1.前言

    蓄電(dian)(dian)池的充(chong)(chong)電(dian)(dian)控制系統是個非線性(xing)[1]的、時變的、有(you)干擾的、具有(you)純(chun)滯(zhi)后的控制系統[2]。在充(chong)(chong)放電(dian)(dian)過程(cheng)中(zhong)涉及(ji)到很多(duo)參數，如充(chong)(chong)電(dian)(dian)率、最(zui)大允(yun)許充(chong)(chong)電(dian)(dian)電(dian)(dian)流(liu)、內(nei)阻、出(chu)氣點(dian)電(dian)(dian)壓、溫度(du)、壽命等。

   傳統(tong)的(de)控(kong)制系統(tong)是建(jian)立在被(bei)控(kong)對象(xiang)精(jing)確數學模型(xing)基(ji)礎(chu)上的(de)，如果被(bei)控(kong)對象(xiang)的(de)數學模型(xing)很復雜或者數學模型(xing)無法建(jian)立，控(kong)制系統(tong)就較難實現。蓄電池(chi)充電正(zheng)是屬于這種情況(kuang)，由于蓄電池(chi)的(de)充電過程有(you)自己獨(du)特的(de)物(wu)理化(hua)學規律，因此考(kao)慮(lv)采用模糊(hu)控(kong)制技術(shu)來進行蓄電池(chi)的(de)充電控(kong)制 [3]。

   模(mo)(mo)糊(hu)控(kong)(kong)制[4]是以模(mo)(mo)糊(hu)集合(he)理論為基礎的(de)(de)(de)控(kong)(kong)制手段(duan)，它是模(mo)(mo)糊(hu)系統理論、模(mo)(mo)糊(hu)技(ji)術與自動(dong)控(kong)(kong)制技(ji)術相結合(he)的(de)(de)(de)產(chan)物,出發(fa)點是操作人員的(de)(de)(de)控(kong)(kong)制經驗或相關專家的(de)(de)(de)知識(shi)，在設計中不需(xu)要建(jian)立被(bei)控(kong)(kong)對象的(de)(de)(de)精確數學模(mo)(mo)型[5]，因(yin)而使得控(kong)(kong)制機理和(he)策略易(yi)于(yu)接受(shou)與理解(jie)，設計簡單，便于(yu)應用。

2.模糊控制器的結(jie)構及算法(fa)

   基于(yu)微(wei)控制(zhi)器(qi)組(zu)成的(de)模糊控制(zhi)器(qi)包括模糊化(hua)接(jie)口、決策邏輯(ji)、知識庫和反(fan)模糊化(hua)接(jie)口四個(ge)部分，如圖1所示。

圖1模糊(hu)控(kong)制器的(de)結構框圖

Fig.1 The structure diagram of fuzzy controller

   在進行(xing)模糊(hu)控(kong)制算法(fa)設(she)計時，須(xu)首先考慮智能充(chong)電系(xi)統的(de)(de)技術要求，體現智能充(chong)電控(kong)制技術的(de)(de)優勢，解(jie)決(jue)長(chang)期困擾蓄電池裝備中的(de)(de)效(xiao)率和壽命問題，所(suo)以應當提高智能充(chong)電控(kong)制中對控(kong)制量的(de)(de)精度；其次，模糊(hu)控(kong)制算法(fa)必須(xu)高速(su)可靠(kao)，對外部檢測得到的(de)(de)物理(li)量要有(you)非常快(kuai)的(de)(de)反(fan)應速(su)度。采(cai)用新型大(da)存儲容量的(de)(de)微控(kong)制器解(jie)決(jue)使用查表法(fa)所(suo)帶來(lai)的(de)(de)需要大(da)容量內存的(de)(de)問題。

