傅玉青
(華僑大學(xué) 工學(xué)院, 福建 泉州 362021)
采用遺傳算法的可見(jiàn)光通信LED陣列分布優(yōu)化
傅玉青
(華僑大學(xué) 工學(xué)院, 福建 泉州 362021)
采用遺傳算法快捷、準(zhǔn)確地優(yōu)化發(fā)光二極管(LED)陣列結(jié)構(gòu),使其在接收器上(目標(biāo)平面)的光照分布均勻.推導(dǎo)LED 陣列的照度分布函數(shù),以光照函數(shù)標(biāo)準(zhǔn)差與平均差的比值構(gòu)建一個(gè)適應(yīng)度函數(shù),以此衡量目標(biāo)平面的光照分布均勻度.獲得光照分布最優(yōu)時(shí)的LED 陣列結(jié)構(gòu)數(shù)據(jù)后,應(yīng)用MATLAB對(duì)幾何模型進(jìn)行仿真.仿真結(jié)果表明:目標(biāo)平面的照度均勻度達(dá)到97.4%,驗(yàn)證了所提算法的有效性與可行性.
發(fā)光二極管; 陣列; 遺傳算法; 光照度; 可見(jiàn)光通信.
可見(jiàn)光通信(VLC)是近幾年迅速發(fā)展起來(lái)的一種新型無(wú)線(xiàn)通信方式[1-5].VLC通過(guò)在發(fā)光二極管(LED)燈中植入芯片,讓LED燈變成無(wú)線(xiàn)訪(fǎng)問(wèn)接入點(diǎn)(AP),利用LED發(fā)出的肉眼看不到的高速明暗閃爍信號(hào)傳輸信息,達(dá)到通信的目的.在使用LED的VLC系統(tǒng)中,為了在整個(gè)室內(nèi)區(qū)域獲得一致的通信效果和燈光照明,防止盲區(qū)的出現(xiàn),要求LED光源發(fā)出的光照度均勻分布.然而,單個(gè)LED的照度分布近似于朗伯分布,無(wú)法滿(mǎn)足均勻分布的要求.二元光學(xué)可以改變單個(gè)LED輸出的照度分布[6-7],但單個(gè)LED輸出功率有限,將影響其傳輸質(zhì)量和傳輸距離.因此,有必要將多個(gè)LED燈按一定方式排列,以獲得均勻的照度分布[8-9].相關(guān)研究人員采用不同的研究模型對(duì)LED的圓形分布、方形分布、環(huán)形分布、菱形分布、球面分布等進(jìn)行了分析[10-14].但這些傳統(tǒng)分析方法都是基于理論計(jì)算的方法,運(yùn)算過(guò)程復(fù)雜且繁瑣.遺傳算法(GA)是一種借鑒生物界自然選擇思想和自然遺傳機(jī)制的全局隨機(jī)搜索算法,是一種自適應(yīng)的迭代尋優(yōu)過(guò)程,可用于對(duì)多目標(biāo)、非線(xiàn)性等問(wèn)題的優(yōu)化[15-20].遺傳算法隨機(jī)初始化一個(gè)種群,以該初始種群作為初始迭代點(diǎn),通過(guò)隨機(jī)選擇、交叉和變異的遺傳操作,產(chǎn)生一群新的更適應(yīng)環(huán)境的個(gè)體,使群體進(jìn)化到搜索空間中越來(lái)越好的區(qū)域.本文采用遺傳算法優(yōu)化LED陣列分布,以獲得較均勻的照度分布.
(a) LED照明系統(tǒng) (b) 平面T
LED照明系統(tǒng)及目標(biāo)平面T,如圖1所示.LED燈隨機(jī)分布在平面S上,目標(biāo)平面T呈M×N格分布.兩平面之間的垂直距離為z.每個(gè)LED燈的照度分布可近似為朗伯分布.
設(shè)平面S上LED燈坐標(biāo)(x,y,0)到達(dá)平面T上點(diǎn)A坐標(biāo)(Xp,Yq,z)的照度[12-15]為
E(Xp,Yq,z)=I0zm+1[(Xp-x)2+(Yq-y)2+z2]-(m+3)/2.
式(1)中:I0是視角為0°方向的光強(qiáng)度;m是光源的輻射模式, 與半光強(qiáng)度角θ1/2(發(fā)光強(qiáng)度下降到軸向強(qiáng)度值一半時(shí),發(fā)光方向與光軸之間的夾角,通常由生產(chǎn)商提供)有關(guān),其計(jì)算公式[1,10-15]為
則所有LED燈到達(dá)點(diǎn)A的總光照度為
式(3)中:(xi,yi,0)為第i個(gè)LED燈在平面S的坐標(biāo);n為L(zhǎng)ED燈的總數(shù).
