程 兵, 于蘭峰, 吳永明, 符 康
(西南交通大學(xué) 機(jī)械工程學(xué)院, 四川 成都 610031)
基于響應(yīng)面法的地坑式架車(chē)機(jī)輕量化研究
程 兵, 于蘭峰, 吳永明, 符 康
(西南交通大學(xué) 機(jī)械工程學(xué)院, 四川 成都 610031)
傳統(tǒng)架車(chē)機(jī)設(shè)計(jì)方法往往趨于保守,為了充分發(fā)揮地坑式架車(chē)機(jī)材料的承載性能,將響應(yīng)面近似模型法與優(yōu)化技術(shù)相結(jié)合,提出一種以地坑式架車(chē)機(jī)結(jié)構(gòu)自重為優(yōu)化目標(biāo)的優(yōu)化方法.該方法基于多學(xué)科優(yōu)化軟件ISIGHT和ANSYS,通過(guò)最優(yōu)拉丁超立方試驗(yàn)設(shè)計(jì)方法分析設(shè)計(jì)變量對(duì)響應(yīng)的靈敏度,獲得最終設(shè)計(jì)變量,并建立響應(yīng)面近似模型,利用多島遺傳算法對(duì)近似模型進(jìn)行優(yōu)化.結(jié)果表明,該方法能極大提高優(yōu)化效率,同時(shí)使地坑式架車(chē)機(jī)結(jié)構(gòu)自重降低20.64%,輕量化效果顯著.研究結(jié)果可為地坑式架車(chē)機(jī)的設(shè)計(jì)提供理論指導(dǎo).
地坑式架車(chē)機(jī); 近似模型; 最優(yōu)拉丁超立方設(shè)計(jì); 多島遺傳算法
地坑式架車(chē)機(jī)作為動(dòng)車(chē)組三級(jí)修程中必備的大型關(guān)鍵設(shè)備之一,主要用于舉升動(dòng)車(chē)組、更換轉(zhuǎn)向架和對(duì)底部進(jìn)行維修[1-4].圖1為車(chē)體舉升單元結(jié)構(gòu)示意圖,其工作原理為:立柱底部與螺母支撐箱連接,通過(guò)電機(jī)帶動(dòng)絲杠,利用絲杠和螺母的螺旋傳動(dòng)帶動(dòng)立柱垂直運(yùn)動(dòng)從而舉升車(chē)體.
作為大型檢修設(shè)備,地坑式架車(chē)機(jī)整體結(jié)構(gòu)非常復(fù)雜,整個(gè)16編組地坑式架車(chē)機(jī)總重達(dá)1 000 t[5].當(dāng)前,國(guó)內(nèi)廠家對(duì)架車(chē)機(jī)結(jié)構(gòu)件的設(shè)計(jì)依賴(lài)于經(jīng)驗(yàn),且沒(méi)有統(tǒng)一的設(shè)計(jì)規(guī)范,校核計(jì)算主要參考《起重機(jī)設(shè)計(jì)規(guī)范》.為安全起見(jiàn),結(jié)構(gòu)強(qiáng)度等性能參數(shù)所留余量較大,因此有必要對(duì)各結(jié)構(gòu)件進(jìn)行優(yōu)化.
圖1 車(chē)體舉升單元結(jié)構(gòu)示意圖Fig.1 The structure diagram of lifting unit
由于結(jié)構(gòu)復(fù)雜,傳統(tǒng)有限元優(yōu)化設(shè)計(jì)計(jì)算成本太大,本文采用多學(xué)科設(shè)計(jì)優(yōu)化軟件ISIGHT,對(duì)參數(shù)進(jìn)行靈敏度分析,并將近似模型技術(shù)應(yīng)用于優(yōu)化設(shè)計(jì)中,探究出適用于地坑式架車(chē)機(jī)的結(jié)構(gòu)優(yōu)化方法.
