1.前言

    蓄電(dian)池的充(chong)電(dian)控(kong)制(zhi)(zhi)系統是(shi)個非線性[1]的、時變的、有(you)干(gan)擾的、具有(you)純滯后的控(kong)制(zhi)(zhi)系統[2]。在充(chong)放電(dian)過程中涉(she)及到很多參數，如充(chong)電(dian)率、最大允(yun)許充(chong)電(dian)電(dian)流、內(nei)阻(zu)、出氣點電(dian)壓、溫度、壽命等。

   傳統(tong)的(de)(de)控(kong)(kong)制(zhi)系統(tong)是(shi)(shi)建(jian)立(li)在(zai)被控(kong)(kong)對象精(jing)確數(shu)(shu)學模型(xing)基礎上的(de)(de)，如果被控(kong)(kong)對象的(de)(de)數(shu)(shu)學模型(xing)很復雜或者數(shu)(shu)學模型(xing)無法建(jian)立(li)，控(kong)(kong)制(zhi)系統(tong)就(jiu)較難(nan)實現。蓄(xu)(xu)電(dian)(dian)池(chi)(chi)充(chong)電(dian)(dian)正是(shi)(shi)屬于這種情況，由于蓄(xu)(xu)電(dian)(dian)池(chi)(chi)的(de)(de)充(chong)電(dian)(dian)過(guo)程有自(zi)己獨特的(de)(de)物(wu)理化學規律，因(yin)此(ci)考慮(lv)采(cai)用模糊控(kong)(kong)制(zhi)技(ji)術(shu)來(lai)進行蓄(xu)(xu)電(dian)(dian)池(chi)(chi)的(de)(de)充(chong)電(dian)(dian)控(kong)(kong)制(zhi) [3]。

   模(mo)糊(hu)(hu)控(kong)(kong)(kong)(kong)制(zhi)[4]是以模(mo)糊(hu)(hu)集合(he)理論為基礎的(de)控(kong)(kong)(kong)(kong)制(zhi)手(shou)段，它是模(mo)糊(hu)(hu)系(xi)統理論、模(mo)糊(hu)(hu)技術(shu)與自動(dong)控(kong)(kong)(kong)(kong)制(zhi)技術(shu)相結合(he)的(de)產物(wu),出(chu)發點是操作人員的(de)控(kong)(kong)(kong)(kong)制(zhi)經驗或相關專(zhuan)家的(de)知(zhi)識，在設(she)(she)計中不需要(yao)建立被控(kong)(kong)(kong)(kong)對象(xiang)的(de)精確數(shu)學(xue)模(mo)型[5]，因而使得控(kong)(kong)(kong)(kong)制(zhi)機理和策略(lve)易(yi)于接受(shou)與理解，設(she)(she)計簡(jian)單，便于應(ying)用。

2.模糊控制器(qi)的結(jie)構及算法

   基于微控(kong)制器(qi)組成的(de)模(mo)(mo)糊控(kong)制器(qi)包括模(mo)(mo)糊化接口(kou)、決策邏輯、知識庫和(he)反模(mo)(mo)糊化接口(kou)四個部分(fen)，如圖1所示(shi)。

圖1模(mo)糊控制器的(de)結(jie)構框(kuang)圖

Fig.1 The structure diagram of fuzzy controller

   在進(jin)行模(mo)糊控(kong)(kong)(kong)制(zhi)(zhi)(zhi)(zhi)算(suan)法(fa)設(she)計時，須(xu)首先考慮智能(neng)充電(dian)(dian)系統的(de)(de)(de)技術要求，體現(xian)智能(neng)充電(dian)(dian)控(kong)(kong)(kong)制(zhi)(zhi)(zhi)(zhi)技術的(de)(de)(de)優勢，解(jie)決(jue)長期困擾蓄(xu)電(dian)(dian)池(chi)裝備中的(de)(de)(de)效率和壽命問題，所(suo)以應當提(ti)高(gao)智能(neng)充電(dian)(dian)控(kong)(kong)(kong)制(zhi)(zhi)(zhi)(zhi)中對(dui)(dui)控(kong)(kong)(kong)制(zhi)(zhi)(zhi)(zhi)量(liang)的(de)(de)(de)精度；其次，模(mo)糊控(kong)(kong)(kong)制(zhi)(zhi)(zhi)(zhi)算(suan)法(fa)必(bi)須(xu)高(gao)速(su)可靠，對(dui)(dui)外部檢測(ce)得到的(de)(de)(de)物理量(liang)要有非(fei)常(chang)快(kuai)的(de)(de)(de)反應速(su)度。采用(yong)新型(xing)大(da)存(cun)儲容(rong)量(liang)的(de)(de)(de)微控(kong)(kong)(kong)制(zhi)(zhi)(zhi)(zhi)器解(jie)決(jue)使用(yong)查(cha)表(biao)法(fa)所(suo)帶(dai)來的(de)(de)(de)需(xu)要大(da)容(rong)量(liang)內存(cun)的(de)(de)(de)問題。

