外文翻譯--基于鋰離子電池管理系統(tǒng)can總線的電動汽車

1、<p>　　中文4080字，2080單詞</p><p>　　畢業(yè)設(shè)計（論文）外文翻譯</p><p><b>　　（附外文原文）</b></p><p>　　系 別： 汽車工程系 </p><p>　　專業(yè)班級： 汽服102班 &l

2、t;/p><p>　　姓 名： </p><p>　　學(xué) 號： </p><p>　　指導(dǎo)教師： </p><p>　　基于鋰離子電池管理系統(tǒng)CAN總線的電動汽車&

3、lt;/p><p>　　摘要 —電池管理系統(tǒng)(BMS)基于CAN總線設(shè)計的由許多串聯(lián)電池并分散地分布在電動汽車（EV）上的鋰離子電池組。 該BMS包括一個主模塊和幾個采樣模塊。 采樣電路的硬件設(shè)計和CAN擴展電路進行了介紹。 還介紹了電池SOC（荷電狀態(tài)）估計和電池安全管理的策略。</p><p><b>　　一、引言</b><

4、;/p><p>　　電池是電動汽車(EV)的一個重要組成部分的性能深深影響著電動汽車的動力性能和經(jīng)濟性能。 閥控式鉛酸（VRLA）電池，鎳氫（Ni-MH）電池和鋰離子（Li-ion）電池通常用作電動汽車的能量來源[1]。鋰離子電池是一種理想的電池由于其高功率比和能量比[2]。 鋰離子電池組通常由幾十電池的串聯(lián)連接，因為一個電池單元的電壓不夠高來驅(qū)動電機的。有這么多的電池單元的電池組，該電池組由于其

5、復(fù)雜的安全管理的使用不方便。 因此，電池管理系統(tǒng)（BMS）被用于電池荷電狀態(tài)的（SOC）估計，電池的安全性管理和通訊與電動車的整車控制器（HCU）[3]。</p><p>　　CAN（控制器區(qū)域網(wǎng)絡(luò)）是一個實時的分布式控制的串行通信網(wǎng)絡(luò)[4]。CAN總線廣泛用于汽車上的電氣控制設(shè)備由于其通信波特率高,可靠性強。如果CAN總線使用可以減少汽車上的線束。</p><p>　　在本文中

6、，我們開發(fā)基于CAN總線的考慮用于電動車的鋰離子電池組的結(jié)構(gòu)和功能要求，在分布式電池管理系統(tǒng)。 我們開發(fā)的BMS可以采集電池的信息，估計電池的SOC，并與HCU溝通。</p><p><b>　　硬件設(shè)計</b></p><p>　　A、 電池管理系統(tǒng)的結(jié)構(gòu)</p><p>　　電動汽車的鋰離子電池組裝備由80個鋰離子電池。電池的額

7、定電壓288V,額定容量55Ah。圖1為電池組的組成。</p><p>　　圖1  電池組的結(jié)構(gòu)</p><p>　　電池組電池分為兩個電池的情況下被固定在車身底部。各電池盒具有兩個電池模塊。在上述情況下前側(cè)的電池，每個模塊有22個電池單元，在背面每個模塊的18個電池。 然后將電池模塊通過外部導(dǎo)線與中間接觸器組合成電池組。</p><p>　　由

8、于其對安全工作嚴格要求，在電池組的每個鋰離子電池單元的電壓應(yīng)進行采樣。 另外，在各電池模塊至少4典型點的溫度應(yīng)該被采樣。 因此大量的信號分布在不同電池模塊被采樣。 如果采用集中式的BMS需要許多取樣線，一定有許多抽樣電線鑿穿電池模塊和電池的情況。 因此安裝和調(diào)試BMS會不方便和危險的事情往往發(fā)生。 因此BMS的基于CAN總線的分布式式開發(fā)。 兩個獨立的CAN總線用在我們開發(fā)的B

