基于双卡尔曼滤波算法的磷酸铁锂电池建模及SOC估计

发布时间：2019-03-13 18:55

【摘要】：当前世界的能源储备急剧减少,环境污染问题也变得越来越严峻。大力开发能够替代传统能源并且对环境无污染的新能源越来越成为大家重视的问题。在汽车领域,世界各国都加大了对新能源汽车的研究。动力电池作为电动汽车的动力来源,是能够影响电动汽车发展程度的一个重要因素。相比于其他电池,磷酸铁锂电池由于其优越的性能在作为电动汽车动力电池方面脱颖而出,应用越来越广泛。但磷酸铁锂电池存在单体电池间同一性较差的问题,因此设计一套对电池组进行管理的电池管理系统(Battery Management System,BMS)是非常关键的。对电池的荷电状态(State of charge,SOC)进行准确地估计是电池管理系统能够有效运行的核心和关键。本文以一款50AH的磷酸铁锂电池作为研究对象,对其建立电池模型,并在该模型的基础上重点研究SOC的估计方法。论文主要工作及成果如下:1、对电池SOC估计的研究背景进行了详细的介绍,介绍磷酸铁了锂电池的优点和特性和当前对电池模型和电池SOC估计研究的现状,为后文对本文研究对象磷酸铁锂电池进行电池建模和SOC估算建立了基础。在对磷酸铁锂电池的工作原理和主要特性进行分析和总结的基础上设计了实验对电池特性进行测定。最后,介绍了当前得到广泛认可的SOC的定义方法,并在传统SOC定义方法的基础上进行了改进,得到了动态SOC的定义方法,这是后文对电池进行建模和对电池进行SOC估计的理论依据。2、对电池的四种等效模型进行了分析和比较,最终确定二阶RC模型作为本文研究电池的模型,考虑到电池单体间的同一性较差,因此在二阶RC模型上做出改进,得到了改进二阶RC模型,并对模型进行了公式推导,并在Matlab中进行了仿真分析,验证了模型的准确性。3、详细介绍了 Kalman算法的基本原理,并在经典卡尔曼滤波算法的基础上对适用于非线性系统的扩展卡尔曼滤波算法进行了原理介绍和公式推导。采用经典卡尔曼滤波器和扩展卡尔曼滤波器相结合的双卡尔曼滤波算法联合估计电池SOC和电池模型参数,通过实验及Matlab仿真在横流放电工况和脉冲放电这两种工况下验证了双卡尔曼滤波算法联合估计电池SOC和电池模型参数方法的准确性。4、研究了基于CKF估计电池SOC的方法,并将这种估算方法和基于UKF估计电池SOC方法进行了比较,最后通过仿真实验发现基于CKF估计电池SOC具有更高的精确性。5、利用最小二乘支持向量机(Least Squares Support Vector Machine,LSSVM)构建LSSVM模型,在此基础上实现对电池SOC的估计,并引入粒子群优化算法(PSO)以提高训练效率与模型精度。通过恒流放电实验和脉冲充放电实验验证了 PSO-LSSVM方法对电池SOC估计的有效性。
[Abstract]:At present, the world's energy reserves are sharply reduced, and environmental pollution is becoming more and more serious. More and more attention has been paid to the development of new energy which can replace the traditional energy and not pollute the environment. In the field of automobiles, countries all over the world have increased their research on new energy vehicles. As the power source of electric vehicle, power battery is an important factor that can influence the development degree of electric vehicle. Compared with other batteries, lithium iron phosphate battery is more and more widely used as electric vehicle power battery because of its superior performance. However, lithium iron phosphate batteries have the problem of poor identity among single batteries, so it is very important to design a battery management system (Battery Management System,BMS) for battery pack management. Accurate estimation of the charge state (State of charge,SOC) of the battery is the core and key to the effective operation of the battery management system. In this paper, a 50AH lithium iron phosphate battery is taken as the research object, and the battery model is established. On the basis of this model, the SOC estimation method is mainly studied. The main work and achievements are as follows: 1. The research background of battery SOC estimation is introduced in detail. The advantages and characteristics of iron phosphate lithium battery and the current research status of battery model and battery SOC estimation are also introduced. The foundation of the battery modeling and SOC estimation for the lithium ferric phosphate battery studied in this paper is established. Based on the analysis and summary of the working principle and main characteristics of the lithium iron phosphate battery, an experiment was designed to measure the characteristics of the battery. Finally, the definition method of SOC, which has been widely accepted, is introduced and improved on the basis of traditional SOC definition method, and the definition method of dynamic SOC is obtained. This is the theoretical basis of the battery modeling and SOC estimation. 2, the four equivalent models of the battery are analyzed and compared. Finally, the second-order RC model is selected as the model of the battery in this paper. Considering the poor identity between cells, this paper improves the second order RC model, obtains the improved second order RC model, deduces the formula of the model, and simulates the model in Matlab to verify the accuracy of the model. The basic principle of Kalman algorithm is introduced in detail. On the basis of classical Kalman filtering algorithm, the extended Kalman filter algorithm suitable for nonlinear systems is introduced in principle and formula derived. The dual Kalman filter algorithm combined with classical Kalman filter and extended Kalman filter is used to jointly estimate the parameters of battery SOC and battery model. The accuracy of the dual Kalman filter algorithm to jointly estimate the battery SOC and battery model parameters is verified by experiments and Matlab simulation under the condition of cross-flow discharge and pulse discharge. 4. 4. The method of estimating battery SOC based on CKF is studied. The method is compared with the SOC method based on UKF. Finally, the simulation results show that the estimation of battery SOC based on CKF has higher accuracy. 5. The least squares support vector machine (Least Squares Support Vector Machine,) is used to estimate the battery SOC with least square support vector machine (LSVM). LSSVM) constructs the LSSVM model, then realizes the estimation of battery SOC, and introduces the particle swarm optimization algorithm (PSO) to improve the training efficiency and model precision. The validity of the PSO-LSSVM method for SOC estimation is verified by the constant current discharge experiment and the pulse charge-discharge experiment.
【学位授予单位】：西南交通大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TM912

【参考文献】