1.1 設(shè)計(jì)變量
以某研究所研制的地坑式架車(chē)機(jī)為例,對(duì)圖1中的立柱、螺母支撐箱及導(dǎo)向箱等主要結(jié)構(gòu)進(jìn)行優(yōu)化設(shè)計(jì).該架車(chē)機(jī)最大舉升高度為2.7 m,額定舉升重量Q=1.7×105N[6],考慮偏載等情況,取載荷放大系數(shù)ψ=1.4,故計(jì)算載荷F=1.4×1.7×105N=2.38×105N.各部件材料均為Q345,材料許用應(yīng)力[σ]=275 MPa[7].為保證機(jī)構(gòu)各零部件與結(jié)構(gòu)件的連接位置不變,只對(duì)各結(jié)構(gòu)件的板厚進(jìn)行優(yōu)化,具體設(shè)計(jì)變量如表1所示.
表1 架車(chē)機(jī)設(shè)計(jì)變量
1.2 優(yōu)化目標(biāo)
為減輕結(jié)構(gòu)自重,達(dá)到輕量化設(shè)計(jì)的目的,以地坑式架車(chē)機(jī)結(jié)構(gòu)系統(tǒng)的質(zhì)量為優(yōu)化設(shè)計(jì)目標(biāo),即
min F(x)=T_MASS,
(1)
式中T_MASS為架車(chē)機(jī)結(jié)構(gòu)自重.
1.3 約束條件
1)靜強(qiáng)度約束為
g1(x)=SG≤[σ],
(2)
式中:SG為架車(chē)機(jī)在工作狀態(tài)下的最大等效應(yīng)力, [σ]=275 MPa.
2)靜剛度約束為
g2(x)=DG≤10 mm,
(3)
式中DG為架車(chē)機(jī)工作狀態(tài)下最大垂直靜撓度,根據(jù)設(shè)計(jì)要求,最大值不超過(guò)10 mm.
3)設(shè)計(jì)變量上下限約束為
xl≤xi≤xu,i=1,2,….
(4)
式中xu,xl為設(shè)計(jì)變量上、下限,具體見(jiàn)表1.
1.4 有限元模型
圖2 地坑式架車(chē)機(jī)有限元模型Fig.2 The finite element model of underfloor lifting system
對(duì)架車(chē)機(jī)結(jié)構(gòu)進(jìn)行部分簡(jiǎn)化,利用ANSYS中的APDL語(yǔ)言建立整機(jī)的有限元參數(shù)化模型,如圖2所示.其中,托頭、滑塊、滑軌等構(gòu)件以及較厚板采用SOLID45單元模擬,其余板結(jié)構(gòu)用SHELL63單元模擬,彈性模量Ex=2.1×1011N/m2,泊松比μ=0.3,材料密度ρ=7 850 kg/m3.
ISIGHT軟件是一款先進(jìn)的基于參數(shù)的多學(xué)科設(shè)計(jì)優(yōu)化軟件[8-9],其主要優(yōu)勢(shì)是將優(yōu)化方法、數(shù)值計(jì)算等技術(shù)有機(jī)結(jié)合,使設(shè)計(jì)流程集成化、自動(dòng)化、最優(yōu)化,從而大大提高優(yōu)化設(shè)計(jì)效率.
通過(guò)將基于ANSYS建立的參數(shù)化模型集成到ISIGHT中,利用試驗(yàn)設(shè)計(jì)方法(design of experiments,DOE)選出對(duì)約束條件和目標(biāo)函數(shù)影響較大的設(shè)計(jì)變量,并建立響應(yīng)面模型,以結(jié)構(gòu)自重為目標(biāo)函數(shù)建立優(yōu)化模型,并利用多島遺傳算法(multi-is land genetic algorithm,MIGA)對(duì)模型進(jìn)行優(yōu)化.架車(chē)機(jī)輕量化設(shè)計(jì)的具體流程如圖3所示.