9、MS為了避免數(shù)據(jù)干擾。 一個CAN總線命名內(nèi)的CAN用于BMS內(nèi)部的通信，另一個外部能夠與HCU通信。 成熟BMS的結(jié)構(gòu)如圖 2所示。</p><p>　　圖2  基于CAN總線開發(fā)BMS的結(jié)構(gòu)</p><p>　　BMS由一個位于背面電池外殼的主模塊和四個位于相應(yīng)的電池模塊的采集模塊。采樣模塊采樣電池模塊的對應(yīng)電池單元的電壓和溫度。然后將電池模塊的

10、信息,如電壓和溫度是由內(nèi)部CAN總線傳輸。主模塊從內(nèi)部CAN總線接收電池的信息和整個電池組的電壓和電流通過傳感器接收到電池的信息。</p><p>　　然后根據(jù)取樣的電池信息對電池的SOC估計和電池安全進行管理。 此外主模塊與HCU根據(jù)通信協(xié)議的電動汽車， 主模塊可與PC機通過RS232接口使電池的信息可以顯示在電腦屏幕上，BMS可在線調(diào)試。</p><p><b

11、>　　B、 采樣模塊</b></p><p>　　電池信息,比如需要知道電池的電壓、溫度,整個電池電壓和電流時對電池SOC估計和電池進行安全管理。</p><p>　　每個電池模塊有18個或22個電池單元的串聯(lián)連接。這些電池的電壓的數(shù)字和他們不是相對于相同的“地球”。所以巡航這些電壓采樣電路是為抽樣設(shè)計不同的“地球”。設(shè)計見圖3。在同一時間只有一個電池的電壓通過繼電器陣

12、列的精確控制連接到單片機的AD采樣電路。 </p><p>　　當(dāng)采樣的電池單元1的電壓，相關(guān)的繼電器是通過控制信號閉合。 例如，采樣的2號電池B2電壓，繼電器2與繼電器3被關(guān)閉，繼電器Relay_Even也被關(guān)閉，因為取樣的電池的序列號是一個。 從而對電池B2的電壓被連接到節(jié)點V +和V-，然后由MCU的AD轉(zhuǎn)換單元進行采樣。</p><p>　　圖3

13、60; 巡航電路，用于采樣的電池單元的電壓</p><p>　　該控制信號必須準確，可靠的。 否則，電池可能會通過繼電器短路，如果控制信號是錯誤的或紊亂的。 需要另外的I / O端口來控制繼電器陣列。 所以，用CPLD（復(fù)雜可編程邏輯器件）是用來產(chǎn)生控制信號。 CPLD適用于復(fù)雜邏輯設(shè)計由于其高可靠性和大量的可編程I / O端口。 只有單片機的5個I / O端口需

14、要通過CPLD控制繼電器如果有22個電池。</p><p>　　在電池模塊中的溫度由單線總線器件DS18B20測量。 許多單線器件可以只連接到只有一個單片機的I / O端口。 不同的單線器件可以彼此通過在他們的唯一標識代碼來區(qū)分。 該DS18B20為-55°C?125°C的溫度測量范圍和0.0625°C的轉(zhuǎn)換精度 該DS18B20的轉(zhuǎn)換時間小于

15、750毫秒。 接口電路簡單和AD轉(zhuǎn)換單元可以省略，如果DS18B20作為溫度傳感器。單片機和DS18B20之間的電路連接圖4所示。</p><p>　　圖4  電路連接到DS18B20的單片機</p><p>　　霍爾元件被用作高電壓傳感器和電流傳感器，以便到BMS的電磁干擾可以被降低。</p><p><b>　　C、 主控模塊&l

16、t;/b></p><p>　　在BMS的主控模塊將執(zhí)行許多功能，包括測量電池的電壓和電流，處理電池的信息數(shù)據(jù)，并估計電池SOC。 所以，一個DSP（數(shù)字信號處理器）芯片，被命名為LF2407A，是由德州儀器公司制造出來使用。 命名SJA1000控制器芯片也擴展，因為BMS需要兩個單獨的CAN接口，但只有一個CAN接口可以通過LF2407A。但是，LF2407A和SJA1000互相不兼容

