基于卡尔曼滤波的动力电池SOC估计算法设计

发布时间：2018-12-06 09:43

【摘要】：电池荷电状态(State Of Charge, SOC)的准确估计是电动汽车电池充放电控制和动力优化管理的重要依据,直接影响电池的使用寿命和汽车性能。本文针对电动汽车动力电池的SOC估计问题,主要进行了以下几个方面的研究工作： 1.采用映射近似对模型进行线性化,引入环境温度比例系数和充放电倍率比例系数来确定折算库伦效率,设计了一种基于复合模型的卡尔曼滤波算法。仿真结果表明,所设计的算法具有更好的修正累计误差和初值误差的能力。 2.采用加权统计线性回归法来实现模型函数线性化,基于电池复合模型状态方程线性的特性,通过将标准卡尔曼滤波算法和基于加权统计线性回归法的卡尔曼滤波算法组合,并引入奇异值分解,设计了一种基于奇异值分解的卡尔曼滤波算法。仿真结果表明,所设计的算法具有比基于复合模型的卡尔曼滤波算法更好的运算效率,以及更好的收敛速度和估计精度。 3.为了实现算法具有应对突变状态的强跟踪能力和应对模型不准确的鲁棒性,基于强跟踪原理,引入次优渐消因子,设计了一种基于强跟踪的卡尔曼滤波算法。仿真结果表明,所设计的算法具有比基于复合模型的卡尔曼滤波算法和基于奇异值分解的卡尔曼滤波算法更高的估计精度和更快的收敛速度。
[Abstract]:The accurate estimation of battery charge state (State Of Charge, SOC) is an important basis for battery charge and discharge control and dynamic optimization management of electric vehicle, which directly affects the battery service life and vehicle performance. In this paper, the SOC estimation of electric vehicle battery is studied in the following aspects: 1. Using mapping approximation to linearize the model and introducing environmental temperature ratio coefficient and charge-discharge ratio coefficient to determine the conversion Coulomb efficiency, a Kalman filter algorithm based on composite model is designed. Simulation results show that the proposed algorithm has better ability to correct cumulative error and initial error. 2. The weighted statistical linear regression method is used to linearize the model function. Based on the linear characteristic of the state equation of the battery composite model, the standard Kalman filter algorithm and the Kalman filter algorithm based on weighted statistical linear regression method are combined. A Kalman filter algorithm based on singular value decomposition is designed by introducing singular value decomposition. Simulation results show that the proposed algorithm has better computational efficiency, better convergence speed and better estimation accuracy than the Kalman filter algorithm based on composite model. 3. In order to realize the strong tracking ability of the algorithm to deal with the mutation state and the robustness of the model inaccuracy, a Kalman filter algorithm based on strong tracking is designed based on the strong tracking principle and the suboptimal fading factor. The simulation results show that the proposed algorithm has higher estimation accuracy and faster convergence speed than the Kalman filtering algorithm based on composite model and singular value decomposition.
【学位授予单位】：浙江工业大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TM912;U469.72

【参考文献】