basis of a black box approach, using data from more than 1900 milling steps with varying roll gap height, sheet width and friction conditions. As a result, the developed program is able to interpolate and extrapolate between these parameters as well as different initial sheet thicknesses, serving as a digital twin for data-based recommendations on schedule changes between different rolling process steps. Furthermore, via the second user interface, it is possible to visualize the influence of this parameters on the result of the milling process. As the whole layer system runs on an internal server at the university, students and other interested parties are able to access the visualization and can therefore use the environment to deepen their knowledge within the characteristics and influence of the sheet metal rolling process as well as data science and especially fundamentals of machine learning. This algorithm also serves as a basis for further integration of materials science based data for the prediction of the influence of different materials on the rolling result. To do so, the rolled specimens were also analyzed regarding the influence of the plastic strain path on their mechanical properties, including anisotropy and materials’ strength.

Big Data in the Metal Processing Value Chain: A Systematic Digitalization Approach under Special Consideration of Standardization and SMEs

“…The MF houses the furnace, hydraulic press, and rolling mill, posing as the second production site in the process. The rolling mill from 1954 was transformed into a CPPS utilizing low-cost retrofitting and suitable sensors such as Linear Variable Differential Transformer (LVDT) and load cells to be able to integrate required data into the process [52]. In cooperation with the Chair of Industrial Logistics (IL), a cross-process database was set up to make data available between the cooperating parties.…”

“…Due to the existing amount and value of data, an initial BBM approach is applied within this example. The BBM can be fed and refined with further process data recorded during production and thus establish BBMLM, as shown in [52]. If required data cannot be obtained from the DAQ of the process and would require a WBM based simulation, e.g., FEA for material models or microstructure models, a GBMLM approach Digitization and digitalization of both enterprises were based on the model of the automation pyramid.…”

“…A further extension of these criteria is proposed by [52], adding two additional conditions to be met, especially considering SMEs: (iv) a user-centered CPPS consists of HMIs that are tailored to the application and the respective end-user; (v) a user-centered CPPS for SMEs is resilient and has a short amortization time.…”

“…However, the results may vary depending on the modeling method used and so the use of either a WB, GB or BB model has to be answered individually depending on the given circumstances and resources [56,58]. Taking Machine Learning (ML) into account, the models can be developed into White Box Machine Learning Models (WBMLMs), Grey Box Machine Learning Models (GBMLMs), and Black Box Machine Learning Models (BBMLMs), as demonstrated in [52,56]. In the case of SMEs, initial WBMs are especially important to consider, as there may not be enough experimental investigations nor resources to generate data from tests to meet the 5Vs of Big Data, leading to an unsatisfactory result.…”

“…In the further production process, data can be collected and fed into the initial WBM, which transforms it into a GBMLM, as shown in [56]. Large enterprises, on the other hand, may have already collected enough data to meet the 5Vs of Big Data and can implement a BBM that is also fed with data collected in the process, transforming it into a BBMLM [52]. Nevertheless, in the event of limited access to sufficient data, a similar approach to SMEs can be adopted.…”