17、，因為數(shù)據(jù)總線和LF2407A的地址總線是分開的，但對SJA1000共享。 因此橋梁是需要連接LF2407A和SJA1000。用CPLD實現(xiàn)這個邏輯功能和方案圖5所示 。 D07?D00這是LF2407A的數(shù)據(jù)總線的數(shù)據(jù)連接到AD7?AD0。SJA1000的數(shù)據(jù)/地址共享總線，可發(fā)送“地址”值或“數(shù)據(jù)”時，當(dāng)LF2407A訪問SJA1000。 A15?A13這是LF2407A的地址總線的一部分，用

18、于通過邏輯編碼器擔(dān)任由CPLD生成芯片選擇信號SJA1000的CS。 LF2407A的包括選擇IO空間的信號，則讀或?qū)懶盘朢 / W和地址總線A15 ~ A13用于生成控制信號通過CPLD SJA1000讀信號RD,寫信號WR和地址鎖定信號。</p><p>　　圖5 通過CPLD LF2407A和SJA1000之間的連接</p><p>　　表 1表示在CPLD為LF

19、2407A和SJA1000之間橋梁的邏輯設(shè)計。 對SJA1000被映射到地址0 x8000 ~ 0 xbfff LF2407A IO和空間的訪問由LF2407A通過CPLD的上讀取和寫入時序為它模擬訪問。 都需要LF2407A的IO空間兩個連續(xù)操作指示LF2407A讀取或?qū)懭隨JA1000。 例如，當(dāng)LF2407A打算讀某個地址上SJA1000的數(shù)據(jù)，LF2407應(yīng)該寫通過“寫'地址'”指令

20、“地址”到D07?D00的值，那么數(shù)據(jù)將出現(xiàn)在D07?D00他們應(yīng)該通過“讀'數(shù)據(jù)'”指令讀取。 同樣，當(dāng)LF2407A打算將數(shù)據(jù)寫入到某個地址上的SJA1000，LF2407應(yīng)該寫通過“寫'地址'”指令“地址”到D07?D00，那么“數(shù)據(jù)”的值應(yīng)寫入到D07?D00通過“寫'數(shù)據(jù)'”指令。</p><p><b>　　表1</b>&

21、lt;/p><p>　　在CPLD的LF2407A和SJA1000之間的橋梁邏輯設(shè)計</p><p><b>　　C、 可靠性設(shè)計</b></p><p>　　在電動汽車運行時，存在著強烈的電磁干擾，這可能會影響測量的精度和通信的可靠性。 所以這是非常重要的，研究BMS如何抵御干擾。BMS通過隔離，過濾等方式試圖減少干擾。</p&

22、gt;<p>　　該BMS是由12V電池的電動汽車提供動力。 DC / DC轉(zhuǎn)換器是用于BMS隔離12-V電池電，因為12-V電池可以帶來一些電磁干擾。 也是一個電阻串聯(lián)連接在所述測量線的電池單元和相應(yīng)的繼電器，從而限制了電流小范圍的的電流之間的，電池單元是通過測量導(dǎo)線短路時，BMS被擾亂。 此外,隔離放大器連接之前的單片機可以減少測量電線的干擾效應(yīng)。</p><p>

23、　　另外，光電耦合器所使用的原因是通信電纜減少干擾和保護BMS的核心電路來隔離通信電路。 數(shù)字低通濾波器是處理數(shù)據(jù)采樣期間使用(1)。</p><p>　　y(n)=ay(n-1)+(1-a)s(n) (1)</p><p>　　這里，y（n）是該時間的結(jié)果，y（n-1）是下一個時間的結(jié)果，s（n）是該時間的采樣值，和一個是

24、過濾器的系數(shù)。 有一個更大的，更好的過濾器但濾波器響應(yīng)緩慢。</p><p><b>　　電池SOC估算</b></p><p>　　估計電池SOC是BMS實現(xiàn)的最重要功能之一。</p><p>　　普通裝置庫侖計數(shù)來估計電池SOC和用開路電壓法（OCV）推斷SOC)[5]。庫侖數(shù)敏感電流比重錯誤,所以它不適合工作了很長一段時間。等要

