Process mining for multi-objective online control
LIACS – CWI – Tata Steel – BMW – MonetDB
In the steel coil production process (Tata Steel), the characteristics of steel coils are of critical importance throughout the manufacturing chain for which process parameters and installation settings have to be carefully controlled. Any deviation of the characteristics would lead to cracks during the steel forming process (BMW) later on. Normally, the process to resolve the problem cost hugely in terms of time, fault products, production loss and loss of source material.
In order to optimize the production and forming process of steel coils, PROMIMOOC aims at developing a generic platform to collect and integrate the data from the production process, model the process based the data, perform multiple-objective decision making based on data-driven models and finally give Pareto-optimal process control settings under complex constraints.
In the first place, the process data from BMW and Tata Steel is extracted, transferred and loaded to a MonetDB database a suitable format for large amounts of data.
Based on the process data, the relationship between process control data and product properties will be modelled via a data driven modelling process which uses state-of-the-art nonlinear regression algorithms. The automatic generation, comparison, selection and update of such models will be a major task within the project.
Based on sufficiently well performing, cross-validated models, an evolutionary multi-objective optimisation algorithm will then be used for developing an efficient frontier of Pareto-optimal solutions. The frontier can then be used to select an optimal compromise solution and associated control parameters to run the process optimally. Evolutionary strategies will be used for this task.
![]() |