圖3 架車(chē)機(jī)輕量化設(shè)計(jì)流程圖Fig.3 The flow chart of lightweight design for underfloor lifting system
2.1 DOE試驗(yàn)研究
由于設(shè)計(jì)變量較多,為了減少部分次要變量以提高優(yōu)化效率,采用ISIGHT軟件中的DOE模塊分析各設(shè)計(jì)變量對(duì)約束條件和目標(biāo)函數(shù)的靈敏度,然后利用最優(yōu)拉丁超立方試驗(yàn)設(shè)計(jì)方法來(lái)獲得最終設(shè)計(jì)變量.最優(yōu)拉丁超立方設(shè)計(jì)通過(guò)改進(jìn)隨機(jī)拉丁超立方設(shè)計(jì)的均勻性,使因子和響應(yīng)的擬合更加精確真實(shí).最優(yōu)拉丁超立方設(shè)計(jì)使試驗(yàn)點(diǎn)盡可能均勻地分布于設(shè)計(jì)空間,具有很好的空間填充性與均衡性.
最優(yōu)拉丁超立方試驗(yàn)設(shè)計(jì)法是在[0,1] 之間選取樣本點(diǎn),所以必須根據(jù)設(shè)計(jì)變量的取值范圍來(lái)確定樣本點(diǎn)的實(shí)際值, 其轉(zhuǎn)換關(guān)系如下:
(5)
式中:Pi為設(shè)計(jì)變量的實(shí)際樣本點(diǎn);Pmax和Pmin分別為取值范圍的最大值與最小值;Ri為最優(yōu)拉丁超立方采樣法選取的樣本點(diǎn)[10].
選擇最優(yōu)拉丁超立方設(shè)計(jì)法,選取150個(gè)樣本點(diǎn)進(jìn)行計(jì)算,DOE試驗(yàn)結(jié)果如圖4所示.
圖4 設(shè)計(jì)變量對(duì)結(jié)構(gòu)剛度、強(qiáng)度、自重的Pareto圖Fig.4 The Pareto diagram of design variables for the structural stiffness, strength and weight
2.2 響應(yīng)面近似模型構(gòu)造
近似模型技術(shù)是利用數(shù)學(xué)模型的方法逼近一組輸入變量與輸出變量的方法.通過(guò)近似模型方法,能極大地提高優(yōu)化算法的尋優(yōu)速度.由于響應(yīng)面法能利用較少的試驗(yàn)獲得比較精確的逼近函數(shù)關(guān)系,計(jì)算簡(jiǎn)單,并且擁有很好的魯棒性,本文將采用響應(yīng)面模型來(lái)構(gòu)造地坑式架車(chē)機(jī)的近似模型.
(6)
式中:N為模型的輸入變量個(gè)數(shù);xi為輸入變量;a,b,c為響應(yīng)面多項(xiàng)式的待定系數(shù).
對(duì)于待定系數(shù)的計(jì)算,是通過(guò)將待定系數(shù)組成一個(gè)列矩陣β,并通過(guò)最小二乘法得到系數(shù)矩陣[14-15]:
β=(XTX)-1(XTY),
(7)
式中,X為響應(yīng)面樣本點(diǎn)矢量,Y為樣本點(diǎn)對(duì)應(yīng)的響應(yīng)矢量,具體可表示為如下:
(8)
將式(8)代入式(7)中即可求出響應(yīng)面函數(shù)的系數(shù)矩陣β,從而得到具體的響應(yīng)面函數(shù)表達(dá)式.
選擇DG,SG和T_MASS為輸出響應(yīng),通過(guò)選取150個(gè)模型樣本點(diǎn)和20個(gè)誤差分析樣本點(diǎn),利用ISIGHT軟件中的誤差分析模塊對(duì)建立的響應(yīng)面模型進(jìn)行誤差分析,其結(jié)果如表2所示,各項(xiàng)誤差指標(biāo)均在允許范圍內(nèi),表明該近似模型可信度較高,可以代替仿真程序進(jìn)行優(yōu)化設(shè)計(jì).
