Comprehensive review of battery state estimation strategies using machine learning for battery Management Systems of Aircraft Propulsion Batteries

The battery-powered propulsion system is introduced in the literature as a suitable solution for the CO2 emission challenge induced by aviation. However, because of design and manufacturing factors, during or after abused operational and environmental situations, Lithium-Ion battery (LIB) safety, and reliability cannot be guaranteed. Thus, an effective Battery Management System (BMS), is an essential unit in the Electric Propulsion System (EPS) of Electric Aircraft. Battery state estimation and prediction are vital to providing required safety strategies through acquiring battery data such as current, voltage, and temperature. Various methods of state estimation are practically and technically analyzed and offered in the literature including physics-based, model-based, and data-driven approaches. Among them, the recent method seems to be a novel solution to overcome the current experimental difficulties and inaccuracies. In a data-driven method, the battery is considered as a black box while a large volume of data is applied to learn the internal dynamics of the battery, using Artificial Intelligence (AI) and Machine Learning (ML) approaches. However, there are still major uncertainties and hurdles in the application and using AI in EPS due to data source scarcity, the complexity of computation, and ambiguities in the airworthiness certification process. In this study, a systematic literature review is performed; 948 papers were selected to be analyzed precisely in both qualitative and quantitative approaches to provide descriptive, metadata, and BMS function analysis reports. The goal of the research is to review BMS strategies supported by intelligent algorithms to propose appropriate solutions for battery management of EPS based on the proposed BMS necessary functions. Moreover, current airworthiness certification regulations are analyzed, and it is shown that the existing status is insufficient to satisfy critical issues for employing data-driven methods in the battery management of future electric aircraft including AI safety risk assessment and learning assurance. Finally, trends show an increase in studies on the subject of AI themes application in battery state estimation during the last ten years, especially for the State of Charge and the State of Health. However, there are still gaps in research for the application of intelligent technology in State of Function (SOF) and State of Power (SOP) estimation as one of the most imperative functions of the BMS in EA, which consists of less than 1 % of the total studies in this field.