Explainable Data-Driven Battery State of Charge Estimation

作者：Shalman Ojukwu, Sidharth Maheshwari, Mohamed Mamlouk, Volker Pickert, Alex Yakovlev, Rishad Shafik · 年份：2025 · DOI：10.1109/ecce-europe62795.2025.11238681 · 被引用次数：2 · 研究领域：Advanced Battery Technologies Research、Electric Vehicles and Infrastructure、Energy Load and Power Forecasting

In the domain of battery energy storage systems for Electric Vehicles (EVs) applications and beyond, the adoption of machine learning techniques has surfaced as a notable strategy for battery modeling. Machine learning models are primarily utilized to forecast the state of batteries, specifically focusing on analyzing the state of charge (SOC). A crucial barrier to the adoption of machine learning algorithms in SOC estimation is their lack of explainability. Current ML SOC estimators researched, are highly focused on model performance while neglecting model explainability. This lack of explainability is crucial for stakeholders as they require trust and assurance in models' predictions, particularly applications like EVs, where erroneous SOC estimation can cause potential safety hazards. This work focuses on the explainability of SOC estimation using the Tsetlin Machine. The Tsetlin machine (TM) is a novel intrinsically explainable machine learning algorithm that offers competitive accuracy across various applications with simple Boolean-based logic computation. This paper aims to bridge the gap of model explainability by also comparing the explainable AI technique SHAP with the novel TM algorithm. The results present the TM as a highly competitive algorithm for accurate SOC estimation while offering a precise quantification of the influence of individual attributes to the final prediction without the need for post-hoc techniques.