應用模糊回歸模型於鋰電池殘電量估測

THE STATE-OF-CHARGE ESTIMATION FOR LI-ION BATTERY BY THE FUZZY C-REGRESSION MODEL

曾唯瑜(Wei-yu Tseng)

指導教授 ： 龔宗均

大同大學/電機資訊學院/電機工程學系所/碩士(2012年)

摘要

在本篇論文中，提出鋰電池(Li-ion battery)的模糊分群模型(Fuzzy C-Regression Models, FCRM)和殘餘電量(state of charge, SOC)的估測。此篇應用等效電路模型(equivalent circuit model, ECM)去近似鋰電池的特性。可以看出所有的參數(Rt, Rps, Rpl, Cps, Cpl and Voc) 不是常數，所有參數都是跟著殘電量變化的。因此，我們應用模糊分群演算法去將離散的參數轉換成連續的函數。基於模糊分群模型和等效電路模型，我們可以針對鋰電池做殘電量的估測。直流內電阻法(direct current internal resistance, DCIR)和開路電壓測量法(open circuit voltage measurement method)被使用去找出所有鋰電池每百分比的殘電量參數。經由提出一個有著電壓誤差的修正項的方法來求殘電量值。提出一個估測殘電量方法使用了有著2組並聯RC網路的等效電路模型比有著1組並聯RC網路的等效電路模型還好。比較殘電量均方根誤差(root mean square, RMS) 在第一個模擬中由原本的 2.61%下降到 2.52%。最後我們估測針對真實情況下的殘電量，如:太陽能燈、手機和電腦。殘電量的估測可利用修正R_t的方法來得到更準確的值。在實驗結果可以發現真實情況下的殘電量均方根誤差小於6%。

關鍵字

殘電量估測

並列摘要

In this thesis, the fuzzy c-regression models (FCRM) of the Li-ion battery and its application to estimate the state of charge (SOC) is proposed. The equivalent circuit model (ECM) is applied to approximate characteristic of the Li-ion battery. It is seen that all the parameters of ECM such as Rt, Rps, Rpl, Cps, Cpl and Voc are not constants. Those parameters are vary with the state of charge. Thus we apply the fuzzy c-regression models clustering algorithm to converted discrete parameters to a continuous function. Based on these fuzzy c-regression models and the ECM, we can estimate the SOC for the given Li-ion battery. The direct current internal resistance (DCIR) method and open circuit voltage measurement method are used to find parameters of each state of charge of Li-ion battery. Through, the proposed method is evaluated SOC by the correction gain with voltage error. The proposed SOC estimation using the ECM with two RC parallel networks is more effective and reliable than that using the ECM with one RC parallel network. The comparison shows that the root mean square (RMS) SOC estimation error reduces from 2.61% to 2.52% in simulation1. Finally, we estimate the SOC for real situations such as solar lamp posts, cell phone and computer. The SOC estimation will be more accurate by the modify Rt method. The simulation results shown that the root mean square (RMS) SOC estimation error less than 8%.

並列關鍵字

state of charge

參考文獻

[2] S. Lee, J. Kim, J. Lee and B.H. Cho, “State-of-charge and capacity estimation of lithium-ion battery using a new open-circuit voltage versus state-of-charge,” J. Power Sources , vol. 185, pp. 1367-1373, August. 2008.

[5] IL. S. Kim, “A Technique for Estimating the State of Health of Lithium Batteries Through a Dual-Sliding-Mode Observer,” J. Power Sources, vol. 25, no. 4, April, 2010.

[6] L. M. Hofland, E. J. Stofel and R. K. Taenaka, “Galileo probe Li-SO2 battery cell life testing,” IEEE Aerosp. Electron Syst., pp. 14-18, June, 1996.

[7] G. Alber, and M. W. Migliaro, “Impedance testing-is it a substitute for capacity tests,” Telecommunications Energy Conference, pp. 245-249, 1994.

[8] IL. S. Kim, “The novel state of charge estimation method for lithium battery using sliding mode observer,” J. Power Sources, vol. 163, pp. 584-590, September, 2006.