3 模糊控制(zhi)器的設計

3.1雙輸(shu)(shu)入單輸(shu)(shu)出模糊控(kong)制(zhi)器(qi)

    常見的模(mo)糊控制器[6]根據輸(shu)(shu)(shu)(shu)入(ru)(ru)(ru)(ru)與輸(shu)(shu)(shu)(shu)出個數分為單輸(shu)(shu)(shu)(shu)入(ru)(ru)(ru)(ru)單輸(shu)(shu)(shu)(shu)出、雙輸(shu)(shu)(shu)(shu)入(ru)(ru)(ru)(ru)單輸(shu)(shu)(shu)(shu)出、多(duo)輸(shu)(shu)(shu)(shu)入(ru)(ru)(ru)(ru)單輸(shu)(shu)(shu)(shu)出和雙輸(shu)(shu)(shu)(shu)入(ru)(ru)(ru)(ru)多(duo)輸(shu)(shu)(shu)(shu)出等幾(ji)種(zhong)。平時應用(yong)較多(duo)的是雙輸(shu)(shu)(shu)(shu)入(ru)(ru)(ru)(ru)單輸(shu)(shu)(shu)(shu)出模(mo)糊控制器。

圖2 雙輸(shu)入單(dan)輸(shu)出模糊控(kong)制器(qi)方框圖

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

    圖2是雙輸入單(dan)輸出(chu)模(mo)(mo)(mo)糊(hu)(hu)控(kong)(kong)制(zhi)(zhi)(zhi)器(qi)的(de)方框圖。其(qi)中(zhong)屬(shu)于論(lun)(lun)域X的(de)模(mo)(mo)(mo)糊(hu)(hu)集(ji)(ji)(ji)合 取(qu)自(zi)系統誤(wu)差(cha)e的(de)模(mo)(mo)(mo)糊(hu)(hu)化(hua)(hua)(hua)，屬(shu)于論(lun)(lun)域Y的(de)模(mo)(mo)(mo)糊(hu)(hu)集(ji)(ji)(ji)合 取(qu)自(zi)系統誤(wu)差(cha)變化(hua)(hua)(hua)率 的(de)模(mo)(mo)(mo)糊(hu)(hu)化(hua)(hua)(hua)，二(er)者構(gou)成模(mo)(mo)(mo)糊(hu)(hu)控(kong)(kong)制(zhi)(zhi)(zhi)器(qi)的(de)二(er)維(wei)(wei)輸入；屬(shu)于論(lun)(lun)域Z的(de)模(mo)(mo)(mo)糊(hu)(hu)集(ji)(ji)(ji)合 是反(fan)映控(kong)(kong)制(zhi)(zhi)(zhi)量(liang)變化(hua)(hua)(hua)的(de)模(mo)(mo)(mo)糊(hu)(hu)控(kong)(kong)制(zhi)(zhi)(zhi)器(qi)的(de)一維(wei)(wei)輸出(chu)，模(mo)(mo)(mo)糊(hu)(hu)控(kong)(kong)制(zhi)(zhi)(zhi)器(qi)的(de)控(kong)(kong)制(zhi)(zhi)(zhi)規則[7]通常(chang)由模(mo)(mo)(mo)糊(hu)(hu)條件語句if    and    then  來表達。

