Data-Driven Controller Tuning for Battery Energy Storage Systems: A Set-Membership Approach

Abstract

Optimizing resources in industrial processes requires precise control strategies but often relies on complex mathematical models that are difficult to develop, posing challenges for traditional control methods. To solve this problem, an innovative methodology for fine-tuning controllers is presented based on the set-membership data-driven (SMDD) approach. This novel technique can change the traditional methods used for tuning controllers in the industry by leveraging real-world data and presents a transformative alternative in controller design. To demonstrate the effectiveness of the proposed methodology, it is applied in a controller design process for a real battery energy storage system (BESS) with a peak capacity of 267 kW, which is connected to the main grid. The study conducts a detailed comparative analysis between conventional tuning techniques based on simple open-loop step response characteristics, an existing controller for the BESS tuned by trial and error, and the SMDD methodology. The data-driven approach shows significant improvements in the system performance with reductions of up to 18% in step response time and overshoots under 1%, proving its robustness at dealing with time-variant communication delays and fixed controller structures. The study concludes that the SMDD methodology emerges as a tool with the potential for tuning controllers in real-world systems.