锂离子动力电池便携式智能检测系统的设计与实现

摘要

摘要

作为二十一世纪主要的动力电源之一,锂离子动力电池广泛的运用在电动汽车、航空航天等军事和民用领域。为了确保锂电池在使用过程中可靠性和安全性,需要对其关键参数进行准确检测和估计。但是剩余电量(SOC,State Of Charge),无法像电流、电压等可通过测量直接获得,需要结合相关参数和数学方法估算得到。因此,在对SOC估算研究的基础上,针对现有检测系统体积笨重、价格昂贵等缺点,设计一套便携式智能化检测系统能对锂电池性能参数进行检测具有重要的现实意义。本文主要完成工作如下:

针对锂电池Thevenin非线性模型中不可线性化的参数难以辨识的问题,在有限量测实验数据基础上,采用了梯度下降算法通过构造目标函数,逐次接近设定阈值从而得到目标辨识值。该方法与最小二乘拟法算法结果进行分析比较,前者方法简洁、精度高。并通过Simcape对已辨识Thevenin模型进行了仿真和验证,并在其基础上进行了改进,通过误差分析新的模型适用性更强。

SOC作为锂电池的一个重要性能参数,目前采用扩展卡尔曼滤波方法在对其进行估算,但是在线性化过程中泰勒展开省略高阶项易造成精度下降和发散。无迹卡尔曼滤波算法通过无迹变换方式处理系统均值和协方差问题,较扩展卡尔曼滤波算法精度有显著提高。而迭代无迹卡尔曼滤波算法通过迭代,把无迹卡尔曼滤波算法结果作为算法预测过程中的初值,进一步提高了SOC算法精度。通过锂电池恒流放电,脉冲放电,QC/T897-2011实验对以上算法进行验证与分析,得出一种迭代无迹卡尔曼滤波算法和安时积分法结合的联合算法,并可运用于工程实际中。

设计与实现了一套便携式锂电池智能检测系统。系统由硬、软件构成,支持锂电池性能参数的采集,包括电压、电流、温度以及锂电池剩余电量SOC的估算,采集精度满足QC/T897-2011标准。系统具有完善的人机交互界面,可查看所需数据的波形图和存储数据,同时具备了包括串口、USB外部接口,提高了系统的实用性。

关键词:锂电池,SOC估算,Thevenin电路模型,卡尔曼滤波,锂电池检测系统

Abstract

ABSTRACT

As one of the major power sources in the 21st century, lithium-ion power batteries are widely used in military and civilian fields such as electric vehicles and aerospace. In order to ensure the reliability and safety of the lithium battery during use, it is necessary to accurately detect and estimate its key parameters. However, the remaining capacity (SOC, State Of Charge) cannot be directly obtained through measurement such as current, voltage, etc. It needs to be estimated by combining relevant parameters and mathematical methods. Therefore, based on the research of SOC estimation, aiming at the shortcomings of the existing detection systems such as bulky and expensive, it is of great practical significance to design a portable intelligent detection system to detect the performance parameters of lithium batteries. This article mainly completes the work as follows:

Aiming at the problem that the non-linearizable parameters in the non-linear model of the lithium battery Thevenin are difficult to identify, based on the limited measurement experimental data, a gradient descent algorithm is adopted to construct the objective function and successively approach the set threshold to obtain the target identification value. Compared with the results of the least squares algorithm, the former method is simple and accurate. Through Simcape, the identified Thevenin model is simulated and verified, and improved on the basis of it, and the new model is more applicable through error analysis.

As an important performance parameter of lithium battery, SOC is used to estimate by extended Kalman filter(EKF), but in the process of linearization, Taylor's expansion of high-order items tended to result in decreased precision and divergence. The unscented Kalman filter (UKF) solves the problem of mean and covariance of the system through unscented transformation, and has a significant improvement over the EKF accuracy. Through iterative methods, the accuracy of the SOC is further improved by using the result of the unscented Kalman filter algorithm as the initial value in the algorithm prediction process. Through the constant current discharge of lithium battery, pulsed discharge, QC/T897-2011 experiment to verify and analyze the relevant algorithms, a joint algorithm combining IUKF and Ampere-hour integration method is

相关主题
相关文档
最新文档