3 模糊控制器的設計

3.1雙(shuang)輸入(ru)單輸出模(mo)糊控制(zhi)器

    常見的(de)模糊控制器(qi)[6]根據輸(shu)(shu)(shu)入(ru)(ru)與輸(shu)(shu)(shu)出(chu)(chu)(chu)個(ge)數分為單輸(shu)(shu)(shu)入(ru)(ru)單輸(shu)(shu)(shu)出(chu)(chu)(chu)、雙輸(shu)(shu)(shu)入(ru)(ru)單輸(shu)(shu)(shu)出(chu)(chu)(chu)、多(duo)輸(shu)(shu)(shu)入(ru)(ru)單輸(shu)(shu)(shu)出(chu)(chu)(chu)和(he)雙輸(shu)(shu)(shu)入(ru)(ru)多(duo)輸(shu)(shu)(shu)出(chu)(chu)(chu)等幾種。平(ping)時應用較多(duo)的(de)是雙輸(shu)(shu)(shu)入(ru)(ru)單輸(shu)(shu)(shu)出(chu)(chu)(chu)模糊控制器(qi)。

圖2 雙輸(shu)入單輸(shu)出(chu)模糊控制器方框(kuang)圖

 Fig.2 The rectangular diagram of fuzzy cONtroller witch has two input units and one output unit

    圖2是(shi)雙輸(shu)入單輸(shu)出模(mo)(mo)糊(hu)(hu)控(kong)制(zhi)(zhi)器(qi)(qi)的(de)(de)方(fang)框圖。其中屬于(yu)論(lun)域(yu)X的(de)(de)模(mo)(mo)糊(hu)(hu)集(ji)合(he) 取自(zi)系(xi)統誤差e的(de)(de)模(mo)(mo)糊(hu)(hu)化，屬于(yu)論(lun)域(yu)Y的(de)(de)模(mo)(mo)糊(hu)(hu)集(ji)合(he) 取自(zi)系(xi)統誤差變(bian)化率 的(de)(de)模(mo)(mo)糊(hu)(hu)化，二者構成模(mo)(mo)糊(hu)(hu)控(kong)制(zhi)(zhi)器(qi)(qi)的(de)(de)二維輸(shu)入；屬于(yu)論(lun)域(yu)Z的(de)(de)模(mo)(mo)糊(hu)(hu)集(ji)合(he) 是(shi)反映控(kong)制(zhi)(zhi)量變(bian)化的(de)(de)模(mo)(mo)糊(hu)(hu)控(kong)制(zhi)(zhi)器(qi)(qi)的(de)(de)一維輸(shu)出，模(mo)(mo)糊(hu)(hu)控(kong)制(zhi)(zhi)器(qi)(qi)的(de)(de)控(kong)制(zhi)(zhi)規則(ze)[7]通常由模(mo)(mo)糊(hu)(hu)條件語(yu)句if    and    then  來表(biao)達(da)。

3.2精確量的模糊化

   模(mo)糊控制(zhi)系統中的(de)被控對象狀態變(bian)化是(shi)連(lian)續的(de)，系統給定值(zhi)也是(shi)連(lian)續的(de),反映(ying)到模(mo)糊控制(zhi)器輸入端的(de)輸入量也必然是(shi)連(lian)續的(de)。但(dan)模(mo)糊控制(zhi)器由計算機構成，它只能執行(xing)離(li)(li)散(san)處理[7]。因此模(mo)糊控制(zhi)器要(yao)求(qiu)輸入量是(shi)離(li)(li)散(san)模(mo)糊量，即論(lun)(lun)域是(shi)離(li)(li)散(san)的(de)。對連(lian)續論(lun)(lun)域要(yao)進行(xing)離(li)(li)散(san)化。連(lian)續論(lun)(lun)域[8]經(jing)過量化后(hou)就成為一個離(li)(li)散(san)論(lun)(lun)域[8]。

設有(you)連續論(lun)域[a，b]，而(er)量化之后的離(li)散論(lun)域為 ，亦即將連續論(lun)域分為2n段，則存在系數

                                                                   （1）

   隨(sui)后(hou)，在求出每條(tiao)規(gui)則的(de)強度之后(hou)，把相(xiang)互矛(mao)盾的(de)規(gui)則中強度較小(xiao)的(de)舍去；把相(xiang)同規(gui)則合成一條(tiao)規(gui)則，得到最(zui)后(hou)控制規(gui)則基。