為了優(yōu)化LED陣列分布,取CV(RMSE)為適應(yīng)度函數(shù),描述目標(biāo)平面照度的均勻性.CV(RMSE)為均方根誤差的變化系數(shù)[14],即
σ是目標(biāo)平面光照度的標(biāo)準(zhǔn)差,可表示為
式(4)的適應(yīng)度函數(shù)是各LED燈的坐標(biāo)函數(shù).為了得到均勻的照度分布,要求取適應(yīng)度函數(shù)的最小值,即通過(guò)遺傳算法優(yōu)化LED分布,讓目標(biāo)平面的光照度最均勻.
在遺傳算法中,模擬自然選擇的過(guò)程主要通過(guò)適應(yīng)度函數(shù)實(shí)現(xiàn).個(gè)體適應(yīng)度的大小決定其繼續(xù)繁衍或消亡,適應(yīng)度高的個(gè)體被復(fù)制到下一代的可能性高于適應(yīng)度低的個(gè)體.采用式(4)作為適應(yīng)度函數(shù),以各LED燈在平面S的坐標(biāo)為函數(shù).
采用遺傳算法優(yōu)化LED分布,主要有以下3個(gè)步驟[20-21].
步驟1 初始化群體集.設(shè)置迭代次數(shù)、交叉概率、變異概率;在初始范圍內(nèi),對(duì)種群進(jìn)行隨機(jī)初始化;構(gòu)造適應(yīng)度函數(shù)(式(4)).
步驟2 循環(huán).對(duì)種群進(jìn)行選擇、交叉、變異的遺傳操作;計(jì)算適應(yīng)度大小,評(píng)價(jià)個(gè)體的優(yōu)劣;迭代次數(shù)增加1次.
步驟3 當(dāng)適應(yīng)度函數(shù)達(dá)到滿(mǎn)意值或達(dá)到最大迭代次數(shù)時(shí),停止算法,輸出優(yōu)化后的LED坐標(biāo);否則,返回步驟2.
遺傳算法的流程圖,如圖2所示.
圖2 遺傳算法的流程圖
應(yīng)用遺傳算法對(duì)LED陣列分布進(jìn)行優(yōu)化,平面T為30 cm×30 cm;平面S為20 cm×20 cm;平面T與平面S的距離為3 m;垂直方向的發(fā)光強(qiáng)度I0為32.69 cd;半功率角θ1/2為54°;陣列中LED數(shù)為54;假設(shè)所有LED燈的發(fā)射角度和軸向光照度都相同.
圖3 優(yōu)化后的LED陣列分布
遺傳算法初始化參數(shù)和截止條件:迭代數(shù)為150;種群大小為25;交叉率為0.9;變異率為0.001;收斂條件 (FFV)為0.01.第51代時(shí),F(xiàn)FV=0.008,達(dá)到了算法運(yùn)行停止條件,這個(gè)數(shù)值小于預(yù)先設(shè)定的值.優(yōu)化后的LED陣列分布,如圖3所示.
利用MATLAB對(duì)LED陣列分布進(jìn)行仿真,歸一化光照度的三維及二維分布,如圖4所示.圖4中:I為歸一化的光照度.由圖4可知:應(yīng)用遺傳算法對(duì)LED陣列分布優(yōu)化后,光照度均勻性非常好,均勻度達(dá)到97.4%.這個(gè)值與優(yōu)化結(jié)果FFV量級(jí)非常吻合,說(shuō)明數(shù)值優(yōu)化是可行的.
(a) 三維分布 (b) 二維分布
構(gòu)造一個(gè)以所有LED燈坐標(biāo)為函數(shù)的適應(yīng)度函數(shù),表征LED光源在目標(biāo)平面的光照度均勻程度.利用遺傳算法對(duì)LED陣列分布進(jìn)行優(yōu)化,將優(yōu)化結(jié)果進(jìn)行仿真.仿真結(jié)果表明:目標(biāo)平面的照度均勻度達(dá)到97.4%,可以滿(mǎn)足可見(jiàn)光通信和照明要求.采用這種數(shù)值優(yōu)化算法可借助計(jì)算機(jī)實(shí)現(xiàn)自動(dòng)優(yōu)化設(shè)計(jì),并可應(yīng)用于不同光照度的LED燈的優(yōu)化.
[1] TANAKA Y,HARUYAMA S,NAKGAWA N.Wireless optical transmissions with white colored LED for wireless home links[C]∥Proceedings of the 11th IEEE International Symposium on Personal, Indoor and Mobile Radio Communications.London:IEEE Press,2000:1325-1329.
[2] MINH H L,O′BRIEN D, FAULKNER G,et al. High-speed visible light communications using multiple-resonant equalization[J].IEEE Photon Technol Lett,2008,20(14):1243-1245.
[3] KWOHYUNG L,HYUNCHEOL P,JOHN R B.Indoor channel characteristics for visible light communications[J].IEEE Comm Lett,2011,15(2):217-219.