25、適合情況的推斷SOC開路或工作在恒流情況下。但電動汽車上的工作電流在一個大范圍變化迅速，所以要意味著也不適合。最近的研究重點是估算電池SOC使用模糊邏輯[6]，自適應(yīng)神經(jīng)模糊推理系統(tǒng)(簡稱ANFIS)[7]等等。這些方法需要存儲大量的電池的歷史數(shù)據(jù)和復(fù)雜的使用它們。本文采用一種基于卡爾曼濾波的方法[8]?？柭鼮V波方法采用一些遞歸方程和電池的歷史數(shù)據(jù)不需要存儲。電池組被視為由輸入的卡爾曼濾波方法驅(qū)動的系統(tǒng)。電池SOC狀態(tài)變量, 電流是輸

26、入到系統(tǒng)中，并在電池電壓是可觀察到的輸出。 該方法的步驟是圖6所示，并給出一個簡短的描述如下:</p><p>　　圖6  基于卡爾曼濾波的電池SOC估算方法</p><p>　　?初始化。當(dāng)車輛,BMS讀取從其上存儲的車輛被關(guān)閉之前。電池的影響被認是“自放電</p><p>　　?估計，評估。當(dāng)時k + 1, 估計根據(jù)該時刻k和時間k的當(dāng)前的電

27、池的SOC。</p><p>　　?更新輸出。 按照電池模型計算的輸出（這里是電池電壓）。</p><p>　　?更新輸出誤差。 根據(jù)所觀察到的輸出和計算出的輸出計算出的輸出的誤差。</p><p>　　?更新的權(quán)重。 根據(jù)輸出誤差估計濾波器的權(quán)重。</p><p>　　?修正。 正確顯示電池SOC的數(shù)

28、值。估計的k +1時刻的SOC值。</p><p>　　圖7顯示了在道路測試基于卡爾曼濾波的SOC估計方法的仿真。 電池的SOC結(jié)果從庫侖計數(shù)法的結(jié)果。 為很短的時間周期中工作時的庫侖計數(shù)方法可以是可信的。 計算周期時間是0.2秒。 估計誤差被初始化為0.15（真正的SOC為約0.55，但推定SOC被初始化為0.70）。 我們可以借鑒的圖 7，其估計誤差

29、小于0.02后250計算步驟。 它表明，基于卡爾曼濾波的SOC推定方法可以估算電池SOC估計方法基于卡爾曼濾波能正確估計誤差快速估算過程中電池SOC。</p><p>　　圖7  基于卡爾曼濾波SOC的估算方法道路測試期的仿真</p><p>　　卡爾曼濾波方法可以糾正錯誤的電池SOC估計根據(jù)新的觀測數(shù)據(jù),從而可以提高電池SOC估算的準確性。此外,該方法可以估計SOC估

30、計的誤差,從而提高SOC估計的可信性。</p><p><b>　　電池安全管理</b></p><p>　　安全管理電池組也是BMS的一個重要功能，尤其是對一個鋰離子電池組。</p><p>　　電池組的安全管理是基于電池組故障診斷。故障診斷根據(jù)電池故障信息采樣并生成故障代碼,然后傳送這些代碼通過CAN總線傳輸這些代碼到整車控制器。</

31、p><p>　　產(chǎn)生兩個級別的故障，即I級故障和II級故障，故障是在表中定義。  I級故障是一種嚴重的故障發(fā)生，這將導(dǎo)致在BMS傳輸通過CAN總線事后切斷接觸中，為了電池的安全性的斷電要求的HCU如果采取HCU沒有相應(yīng)的操作當(dāng)它生。 II級故障是一個警告信號，這將導(dǎo)致在BMS通過CAN總線當(dāng)它發(fā)生在一個傳輸報警信號，HCU，以及HCU應(yīng)及時調(diào)整控制策略，以降低電池電流，以避免進一步損壞電

32、池。</p><p><b>　　結(jié)論和總結(jié)</b></p><p>　　在BMS（電池管理系統(tǒng)）是一種電池組和EV（電動車）的橋梁，這可以提高性能和電池組的可行性。 本文開發(fā)了基于CAN總線BMS可以減少線束和方便安裝和調(diào)試。 成熟的BMS可以檢查到電池單元電壓，電流和溫度與可靠的采樣電路。 另外，根據(jù)電池的SOC可以通過基于卡爾曼濾波

