State of charge estimation approaches for domestic thermal storage using phase change material

This paper discusses the development and implementation of State of Charge (SoC) estimation methods for a domestic-scale Phase Change Materials (PCM) system, aimed at enhancing energy storage system management and efficiency. Addressing challenges such as the complexity of physical processes and the need for real-time algorithm operation, the study integrates physics-based insights with statistical analysis of training data to accurately predict the thermal energy stored. The estimation process relies on a historical dataset of temperature measurements collected from multiple points within the PCM system, providing a comprehensive understanding of thermal behaviour over time. The research involves an experimental program for data collection, starting with linear regression analysis of temperature data to assess thermal performance and standing loss, followed by piecewise linear approximation to characterise the transition between sensible and latent heat processes. Advanced regression techniques, specifically Gaussian Process (GP) models trained on diverse operational data, demonstrate strong predictive capability in forecasting system performance. By leveraging historical temperature profiles, this approach significantly enhances SoC estimation accuracy for PCM systems and optimises charging and discharging cycles in domestic settings. The findings contribute valuable insights and methodologies to the field of thermal energy storage, supporting the development of more efficient and intelligent energy management strategies.