[4] CHIEN H Y,YEN L L,CHI W C.Realtime white light phosphor-LED visible light communications (VLC) with compact size[J].Opt Express,2013,21(22):26192-26197.
[5] KOMINE T,LEE J H,HARUYAMA S,et al.Adaptive equalization for indoor visible-light wireless communication systems communications[C]∥Asia-Pacific Conference on IEEE.Perth:IEEE Press,2005:294-298.
[6] DING Yi,LIU Xu,ZHENG Zhenrong,et al.Freeform LED lens uniform illuminations[J].Opt Express,2008,16(17):12958-12966.
[7] KIM B,KIM H,KANG S.Reverse functional design of discontinuous refractive optics using an extended light source for flat illuminance distributions and high color uniformity[J].Opt Express,2011,19(3):1794-1807.
[8] 劉浩杰,藍(lán)天,倪國(guó)強(qiáng).室內(nèi)可見(jiàn)光通信發(fā)光二極管陣列發(fā)射性能的研究[J].物理學(xué)報(bào),2014,63(23):238503(1-8).
[9] 趙芝璞,季凌燕,沈艷霞,等.基于PSO粒子群算法的LED照明系統(tǒng)光照均勻性研究[J].發(fā)光學(xué)報(bào),2013,34(12):1677-1682.
[10] MORENO I,AVENDAO-ALEJO M,TZONCHEV R I.Designing light-emitting diode arrays for uniform nearfield irradiance[J].Appl Opt,2006,45(10):2265-2272.
[11] MORENO I,MUOZ J,IVANOV R.Uniform illumination of distant targets using a spherical light-emitting diode array[J].Opt Eng,2007,46(3):033001(1-7).
[12] YANG Hongming,BERGMANS J W M,SCHENK T C W,et al.Uniform illumination rendering using an array of LEDs: A signal processing perspective[J].IEEE Signal Proces,2009,57(3):1044-1057.
[13] WHANG A J W,CHEN Y Y,TENG Y T.Designing uniform illuminance systems by surface-tailored lens and configurations of LED arrays[J].J Disp Technol,2009,5(3):94-103.
[14] QIN Zong,WANG Kai,CHEN Fei,et al.Analysis of condition for uniform lighting generated by array of light emitting diodes with large view angle[J].Opt Express,2010,18(16):17460-17476.
[15] SU Zhouping,XUE Donglin,JI Zhicheng.Designing LED array for uniform illumination distribution by simulated annealing algorithm[J].Opt Express,2012,20(S6):A843-A855.
[16] WU N H,CHAN K C.A genetic algorithm based approach to optimal fixture configuration comput[J].Ind Eng,1996,31(3/4):919-924.
[17] KRISHNAKUMAR K,MELKOTE S N.Machining fixture layout optimization using the genetic algorithm[J].Int J Mach Tool Manu,1999,40(4):579-598.
[18] KULANKARA K,SATYANARAYANA S,MELKOTE S N.Iterative fixture layout and clamping force optimization using the genetic algorithm[J].J Manufacturing Science and Engin,2002,124(1):119-125.
[19] KAYA N.Machining fixture locating and clamping position optimization using genetic algorithms[J].Comput Ind,2006,57(2):112-120.
[20] HAMEDI M.Intelligent fixture design through a hybrid system of artificial neural network and genetic algorithm[J].Artif Intell Rev,2005,23(3):295-311.
[21] RAJENDRAN I,VIJAYARANGAN S.Optimal design of a composite leaf spring using genetic algorithms[J].Comput Struct,2001,79(11):1121-1129.
(責(zé)任編輯: 錢(qián)筠 英文審校: 吳逢鐵)
Research on LED Array for Illumination Uniformity in Visible Light Communication Based on Genetic Algorithm
FU Yuqing
(College of Engineering, Huaqiao University, Quanzhou 362021, China)
For obtaining uniform distribution of illumination on the receiver plane (target), the genetic algorithm is used to accurately and quickly optimize the light emitting diode (LED) array. LED array illumination distribution function is derived. A fitness function with the ratio of standard deviation and mean difference of the illumination function is built. The optimal distribution of LED array by applying the light distribution uniformity on the target plane is then achieved. Finally, the geometric model is simulated by using MATLAB. The simulation results show that the illumination uniformity on the target plane reached 97.4% and indicate that the proposed algorithm is feasibility and effectiveness.
light emitting diode; array; genetic algorithm; illumination distribution; visible light communication
1000-5013(2015)06-0632-04
10.11830/ISSN.1000-5013.2015.06.0632
2015-01-15
傅玉青(1984-),女,講師,博士,主要從事光通信技術(shù)、激光技術(shù)、可見(jiàn)光通信的研究.E-mail:fuyq@hqu.edu.cn.
福建省科技創(chuàng)新平臺(tái)資助項(xiàng)目(2013H2002)
TN 401
A