33、的SOC估計策略的手段來估計。 該BMS可以實現(xiàn)安全管理的電池組根據(jù)不同級別的故障診斷的結(jié)果。開發(fā)BMS的結(jié)果成功地工作由天津清遠電動汽車制造公司。</p><p><b>　　致謝</b></p><p>　　這項工作是由國家高技術(shù)研究發(fā)展計劃中國（863計劃）（No.2003AA501630）的贊助。 對天津清源電動車輛有限公司也表示感謝。&l

34、t;/p><p>　　A Li-ion Battery Management System Based on</p><p>　　CAN-bus for Electric Vehicle</p><p>　　Abstract- A battery management system (BMS) based on the CAN-bus was designed for

35、the Li-ion battery pack which consisted of many series-connected battery cells and was distributed dispersedly on the electric vehicle (EV). The BMS consisted of one master module and several sampling modules. The hard

36、ware design of the sampling circuit and the CAN expanding circuit was introduced. The strategies of the battery SOC (state of charge) estimation and the battery safety mana-gement were also present</p><p>　

37、　I. INTRODUCTION</p><p>　　Battery pack is an important component for the electric vehicle (EV) of which the performance deeply affects the power performance and the economy performance of the EV. Valve reg

38、ulated lead acid (VRLA) battery, nickel metal hydride (Ni-MH) battery and lithium ion (Li-ion) battery are generally used as the energy source on the EV [1]. Li-ion battery is an ideal battery due to its higher power ra

39、tio and energy ratio [2]. The Li-ion battery pack consists usually of tens of battery cells which a</p><p>　　CAN (Controller Area Network) is a serial communication network for real-time distributed control

40、 [4]. CAN-bus is widely used in the electrical controlled equipment on the automobiles because of its high communication baud rate and strong reliability. The wire harness on the automobile can be reduced if a CAN-bus i

41、s used.</p><p>　　In this paper, we develop a distributed battery management system based on the CAN-bus considering the structure and function requirement of the Li-ion battery pack used for the EV. The BMS

42、we developed can sample the battery’s information, estimate the battery SOC and communicate with the HCU.</p><p>　　II. HARDWARE DESIGN</p><p>　　Structure of Battery Management System</p>

43、<p>　　The Li-ion battery pack equipped on the EV consists of 80 i-ion battery cells. The nominal voltage of the battery pack is 288V and the nominal capacity 55Ah. Fig. 1 shows thestructure of the battery pack. &l

44、t;/p><p>　　The battery pack is divided into two battery cases and is fixed on the bottom of the car body. Each battery case has two battery modules. Each module in the foreside battery case has 22 battery cel

45、ls and each module in the backside 18 battery cells. Then the battery modules are combined to a battery pack through external wires and the mid contactor. </p><p>　　The voltage of each Li-ion battery cell i

46、n the battery pack should be sampled because of its rigorous demands on the safe working. Also the temperatures of at least 4 typical spot in each battery module should be sampled. Thus a large amount of signals distri

47、buted on the different battery modules should be sampled. There must be many sampling wires which should be drilled through the battery modules and battery cases if a centralized type of BMS is adopted. As a result ins

48、talling and debuggi</p><p>　　The BMS is made up of one master module which is located in the backside battery case and four sampling modules which are located in the corresponding battery modules. The sampl

49、ing module samples the voltages of the battery cells and the temperatures in the corresponding battery module. Then the battery module’s information such as the voltages and the temperatures is transmitted to the inner-

50、CAN bus. The master module receives the battery’s information from the inner-CAN bus and samples the </p><p>　　Fig. 1. Structure of the Battery Pack</p><p>　　Fig. 2. Structure of the develope

51、d BMS based on CAN-bus</p><p>　　Then the battery SOC is estimated and battery safe management is carried out according to the battery information sampled. Moreover the master module communicates with the HCU

52、 according to the CAN protocol of the EV. The master module can also communicate with PC through the RS232 interface so that the battery’s information can be displayed on the screen of the PC and the BMS can be debugged

53、on line.</p><p>　　Sampling Module</p><p>　　The battery information such as the voltages of the battery cells, the temperatures, the whole battery voltage and the current is needed to know when t

54、he battery SOC is estimated and the battery safe management is carried out.</p><p>　　There are 18 or 22 battery cells which are series-connected in each battery module. These battery cells are of great numb

