近年來綠能(Green Power) 意識高漲,其中碳排放量和石油危機是兩個重要的 議題,電動車即是順應這股趨勢所發展的產物。電動車不可缺少的元件之一為其 電池,而由於鋰電池放電量大且具穩定、充放電效率高、循環壽命長、體積小等 優點,已逐漸成為電動車動力來源的主流。評估電動車之電池效率常用到SOH (State of Health)與SOC (Stare of Charge),其中有關SOC 電池電量的量測技術以庫 倫積分法(Coulomb Integral Method) 為主,由於其方便與精確等特性,已廣泛被 應用在電動車SOC 之量測,但現今尚未有技術能準確估測SOC,特別的,量測與 評估的不確定性常被忽略。就此,本研究針對庫倫積分法量測計算上的隨機誤差, 建立數學上的馬可夫模型(Markov Model) ,藉以評估電動車行駛過程中SOC 逐 時下降的趨勢,以及任意使用時間點,處於各不同狀態SOC 之機率。本研究依據 他人實驗所得之電壓、放電量等數據,建立馬可夫模型,並考量電池出廠時的電 壓隨機變異性,以及電池在特定條件下充放電循環過程中老化衰減之隨機性,得 到鋰單電池在各使用時間點電量用盡之機率,最後再應用系統可靠度理論,推演 討論電池模塊(Battery Pack) 之 SOC。本研究所提出的方法能簡單有效模擬估算 真實環境下電池系統動態放電之SOC,並透過馬可夫分析模型得到各使用時間點 電量耗盡的機率,提供製造廠商進行符合ISO26262 等規範所需之風險評估。
With the growing consciousness of green power in recent years, the emissions of carbon dioxide and oil crisis have been two critical issues. The electric vehicle is one of many products developed with this trend. Batteries are indispensable to electric vehicles. In particular, lithium batteries have gradually become the main power sources of electric vehicles for the merits such as the stability and high efficiency of discharge, long cycle life, small size, etc. When it comes to the way to evaluate the efficiency of electric vehicles, State of Health (SOH) and State of Charge (SOC) are often applied. The latter is mainly based on Coulomb integral method, and its property of convenience and precision is widely applied to the SOC estimation of electric vehicles. However, no available techniques can thoroughly be applied to estimate SOC. Therefore, this research aims to direct the random errors of the estimation of the Coulomb integral method and establish a Markov model for assessing the exhaustion of SOC along with any time of discharge. This research is based on the voltage and capacity data from the reference to establish the Markov model and take random differences and aging of batteries into account. By doing so, we can come to realize the possibility of any battery exhaustion clearly and then apply the system reliability theory to estimate the SOC of batteries pack. With this research, we can efficiently estimate the SOC of dynamic discharge of battery systems, and through the Markov model obtain the probability distribution of battery exhaustion at any time. The results can be used for manufacturing companies to carry out risk assessment for standards such as ISO26262.