3.2精確量(liang)的模糊化

   模(mo)糊(hu)控制(zhi)(zhi)系統(tong)中的(de)(de)被控對象(xiang)狀態(tai)變化是連續(xu)的(de)(de)，系統(tong)給(gei)定值也是連續(xu)的(de)(de),反映到模(mo)糊(hu)控制(zhi)(zhi)器輸入(ru)端的(de)(de)輸入(ru)量(liang)(liang)也必然是連續(xu)的(de)(de)。但模(mo)糊(hu)控制(zhi)(zhi)器由計算機構成，它只能執行離(li)散處(chu)理[7]。因(yin)此模(mo)糊(hu)控制(zhi)(zhi)器要求輸入(ru)量(liang)(liang)是離(li)散模(mo)糊(hu)量(liang)(liang)，即(ji)論(lun)域(yu)是離(li)散的(de)(de)。對連續(xu)論(lun)域(yu)要進行離(li)散化。連續(xu)論(lun)域(yu)[8]經(jing)過量(liang)(liang)化后就成為一個離(li)散論(lun)域(yu)[8]。

設有連(lian)續論(lun)域(yu)(yu)[a，b]，而量化(hua)之后的(de)離(li)散論(lun)域(yu)(yu)為(wei) ，亦即(ji)將(jiang)連(lian)續論(lun)域(yu)(yu)分為(wei)2n段，則(ze)存在(zai)系數

                                                                   （1）

   隨(sui)后，在(zai)求出(chu)每(mei)條規(gui)(gui)則(ze)(ze)(ze)的強度之后，把相(xiang)互(hu)矛(mao)盾的規(gui)(gui)則(ze)(ze)(ze)中強度較小(xiao)的舍去；把相(xiang)同規(gui)(gui)則(ze)(ze)(ze)合(he)成一條規(gui)(gui)則(ze)(ze)(ze)，得到最后控制規(gui)(gui)則(ze)(ze)(ze)基。

3.3.2根據學習(xi)算(suan)法生(sheng)成控制規則

   對被控(kong)(kong)對象執行手動控(kong)(kong)制(zhi)(zhi)所得到的(de)(de)控(kong)(kong)制(zhi)(zhi)規(gui)則(ze)是較粗糙的(de)(de)，有(you)時還(huan)可(ke)能(neng)會出(chu)現控(kong)(kong)制(zhi)(zhi)死區，一個控(kong)(kong)制(zhi)(zhi)規(gui)則(ze)表(biao)中會出(chu)現空項,這是不能(neng)滿足實際控(kong)(kong)制(zhi)(zhi)要求的(de)(de)。為了取得更滿意的(de)(de)控(kong)(kong)制(zhi)(zhi)效果，可(ke)以(yi)(yi)對原始(shi)的(de)(de)控(kong)(kong)制(zhi)(zhi)規(gui)則(ze)進行改進。這時，應以(yi)(yi)粗糙的(de)(de)控(kong)(kong)制(zhi)(zhi)規(gui)則(ze)為基(ji)礎，通過仿(fang)真實驗和(he)系統調(diao)試(shi)加以(yi)(yi)完善。

4.模(mo)糊智能(neng)充電(dian)系統的工(gong)作原理及結(jie)構

   智能充(chong)電(dian)(dian)(dian)系統(tong)主(zhu)要(yao)由充(chong)電(dian)(dian)(dian)電(dian)(dian)(dian)源和(he)單片(pian)機控(kong)制電(dian)(dian)(dian)路兩部(bu)分組成。220V的(de)(de)(de)(de)交流市電(dian)(dian)(dian)經(jing)整流濾波電(dian)(dian)(dian)路變(bian)為(wei)脈(mo)動(dong)的(de)(de)(de)(de)310V高(gao)壓(ya)直流。然后經(jing)DC-DC變(bian)換電(dian)(dian)(dian)路（脈(mo)沖功(gong)率變(bian)壓(ya)器）變(bian)為(wei)充(chong)電(dian)(dian)(dian)所(suo)需(xu)的(de)(de)(de)(de)60V直流電(dian)(dian)(dian)壓(ya)。為(wei)了(le)保證輸(shu)出電(dian)(dian)(dian)壓(ya)的(de)(de)(de)(de)穩定性，采(cai)(cai)用了(le)UC3842對60V直流電(dian)(dian)(dian)壓(ya)進行穩壓(ya)。二(er)次斬(zhan)(zhan)波電(dian)(dian)(dian)路主(zhu)要(yao)由MOSFET管、電(dian)(dian)(dian)感、電(dian)(dian)(dian)容(rong)和(he)二(er)極管組成，輸(shu)出24-36V的(de)(de)(de)(de)充(chong)電(dian)(dian)(dian)電(dian)(dian)(dian)壓(ya)。控(kong)制部(bu)分采(cai)(cai)用C504單片(pian)機，通(tong)過對蓄(xu)電(dian)(dian)(dian)池端(duan)電(dian)(dian)(dian)壓(ya)信號的(de)(de)(de)(de)采(cai)(cai)集(ji)(ji)、分析(xi)處理、模(mo)糊推理[8]、模(mo)糊決策(ce)等，控(kong)制二(er)次斬(zhan)(zhan)波電(dian)(dian)(dian)路中的(de)(de)(de)(de)MOSFET管的(de)(de)(de)(de)通(tong)斷(duan)時間來控(kong)制充(chong)電(dian)(dian)(dian)電(dian)(dian)(dian)壓(ya)。控(kong)制部(bu)分還包括(kuo)對電(dian)(dian)(dian)流和(he)溫度的(de)(de)(de)(de)采(cai)(cai)集(ji)(ji)以(yi)及電(dian)(dian)(dian)壓(ya)和(he)電(dian)(dian)(dian)流的(de)(de)(de)(de)顯示。總(zong)體結構如圖3所(suo)示。