55、ers and the voltages of them are not relative to the same “earth”. So a cruising sampling circuit is designed for sampling these voltages of different “earth”. The design is illustrated in Fig. 3. At the same time onl

56、y one battery cell’s voltage is connected to the AD sampling circuit of the MCU via the exact control of the relay array.</p><p>　　When sampling the voltage of one of the battery cells, the relevant relays a

57、re closed via the controlling signals. For example, when sampling the voltage of the 2nd battery B2, the relays Relay2 and Relay3 are closed and the relay Relay_Even is also closed because the sequence number of the s

58、ampled battery is an even. Thus the voltage of the battery B2 is connected to the nodes V+ and V– and then be sampled by AD converting unit of the MCU.</p><p>　　Fig. 3. Cruising circuit for sampling the vo

59、ltages of the battery cells</p><p>　　The controlling signals must be generated exactly and reliably. Otherwise the battery may be shorted via the relays if the controlling signals are mistaken or inordinate

60、. Moreover many I/O ports are needed to control the relay array. So a CPLD (Complex Programmable Logical Device) is used to generate the controlling signals. The CPLD is suitable for the complex logical design because

61、 of its high reliability and large numbers of programmable I/O ports. Only 5 I/O ports of the MCU are needed </p><p>　　The temperature in the battery module is measured by 1-Wire bus device DS18B20. Many 1

62、-Wire devices can be connected to only one I/O port of the MCU. Different 1-Wire devices can be distinguished from each other through the unique identity code in them. The DS18B20 has a measurement range of –55°C~

63、125°C and a converting precision of 0.0625°C. The converting time of the DS18B20 is less than 750ms. The interface circuit is simple and the AD converting unit can be omitted if the DS18B20 is use</p>

64、<p>　　Hall components are used as the high voltage sensor and the current sensor so that the electromagnetic interference to the BMS can be reduced.</p><p>　　Fig. 4. Circuit connecting DS18B20 to MCU&l

65、t;/p><p>　　Master Module</p><p>　　The master module of the BMS will carry out a lot of functions including measuring the battery pack’s voltage and the current, dealing with the data of the battery

66、 information and estimating the battery SOC. So a DSP (Digital Signal Processor) chip which is named LF2407A and is manufac-tured by Texas Instrument Ltd. is used. A CAN controller chip named SJA1000 is also expanded bec

67、ause the BMS need two individual CAN interface but only one CAN interface can be supplied by LF2407A. Unfortunately, </p><p>　　Fig. 5. Bridge between LF2407A and SJA1000 via CPLD</p><p>　　Table

68、. I denote the logic design of the CPLD for the bridge between LF2407A and SJA1000. The SJA1000 is mapped to the address 0x8000~0xBFFF of LF2407A’s IO space and is accessed by LF2407A via the CPLD’s simulation on the re

69、ad and written timing to it. Two sequential operating instructions to LF2407A’s IO space are needed when LF2407A intends to read or write SJA1000. For example, when LF2407A intends to read the data of a certain address

70、 on SJA1000, LF2407 should write the value of the “addr</p><p><b>　　TABLE I</b></p><p>　　LOGIC DESIGN OF THE CPLD FOR THE BRIDGE BETWEEN LF2407A AND SJA1000</p><p>　　Des

71、ign for Reliability</p><p>　　A strong electromagnetic interference exists when the EV is running, which can affect the accuracy of the measurement and the reliability of the communication. So it is very imp

72、ortant to study how to resist the interference to the BMS. The BMS tries reducing the interference to the BMS by means of isolating, filtering and so on.</p><p>　　The BMS is powered by the 12-V battery on t

73、he EV. A DC/DC converter is used to power the BMS which can isolate the 12-V battery electrically because the 12-V battery can bring some electromagnetic interference. Also a resistor is connected in series between the

74、 measurement wire for the battery cell and the corresponding relay, which can limit the current to a small scale in case of that the battery cell is shorted via the measurement wires when the BMS is disturbed-severely an

75、d the relay array</p><p>　　In addition, photo-couplers are used to isolate the communication circuit by reason of reducing the interference from the communication cables and protect the core circuit of the B