3.3.2根據學(xue)習(xi)算法生成控制規則(ze)

   對(dui)被控(kong)(kong)對(dui)象執行手動控(kong)(kong)制(zhi)所得(de)到的(de)控(kong)(kong)制(zhi)規(gui)(gui)則(ze)(ze)是較粗(cu)糙(cao)的(de)，有(you)時還可能會出現控(kong)(kong)制(zhi)死區，一個控(kong)(kong)制(zhi)規(gui)(gui)則(ze)(ze)表(biao)中會出現空項(xiang),這是不能滿足實際控(kong)(kong)制(zhi)要求(qiu)的(de)。為(wei)了取(qu)得(de)更滿意的(de)控(kong)(kong)制(zhi)效果，可以(yi)對(dui)原始的(de)控(kong)(kong)制(zhi)規(gui)(gui)則(ze)(ze)進(jin)行改進(jin)。這時，應以(yi)粗(cu)糙(cao)的(de)控(kong)(kong)制(zhi)規(gui)(gui)則(ze)(ze)為(wei)基礎，通過仿真實驗和系(xi)統(tong)調試加以(yi)完善。

4.模糊智能充電系統的(de)工作原理及結構

   智能充(chong)(chong)電(dian)(dian)(dian)(dian)(dian)(dian)系(xi)統主要(yao)由(you)充(chong)(chong)電(dian)(dian)(dian)(dian)(dian)(dian)電(dian)(dian)(dian)(dian)(dian)(dian)源(yuan)和單(dan)片(pian)機(ji)控制電(dian)(dian)(dian)(dian)(dian)(dian)路兩(liang)部分(fen)(fen)組成(cheng)。220V的(de)(de)(de)交流(liu)市電(dian)(dian)(dian)(dian)(dian)(dian)經整(zheng)流(liu)濾(lv)波電(dian)(dian)(dian)(dian)(dian)(dian)路變(bian)為(wei)脈動的(de)(de)(de)310V高壓(ya)直流(liu)。然后經DC-DC變(bian)換電(dian)(dian)(dian)(dian)(dian)(dian)路（脈沖(chong)功率變(bian)壓(ya)器(qi)）變(bian)為(wei)充(chong)(chong)電(dian)(dian)(dian)(dian)(dian)(dian)所需(xu)的(de)(de)(de)60V直流(liu)電(dian)(dian)(dian)(dian)(dian)(dian)壓(ya)。為(wei)了保證(zheng)輸出電(dian)(dian)(dian)(dian)(dian)(dian)壓(ya)的(de)(de)(de)穩定性，采用(yong)了UC3842對60V直流(liu)電(dian)(dian)(dian)(dian)(dian)(dian)壓(ya)進行穩壓(ya)。二次斬(zhan)波電(dian)(dian)(dian)(dian)(dian)(dian)路主要(yao)由(you)MOSFET管(guan)、電(dian)(dian)(dian)(dian)(dian)(dian)感、電(dian)(dian)(dian)(dian)(dian)(dian)容和二極管(guan)組成(cheng)，輸出24-36V的(de)(de)(de)充(chong)(chong)電(dian)(dian)(dian)(dian)(dian)(dian)電(dian)(dian)(dian)(dian)(dian)(dian)壓(ya)。控制部分(fen)(fen)采用(yong)C504單(dan)片(pian)機(ji)，通過對蓄(xu)電(dian)(dian)(dian)(dian)(dian)(dian)池端電(dian)(dian)(dian)(dian)(dian)(dian)壓(ya)信號(hao)的(de)(de)(de)采集、分(fen)(fen)析處理、模糊(hu)(hu)推(tui)理[8]、模糊(hu)(hu)決策等，控制二次斬(zhan)波電(dian)(dian)(dian)(dian)(dian)(dian)路中的(de)(de)(de)MOSFET管(guan)的(de)(de)(de)通斷(duan)時間來(lai)控制充(chong)(chong)電(dian)(dian)(dian)(dian)(dian)(dian)電(dian)(dian)(dian)(dian)(dian)(dian)壓(ya)。控制部分(fen)(fen)還包(bao)括對電(dian)(dian)(dian)(dian)(dian)(dian)流(liu)和溫(wen)度的(de)(de)(de)采集以及電(dian)(dian)(dian)(dian)(dian)(dian)壓(ya)和電(dian)(dian)(dian)(dian)(dian)(dian)流(liu)的(de)(de)(de)顯示。總體結構如圖(tu)3所示。

