Multi-Fault Diagnosis for Series-Connected Lithium-Ion Battery Packs Based on Improved Sensor Topology and Correlation Coefficient Method

In this paper, the multi-fault diagnosis problem is investigated for series-connected lithium-ion battery packs based on an improved correlation coefficient method. Different from existing correlation-based fault diagnosis methods having difficulties in distinguishing between sensor faults and connection faults, a novel sensor topology is proposed to separate sensor faults from connection faults. Furthermore, to solve the problem of inconsistent correlation coefficients in different working states, an adaptive threshold mechanism is proposed. By combining the improved sensor topology and adaptive threshold mechanism, multi-fault detection and diagnosis can be realized. Subsequently, through different characteristics of the correlation coefficient and voltages, the types of sensor faults, such as the bias fault, scaling fault, drift fault, sticking state and noise, can be determined. Experiments prove the effectiveness and superiority of the proposed scheme. Note to Practitioners—This paper focuses on the multi-fault diagnosis of lithium-ion battery packs, aiming to enhance system robustness and longevity. The string-level redundancy topology structure greatly simplifies the fault diagnosis process without increasing the hardware cost and system complexity, which greatly reduces the calculation amount. With the adaptive threshold mechanism, the consistency of correlation coefficients in different working states is improved, so the accuracy of fault diagnosis and the robustness to different working states are increased. This paper uses four batteries in series to form a battery pack. In theory, the proposed method is equally effective when the number of cells in the pack rises. Thus, our multi-fault diagnosis method has great potential to be applied to electric vehicles and energy storage systems and adeptly addresses complex and dynamic operational environments. Our research has been validated in practical battery packs, yielding significant successes. By effectively identifying and addressing various faults, we have improved the overall performance and reliability of battery packs.