76、MS. A digital low pass filter is used during processing the data sampled as in (1).</p><p>　　y (n) =ay (n-1) + (1-a) s (n) (1)</p><p>　　Here, y(n) is the result

77、of this time, y(n–1) is the result of lasttime, s(n) is the sampling value of this time, and a is the coefficient for the filter. The greater a is, the better the filter works but the slowly the filter responses.</

78、p><p>　　III. BATTERY SOC ESTIMATION</p><p>　　Estimating the battery SOC exactly is one of the most important functions the BMS should realize.</p><p>　　Ordinary means to estimate the

79、battery SOC are coulomb counting and inferring SOC from open circuit voltage (OCV) [5]. The coulomb counting is sensitive to current measurement error, so it is not suitable for working for a long time. Inferring SOC f

80、rom OCV is suitable for situations such as open circuit or working at a constant current. But the working current on the EV is varying rapidly and in a large range, so the OCV means is also not suitable. More recent st

81、udies are focused on estimati</p><p>　　Fig. 6. Battery SOC Estimation method based on Kalman filtering</p><p>　　? Initialization. When the vehicle is turned on, the BMS reads the value of the

82、 battery SOC from the EEPROM on the BMS which was stored before the vehicle was turned off. The effect of the battery “self-discharged” is considered. </p><p>　　? Estimation. At the time k+1, estimate the

83、 battery SOC according to the battery SOC of the time k and the current of the time k. </p><p>　　? Updating the output. Calculate the output (here, the battery voltage) according to the battery model. <

84、/p><p>　　? Updating the output error. Calculate the error of the output according to the output observed and the output calculated. </p><p>　　? Updating the weight. Estimate the weight of the f

85、ilter according to the output error. </p><p>　　? Correction. Correct the battery SOC according to the weight. The SOC estimated of the time k+1 is exported.</p><p>　　Fig. 7 shows the simulati

86、on on the SOC estimation method based on the Kalman Filtering during the roadway testing. The true battery SOC results from the coulomb counting method. The coulomb counting method can be credible when working for a sh

87、ort period of time. The calculating period time is 0.2s. The estimation error is initialized to 0.15 (the true SOC is about 0.55 but the estimated SOC is initialized to 0.70). We can draw from the Fig. 7 that the esti

88、mation error is less than 0.02 afte</p><p>　　The Kalman filtering method can correct the error of the battery SOC estimated according to the new data observed and thus can improve the accuracy of estimating

89、the battery SOC. Furthermore, the method can estimate the error of the estimated SOC, which can improve the creditability of estimated SOC.</p><p>　　Fig. 7. Simulation on the SOC estimation method based on

90、 Kalman filtering</p><p>　　During roadway testing</p><p>　　IV. BATTERY SAFETY MANAGEMENT</p><p>　　The safety management for the battery pack is also an important function of the BM

91、S, especially for a Li-ion battery pack.</p><p>　　The safety management for the battery pack is based on diagnosing malfunction to the battery pack. The BMS diagnoses the malfunction according to the batter

92、y information sampled and generates the malfunction codes, then transmits these codes to HCU via CAN-bus.</p><p>　　Two levels malfunctions are generated, i.e. level-I malfunction and level-II malfunction, an

93、d malfunctions are defined in Table. II. Level-I malfunction is a severe malfunction which will lead the BMS to transmit a power-down requirement to the HCU via CAN-bus and afterwards cut off the mid contactor for the

94、 sake of battery safety if no appropriate operation is taken by HCU when it happens. Level-II malfunction is a warning signal which will lead the BMS to transmit a warning signal to the H</p><p>　　V. CONCL

95、USION AND SUMMARY</p><p>　　The BMS (Battery Management System) is a bridge between the battery pack and the EV (Electric Vehicle), which can be used to improve the performance and the reliability of the batt

96、ery pack. This paper has developed a BMS based on the CAN-bus which can reduce the wire harness and is convenient to install and debug. The developed BMS can sample the voltage of the battery cells, current and tempera

97、ture with a reliable sampling circuit. Also the battery SOC can be estimated by means of a SOC esti</p><p>　　ACKNOWLEDGMENT</p><p>　　This work was supported by a grant from the National High Te