A Data-Driven Combining Equivalent Circuit Model Robust Fault Diagnosis Method for Lithium-Ion Batteries Based on Signal Reconstruction

Fault diagnosis of lithium-ion batteries (LIBs) has been an important issue in recent years. This article proposes a robust fault diagnosis method for LIBs by combining the equivalent circuit model (ECM) and data-driven model. First, the extended Kalman filter (EKF) is employed to identify the parameters of LIBs. Second, the sample entropy of voltage and current is calculated, along with a fault factor <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">T</i><sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">sign</sub> as an extra fault feature to enhance the depth of fault features. Then, seven features, including voltage and current, are combined into a multidimensional mixed fault feature as the input of the model. Finally, the autoencoder with long short-term memory (LSTM) network is utilized to reconstruct the input signal. The residuals of the reconstructed features are calculated as the fault indicator to achieve accurate fault diagnosis. Using fault factor <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">T</i><sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">sign</sub> as an extra fault feature enhances the robustness of the proposed method and facilitates the identification of special faults in certain cases. Experiments are conducted by designing the short-circuit faults of batteries and the effectiveness and robustness of the method are validated. The results show that in fault diagnosis, sometimes key feature extraction is more effective than building complex networks.