圖3 智能充電系統總體結構框圖

5結論

  蓄(xu)電(dian)(dian)(dian)(dian)池(chi)(chi)的(de)(de)(de)(de)充(chong)(chong)放(fang)電(dian)(dian)(dian)(dian)過(guo)程(cheng)(cheng)是(shi)一個復(fu)雜的(de)(de)(de)(de)過(guo)程(cheng)(cheng)，要用精確數(shu)(shu)學(xue)模型對(dui)蓄(xu)電(dian)(dian)(dian)(dian)池(chi)(chi)充(chong)(chong)電(dian)(dian)(dian)(dian)的(de)(de)(de)(de)控制則(ze)有(you)相(xiang)當的(de)(de)(de)(de)難度。蓄(xu)電(dian)(dian)(dian)(dian)池(chi)(chi)的(de)(de)(de)(de)充(chong)(chong)電(dian)(dian)(dian)(dian)控制系(xi)統是(shi)個非線性的(de)(de)(de)(de)、時變的(de)(de)(de)(de)、有(you)干擾(rao)的(de)(de)(de)(de)、具有(you)純滯(zhi)后的(de)(de)(de)(de)控制系(xi)統，在充(chong)(chong)放(fang)電(dian)(dian)(dian)(dian)過(guo)程(cheng)(cheng)中涉及到很(hen)多參數(shu)(shu)，如充(chong)(chong)電(dian)(dian)(dian)(dian)率、最大(da)允(yun)許充(chong)(chong)電(dian)(dian)(dian)(dian)電(dian)(dian)(dian)(dian)流、內(nei)阻、出氣(qi)點電(dian)(dian)(dian)(dian)壓(ya)、溫(wen)度、壽命等。

   作者創新點為(wei)：

(1) 隸(li)屬函(han)數 的(de)形(xing)狀，對控制(zhi)效果影響較大(da)。窄(zhai)型隸(li)屬函(han)數，反(fan)映模(mo)糊集合(he)具有(you)高分辨(bian)特(te)性(xing)。如果系統誤(wu)差(cha)，采(cai)用高分辨(bian)率模(mo)糊集合(he)，則(ze)誤(wu)差(cha)控制(zhi)的(de)靈敏度就會提高。在系統誤(wu)差(cha)較大(da)的(de)范圍(wei)內，采(cai)用具有(you)低分辨(bian)率隸(li)屬函(han)數的(de)模(mo)糊集合(he)；而(er)在系統誤(wu)差(cha)較小，或接近于零時，宜采(cai)用具有(you)高分辨(bian)率隸(li)屬函(han)數的(de)模(mo)糊集合(he)。

(2)在定(ding)義某一(yi)語言變量(liang)，如誤差(cha)、誤差(cha)變化(hua)率和控制量(liang)變化(hua)的(de)(de)全部集合時，要考慮其對(dui)論(lun)域[-n,+n]的(de)(de)覆蓋程度，語言變量(liang)的(de)(de)全部模(mo)糊集合所(suo)包含(han)的(de)(de)非零(ling)隸屬度對(dui)應(ying)的(de)(de)論(lun)域元素個數(shu)，應(ying)是模(mo)糊集合總數(shu)的(de)(de)3-4倍。

（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。