Joint project I4.0SDD
Development of an Industry 4.0 Smart Data Device for Modernizing Existing Plants
- 01.01.2021 to 31.03.2024
- Industrial Systems
Many companies in the field of production and process technology strive for higher levels of automation to increase product quality, production flexibility, and reduce control setup effort. However, they often face challenges with existing plants that require upgrades and expansions to achieve such automation levels. Modernizing existing plants may lead to problems like high investment costs, or incompatibility of new hardware and software with the existing infrastructure, and production downtime. The technical limitations of older systems and the increased complexity in their updating can make it problematic for companies to bring existing plants up to the latest (control) technology.
This research project's motivation is to develop, together with our project partner ELZET 80, a device for intelligent control that can be integrated into existing plants without replacing existing components.
Project Goals and Methods
The Smart Data Device (SDD) to be developed in this project aims to modernize existing plants by providing an intelligent control solution through a combined hardware and software approach. It will enable highly precise, self-adjusting control of existing systems. The device will analyze measured process data, identify a mathematical description of the system behavior, and utilize it in a model-based control approach. By efficiently selecting data and automating model adaptation, the process control can be gradually improved. The integration of the SDD into the control of existing plants makes use of established and standardized communication protocols.
The SDD will analyze and evaluate measured process data, aggregate it, and make it available for process modeling. These models will be utilized in an automated model-based control system. The SDD's base module will use the initial system model for control. Before deploying the control or after significant changes to the system, an identification process will be performed to update this initial system model.
To autonomously respond to potential changes in the process during operation, the described SDD base module will be supplemented with an optional learning module. This module continuously identifies the process model and adapts it based on available process data to minimize potential model errors. The Gaussian Process Regression (GPR) method will be used for this purpose. GPR offers significant advantages for online system identification. It allows predicting the model error's expected value and provides information about the prediction's confidence, hence enabling conclusions about the prediction quality.
Additionally, GPR is computationally efficient and can be automatically adjusted to the underlying data. The model is determined based on input and output data, and prior knowledge about the process can be incorporated into the GPR structure, further enhancing the prediction. The flexibility and efficiency of the GPR method make it suitable for use in the SDD for online system identification.
Innovations and Perspectives
The SDD allows the expansion of existing plants to integrate modern automation and control concepts. Its data-driven, model-predictive control enables continuous model adaptation at runtime. Due to its compact design, it can be integrated into existing control cabinets of plants with minimal space requirements. The intelligent control helps avoid production downtime and reduces rejects regarding quality requirements. Therefore, the SDD significantly enhances the level of automation in existing processes with low integration efforts.
In summary, the SDD offers an attractive opportunity for modernizing industrial existing plants by combining online/offline system identification and machine learning. Its easy integration into the existing control eliminates the need for costly and time-consuming control replacements. Moreover, the use of SDD allows continued operation of existing plants, contributing to the resource-efficient implementation of Industry 4.0 requirements.