Energy awareness for supercapacitors using Kalman filter state-of-charge tracking
Energy awareness for supercapacitors using Kalman filter state-of-charge tracking
复制标题
DOI：
10.1016/j.jpowsour.2015.07.050
复制
发表时间：
2015-11-20
影响因子：
9.2
通讯作者：
Soyata, Tolga
中科院分区：
文献类型：
--
作者：
Nadeau, Andrew;Hassanalieragh, Moeen;Soyata, Tolga
在能量缓冲替代品中，超级电容器可以提供无与伦比的效率和耐用性。此外，超级电容器的端电压和存储的能量之间的直接关系可以提高能量意识。然而，简单的电容近似不能充分地表示超级电容器中存储的能量。结果表明，三个分支等效电路模型提供了更准确的能量意识。该等效电路使用三个电容和相关联的电阻来表示超级电容器的内部SOC（充电状态）。然而，SOC不能从端子电压的一次观察确定，并且必须使用不精确的测量随时间跟踪。我们提出; 1）用于跟踪SOC的卡尔曼滤波解决方案; 2）用于有效地估计等效电路的参数的在线系统识别过程;以及3）针对5 F、10 F、50 F和350 F超级电容器的参数估计和SOC跟踪两者的实验验证。验证是在太阳能供电应用的操作范围内以及由于能量收集而导致的相关功率可变性内完成的。所提出的技术是基准对简单的电容模型和先验参数估计技术，并提供了一个67%的预测可用的缓冲能量的均方根误差减少。(C)2015爱思唯尔B. V.保留所有权利。
Among energy buffering alternatives, supercapacitors can provide unmatched efficiency and durability. Additionally, the direct relation between a supercapacitor's terminal voltage and stored energy can improve energy awareness. However, a simple capacitive approximation cannot adequately represent the stored energy in a supercapacitor. It is shown that the three branch equivalent circuit model provides more accurate energy awareness. This equivalent circuit uses three capacitances and associated resistances to represent the supercapacitor's internal SOC (state-of-charge). However, the SOC cannot be determined from one observation of the terminal voltage, and must be tracked over time using inexact measurements. We present; 1) a Kalman filtering solution for tracking the SOC; 2) an on-line system identification procedure to efficiently estimate the equivalent circuit's parameters; and 3) experimental validation of both parameter estimation and SOC tracking for 5 F, 10 F, 50 F, and 350 F supercapacitors. Validation is done within the operating range of a solar powered application and the associated power variability due to energy harvesting. The proposed techniques are benchmarked against the simple capacitive model and prior parameter estimation techniques, and provide a 67% reduction in root-meansquare error for predicting usable buffered energy. (C) 2015 Elsevier B.V. All rights reserved.