The project is focused in driving the industrial chemical plants towards effective use of resources and energy. Effective plant management will be reached as a synergy of tools for production planning and for advanced automatic feedback control. The technology enabling the reaching of these goals is based on the use of data a) for creation of input-output data-based models or of first-principles models with corrective data-based terms and b) for reliable monitoring of unmeasured process variables. The improved mathematical models are subsequently used for optimization of steady-state operating regimes and for optimization-based control of industrial plants. The designed algorithms and control structures are tested in simulations as well as in laboratory conditions. The project also stimulates cooperation with industry.
Keywords: process control, modelling, uncertainty, robust model predictive control, neuro-fuzzy system, energy savings
Scientific goals
Scientific goals for whole period of this project Economic pressures require greater flexibility of (chemical or biochemical) production plants in processing of raw materials of different quality or taking into account various restrictions, e.g. to reduce the environmental impact of production. To achieve operational efficiency in such a context, production units must be managed in an integrated way, i.e. one of the main principles of Industry 4.0. Most plants use planning tools to determine the production targets of individual production units, taking into account the price of products and customer orders. However, due to the integrated nature of production facilities, it is possible to achieve the set goals through several production routes, which is not always used even in the modern facilities. The aim of the work is therefore the development of methods for selecting efficient production routes in terms of consumption of materials and energy in real time. This objective, contained in the RIS3 priorities, will be achieved through the development of accurate predictive mathematical models enabling plant-wide optimization and by improving the functionality and efficiency of control systems to ensure that the desired objectives are met for each unit. The main feature of the developed algorithms will be the use of data a) for off-line design of monitoring, optimization and control algorithms using (large amounts of) historical data (big data) and b) for real-time processing of on-line data to improve the response of automatic control. The obtained results will be published at renowned conferences and in top-tier journals. They will also be implemented in open source software packages and made available on the internet. The project will also make efforts to transfer the results into practice with an industrial partner, Slovnaft, a.s. The developed methods should achieve the acceptance of the proposed solution by the company's management and operators. Thus, the project not only proposes theoretical methods but also implements them in open software.