圖3 智能充電系統總體結構框圖

5結論

  蓄電(dian)(dian)(dian)池(chi)的(de)(de)充(chong)(chong)(chong)放(fang)電(dian)(dian)(dian)過程是一個(ge)復(fu)雜的(de)(de)過程，要用精確數(shu)學模型(xing)對蓄電(dian)(dian)(dian)池(chi)充(chong)(chong)(chong)電(dian)(dian)(dian)的(de)(de)控制則有(you)(you)相當的(de)(de)難度。蓄電(dian)(dian)(dian)池(chi)的(de)(de)充(chong)(chong)(chong)電(dian)(dian)(dian)控制系統(tong)(tong)是個(ge)非(fei)線(xian)性的(de)(de)、時(shi)變的(de)(de)、有(you)(you)干擾(rao)的(de)(de)、具有(you)(you)純滯后(hou)的(de)(de)控制系統(tong)(tong)，在充(chong)(chong)(chong)放(fang)電(dian)(dian)(dian)過程中(zhong)涉及(ji)到很(hen)多參數(shu)，如充(chong)(chong)(chong)電(dian)(dian)(dian)率、最(zui)大允許充(chong)(chong)(chong)電(dian)(dian)(dian)電(dian)(dian)(dian)流、內(nei)阻、出氣點電(dian)(dian)(dian)壓、溫度、壽(shou)命等。

   作者創新點(dian)為：

(1) 隸屬函數(shu) 的形狀，對(dui)控(kong)制效果(guo)影響(xiang)較大。窄型隸屬函數(shu)，反(fan)映模(mo)糊集(ji)合具有(you)高分(fen)(fen)辨特性。如(ru)果(guo)系(xi)統誤(wu)差，采用(yong)高分(fen)(fen)辨率(lv)模(mo)糊集(ji)合，則誤(wu)差控(kong)制的靈敏(min)度就會提(ti)高。在系(xi)統誤(wu)差較大的范圍內(nei)，采用(yong)具有(you)低分(fen)(fen)辨率(lv)隸屬函數(shu)的模(mo)糊集(ji)合；而在系(xi)統誤(wu)差較小，或接近于零時，宜(yi)采用(yong)具有(you)高分(fen)(fen)辨率(lv)隸屬函數(shu)的模(mo)糊集(ji)合。

(2)在定義某一語言(yan)(yan)變量(liang)，如誤(wu)差、誤(wu)差變化(hua)率和控制量(liang)變化(hua)的(de)全(quan)部集合時，要考慮其對論(lun)域[-n,+n]的(de)覆蓋程度，語言(yan)(yan)變量(liang)的(de)全(quan)部模(mo)糊集合所包含(han)的(de)非零(ling)隸屬度對應(ying)的(de)論(lun)域元素個數，應(ying)是模(mo)糊集合總數的(de)3-4倍(bei)。

（3）查表法作為模糊控制算法有表格結構單一，修改繁瑣，缺乏靈活性的缺點。針對使用查表法作為模糊控制算法暴露的缺點，在硬件設計中與以補償，加入了一片X5045電可擦除芯片，將模糊控制表格中的數據存儲于微控制器外部存儲空間中，基本上克服了這個缺點。