基于电池SOC状态和未来能耗预测的电动汽车续驶里程估计

发布时间：2018-01-19 09:57

本文关键词： 粒子滤波 SOC估计能耗预测剩余里程估计　出处：《吉林大学》2016年硕士论文　论文类型：学位论文

【摘要】：随着环境污染的逐年加重和石油资源的逐年减少,环保和节能问题越来越引起人们的重视。为改变这种状况,全球各国政府与各汽车生产商,都在大力度投入发展新能源汽车。对比现有的各种类新能源汽车,纯电动汽车因其相较于其他新能源车具有更加明显的优势,已经成为了目前行业内研发的热点。但是由于许多关键技术还没有突破,其本身还存在着诸多问题。这诸多关键技术中就包括车载电源技术—电池能量管理。估计电池SOC状态和车辆续驶里程是能量管理系统两个重要功能,准确的SOC状态值是电动汽车剩余行驶里程预测的基础。能实时地为驾驶员提供续驶里程信息,能够提高用户对电动汽车的使用信心。因此,能够准确估计电池SOC并据此预测出车辆剩余续驶里程,能够极大地推动电动汽车普及,促进电动汽车发展。本文中对这两个问题分别展开了讨论。目前对于SOC估计的研究主要都是围绕电池单体展开,但车载动力电池是由很多单体电池组成的,如何对电池组剩余容量进行估计是应该重点研究的课题。本文利用实际的电池组实验数据,通过理论分析将电池组看作一个整体,进行SOC估计。首先建立电池等效电路模型并根据采集的电池组放电实验数据,利用递推最小二乘法对电池模型参数进行了辨识,之后对电池组SOC进行估计。根据建立的等效电路模型确定电池系统状态方程,并结合模型参数辨识的结果及电池组实测得的实验数据,应用粒子滤波完成电池组SOC状态的动态估计,并把估计结果与扩展卡尔曼滤波算法估计结果做了比较,表明基于粒子滤波的电池组SOC估计方法的效果更好,估计结果更准确。在确定了电池组剩余容量状态之后,为对电动汽车剩余里程进行预测,就需要对车辆未来行驶能耗进行估计。本文中提出了一种基于车辆过去平均行驶能耗预测未来平均行驶能耗的方法,利用支持向量机回归理论,对能耗差异与车辆行驶工况变化之间的关系进行分析。在确定了未来行驶能耗需求之后,基于SOC值计算电池的剩余能量值,将两者结合估算出电动汽车剩余可行驶里程。最后通过仿真实验,证明了应用该方法能达到比较好的里程估计效果。
[Abstract]:With the increasing of environmental pollution and the decreasing of petroleum resources, the problem of environmental protection and energy saving has been paid more and more attention. In order to change this situation, governments and auto manufacturers all over the world. Compared with the existing kinds of new energy vehicles, pure electric vehicles have more obvious advantages than other new energy vehicles. Has become the industry research and development hotspot. But because many key technologies have not yet broken through. There are still many problems in the system. Among these key technologies are the on-board power supply technology-battery energy management. Estimating the battery SOC status and vehicle driving mileage are two important functions of the energy management system. Accurate SOC state value is the basis of residual mileage prediction for electric vehicles. It can provide drivers with driving range information in real time and improve the users' confidence in the use of electric vehicles. It can estimate the battery SOC accurately and predict the remaining driving mileage of the vehicle, which can greatly promote the popularity of electric vehicles. In order to promote the development of electric vehicles, the two problems are discussed separately in this paper. At present, the research of SOC estimation is mainly focused on the battery cell, but the on-board power battery is composed of a lot of single cells. How to estimate the residual capacity of the battery pack is an important research topic. In this paper, the battery pack is regarded as a whole by theoretical analysis using the actual experimental data of the battery pack. Firstly, the equivalent circuit model of the battery is established and the parameters of the battery model are identified by using the recursive least square method according to the collected discharge experimental data. Then the SOC of the battery pack is estimated. According to the established equivalent circuit model, the state equation of the cell system is determined, and combined with the result of the model parameter identification and the experimental data of the battery pack measured. The particle filter is used to complete the dynamic estimation of the SOC state of the battery pack, and the estimation results are compared with the estimation results of the extended Kalman filter (EKF) algorithm. The results show that the SOC estimation method based on particle filter is more effective and accurate. After determining the residual capacity state of battery pack, the residual mileage of electric vehicle is predicted. It is necessary to estimate the vehicle energy consumption in the future. In this paper, a method based on the vehicle energy consumption in the past to predict the future average driving energy consumption is proposed, and the support vector machine regression theory is used. The relationship between energy consumption difference and vehicle driving condition change is analyzed. After determining the future driving energy consumption demand, the residual energy value of the battery is calculated based on the SOC value. The residual mileage of electric vehicle is estimated by combining the two methods. Finally, the simulation results show that the proposed method can achieve better mileage estimation results.
【学位授予单位】：吉林大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：U469.72

【相似文献】