Control and estimation strategies for Battery Management Systems

Battery management systems (BMSs) play a critical role in the control and monitoring of battery operations, particularly for lithium-ion batteries. This thesis focuses on several key aspects of BMSs, aiming to improve their effectiveness and reliability. BMSs are essential to the control and optimization of battery performance, ensuring safety, efficiency, and longevity. This thesis focuses on strategies for control and estimation in battery systems. The challenges are posed by the complexity of lithium-ion battery systems, and only a few quantities such as temperature, current, and voltage are usually measurable. The thesis addresses state estimation in battery systems, considering the challenges of limited number of sensors. Most used approaches in this field rely on mathematical models like Equivalent Circuit Models (ECMs) and Electrochemical Models (EMs). Novel set-based techniques, including intervals, zonotopes, and constrained zonotopes, are introduced for the first time in this field. These set-based approaches, assuming bounded uncertainty and noise, improve fault detection and analysis. The proposed scheme, utilizing constrained zonotopes, efficiently detects thermal faults in battery cells despite unknown but bounded uncertainties, surpassing traditional methods. Furthermore joint estimation of states and parameters in battery systems has been explored, highlighting the impact of uncertain parameters on equivalent circuit models (ECMs). The analysis emphasizes the importance of careful model selection and design for accurate estimation and observability. The combination of accurate state parameter estimation and advanced modeling plays a crucial role in addressing control challenges in lithium batteries. This approach helps optimize charging strategies and mitigate the effects of aging. Whitin this aspect the thesis focuses on optimizing the control of Battery Management Systems (BMSs), with a specific aim at finding the best charging strategies for lithium-ion batteries while minimizing the impact of aging. The approach involves utilizing surrogate models that combine static nonlinear models dependent on the state of charge with finite-dimensional linear-time-invariant model. The strategy is designed to enhance the charging process while making sure aging-related and safety limits are met, thus promoting the long-term health and performance of battery systems. To enable accurate closed-loop control, a Kalman Filter with a forgetting factor is employed for state estimation, improving the precision of the control system. In the context of lithium-ion batteries, known for their balanced performance, cost-effectiveness, and lifespan in energy storage, the Battery Management System plays a crucial role. The BMS aims to strike the right balance between fast charging and minimizing aging effects while ensuring safety requirements are met. The goal is to minimize side reactions, optimizing charging while meeting aging-related and safety constraints. Simulation results using a DFN battery model as a representation of the actual system demonstrate the effectiveness of the proposed strategy. To complete the closed-loop system, a Kalman Filter with a forgetting factor is used for state estimation, providing adaptability to the control system. To sum up, the results presented in this thesis establish a crucial foundation for the improvement of Battery Management Systems (BMSs), contributing to enhanced robustness and efficiency in the safe and optimal operation of battery systems.