电池状态估计是电池管理系统（BMS，Battery Management System）最核心的功能之一。电池状态主要包括荷电状态SOC（State of Charge），健康状态SOH（State of Health），功率状态SOP（State of Power）和能量状态SOE（State of Energy）。精确的电池状态估计是对车用电池及整车进行有效管理与控制的前提。本文以车用锂离子电池为研究对象，研究电池的状态估计问题，并针对电池各状态间存在的紧密关联，提出一种状态联合估计的研究思路。作为状态联合估计的基础，分别研究了电池SOC、SOH、SOP和SOE的单独估计算法；为了明确电池状态间的关联性，进一步分析了在进行某一状态估计时，其他电池内部状态或参数不准确对该状态估计结果的影响；在此基础上，研究电池SOC、SOH、SOP和SOE的联合估计方法，并基于实际BMS验证算法的实车适用性与估计精度。首先，对电池各状态的单独估计算法进行了研究。具体的，基于电池测试建立一阶RC等效电路模型，并采用扩展卡尔曼滤波算法实现电池SOC估计。为了实现电池SOH估计，采用带遗忘因子的递归最小二乘法在线辨识电池的开路电压和欧姆内阻，进一步的，基于开路电压辨识结果，结合两点间累计电量法实现电池容量在线估计。考虑电池的电压限制、电流限制、SOC限制条件，研究了基于一阶RC模型的动态SOP估计方法。针对剩余放电能量估计问题，提出了一种改进的基于未来电压预测的SOE估计方法，在未来工况预测的基础上，对未来电压、状态、参数进行耦合预测，最后根据未来电压预测值估计剩余放电能量。在进行电池状态单独估计研究的同时，着重分析了电池状态估计的关键影响因素，明确了各状态间的相互关联。采用理论推导与仿真分析相结合的方式，对电池SOC估计方法展开了详尽的误差分析，明确了主要误差源的误差传递过程及其对SOC估计误差的影响程度。通过仿真分析的方式，分别讨论了电池SOC、容量、内阻对电池SOP估计以及SOE估计的影响。最后，在建立单一状态估计算法的基础上，综合考虑各状态间的相互关联后，提出了基于一阶RC模型的电池SOC、SOH、SOP和SOE联合估计方法。首先通过不同工况下的台架试验验证，证明了联合估计方法相比单独估计的优势；然后通过在实际BMS平台的试验，验证了算法在BMS中的适用性及其实际估计精度。

Battery state estimation is a key function of the battery management system (BMS). The common battery states mainly include State of Charge (SOC), State of Health (SOH), State of Power (SOP), and State of Energy (SOE). Accurate state estimation is the precondition to guarantee the effective management of the batteries and electric vehicles.This paper mainly carries out a study on battery state estimation, taking a type of lithium-ion battery as the research object. Because of the close connection between the battery states, a co-estimation scheme which estimates the SOC, SOH, SOP and SOE simultaneously is proposed. As the basis of the states co-estimation, estimation approaches of each single battery state are established firstly. Further analysis is then implemented to clarify the interaction between the battery states. Finallly, a model-based co-estimation approach of the SOC, SOH, SOP and SOE is proposed and validated on a real BMS platform.Firstly, the estimation approach of single battery state is studied. A first-order RC equivalent circuit model is built based on the basic performance test. Then a model-based SOC estimation algorithm applying the extended Kalman filter is proposed. After that, the open circuit voltage (OCV) and the internal resistance are identified online by the recursive least square algorithm with a forgetting factor. And the identified OCV is further used to estimate the capacity. The dynamic model-based approach is applied to estimate the peak power, taking the voltage limits, the current limits and the SOC limits into consideration. And an improved battery remaining energy estimation method is developed. On the basis of the future operation condition prediction, the future terminal voltage, battery states and parameters are predicted, and the remaining energy is subsequently accumulated.Secondly, theoretical and simulation analysis is carried out to clarify the key factors of the battery state estimation and the interaction between the four states. A detailed error analysis of SOC estimation is conducted by theoretical derivation and simulation analysis. The transfer route of the main error sources and the exact impact on SOC estimation are figured out. Besides, the influence of the battery SOC, capacity and resistance on the SOP estimation and SOE estimation is studied by simulation analysis respectively.Finally, based on the previous study on the single state estimation and the interaction of the four states, a co-estimation approach of the SOC, SOH, SOP and SOE is developed. Validation tests are conducted both on the test bench and on the real BMS platform, which prove the high precision and the good applicability of the proposed approach.