The paper presents an improved multiple feature-electrochemical thermal coupling (MF-ETC) modeling method for lithium-ion batteries, focusing on low-temperature performance degradation. The method aims to accurately predict the state of charge (SOC) by considering multi-dimensional information and real-time coefficient correction. Key contributions include: 1. **MF-ETC Modeling**: An improved model that captures the dynamic characteristics and multi-parameter coupling relationships in battery models, separating short-term and long-term voltage changes. 2. **Adaptive Parameter Identification**: An adaptive asynchronous parameter identification strategy for low-temperature environments, improving prediction accuracy across a wide range with real-time feedback correction. 3. **Decoupled Deviated Extended Kalman Filtering (DD-EKF)**: A real-time estimation strategy for battery SOC based on decoupled deviation-extended Kalman filtering, reducing cumulative errors caused by current variations. 4. **Experimental Validation**: The proposed method is validated under complex conditions, including Beijing bus dynamic stress tests (BBDST) and dynamic stress tests (DST), with maximum SOC prediction errors of 4.57% and 0.223%, respectively. The paper also discusses the mathematical analysis of the MF-ETC model, the compound modeling of low-temperature conditions, adaptive asynchronous parameter identification, and the real-time statistical coefficient correction process. Experimental results demonstrate the effectiveness of the proposed method in accurately predicting SOC under various conditions.The paper presents an improved multiple feature-electrochemical thermal coupling (MF-ETC) modeling method for lithium-ion batteries, focusing on low-temperature performance degradation. The method aims to accurately predict the state of charge (SOC) by considering multi-dimensional information and real-time coefficient correction. Key contributions include: 1. **MF-ETC Modeling**: An improved model that captures the dynamic characteristics and multi-parameter coupling relationships in battery models, separating short-term and long-term voltage changes. 2. **Adaptive Parameter Identification**: An adaptive asynchronous parameter identification strategy for low-temperature environments, improving prediction accuracy across a wide range with real-time feedback correction. 3. **Decoupled Deviated Extended Kalman Filtering (DD-EKF)**: A real-time estimation strategy for battery SOC based on decoupled deviation-extended Kalman filtering, reducing cumulative errors caused by current variations. 4. **Experimental Validation**: The proposed method is validated under complex conditions, including Beijing bus dynamic stress tests (BBDST) and dynamic stress tests (DST), with maximum SOC prediction errors of 4.57% and 0.223%, respectively. The paper also discusses the mathematical analysis of the MF-ETC model, the compound modeling of low-temperature conditions, adaptive asynchronous parameter identification, and the real-time statistical coefficient correction process. Experimental results demonstrate the effectiveness of the proposed method in accurately predicting SOC under various conditions.