表2 響應(yīng)面模型的誤差評(píng)估結(jié)果
2.3 優(yōu)化分析
傳統(tǒng)優(yōu)化算法經(jīng)常收斂于局部最優(yōu)解,導(dǎo)致尋優(yōu)不徹底,而ISIGHT軟件提供了全局優(yōu)化算法——多島遺傳算法(MIGA).該算法改進(jìn)了傳統(tǒng)遺傳算法進(jìn)行優(yōu)化設(shè)計(jì)時(shí)容易陷入早熟的缺陷.MIGA將進(jìn)化種群劃分成若干個(gè)子種群,在子種群中獨(dú)立地進(jìn)行遺傳算法的選擇、交叉、變異等操作,從而有效抑制早熟現(xiàn)象[16-17].作為一種偽并行遺傳算法,MIGA能夠更好地在優(yōu)化域中尋找全局最優(yōu)解,使優(yōu)化過(guò)程更高效和精確.因此,本文采用MIGA進(jìn)行全局尋優(yōu).
在實(shí)際生產(chǎn)中,由于鋼材厚度為離散整數(shù),因此,在優(yōu)化過(guò)程中對(duì)涉及板厚的設(shè)計(jì)參數(shù)進(jìn)行離散化處理[18].
在本文的優(yōu)化設(shè)計(jì)中,子群規(guī)模設(shè)定為10,子群數(shù)為10,總進(jìn)化代數(shù)為100,交叉概率Pc=0.8,變異概率Pm=0.08.目標(biāo)函數(shù)的迭代尋優(yōu)歷程如圖5所示,各設(shè)計(jì)變量、狀態(tài)變量及目標(biāo)函數(shù)優(yōu)化前后對(duì)比如表3所示.
圖5 目標(biāo)函數(shù)尋優(yōu)歷程Fig.5 Objective function optimization process
Table 3 Value comparison of design variables, state variables and objective function before and after optimization
參數(shù)初始值優(yōu)化值x1/mm4030x2/mm1613x5/mm3626x8/mm155x9/mm155x10/mm2010x11/mm155DG/mm4.454.99SG/MPa212.51244.32T_MASS/kg4534.293598.19
從優(yōu)化結(jié)果可以看出,各個(gè)設(shè)計(jì)變量的值均有所減小,而應(yīng)力和位移有所增大,但在允許范圍內(nèi),結(jié)構(gòu)自重明顯降低.
為驗(yàn)證基于響應(yīng)面近似模型的優(yōu)化設(shè)計(jì)結(jié)果的準(zhǔn)確性,將優(yōu)化后各設(shè)計(jì)變量的值代入有限元模型,利用ANSYS對(duì)其進(jìn)行有限元分析.仿真結(jié)果如圖6和圖7所示.
圖6 優(yōu)化后架車(chē)機(jī)等效應(yīng)力圖Fig.6 The Von Mises stress diagram of lifting system after optimization
圖7 優(yōu)化后架車(chē)機(jī)位移等值線圖Fig.7 The displacement contour diagram of lifting system after optimization
結(jié)果表明,優(yōu)化后架車(chē)機(jī)最大等效應(yīng)力為256.48 MPa,垂直方向最大位移為4.899 mm,均小于許用值,與響應(yīng)面近似模型優(yōu)化結(jié)果(表3)相比,誤差分別為4.7%,1.8%.優(yōu)化后架車(chē)機(jī)結(jié)構(gòu)自重為3 598.19 kg,相較于原自重減少了20.64%,優(yōu)化效果顯著.
1)利用最優(yōu)拉丁超立方設(shè)計(jì)法對(duì)設(shè)計(jì)變量進(jìn)行靈敏度分析,通過(guò)減少設(shè)計(jì)變量的數(shù)量,有效提高優(yōu)化效率.
2)將響應(yīng)面近似模型法引入地坑式架車(chē)機(jī)結(jié)構(gòu)輕量化設(shè)計(jì)中,大大縮減計(jì)算時(shí)間,顯著提高優(yōu)化設(shè)計(jì)效率.
3)通過(guò)多學(xué)科優(yōu)化軟件ISIGHT和ANSYS的集成,利用多島遺傳算法對(duì)地坑式架車(chē)機(jī)進(jìn)行優(yōu)化,使結(jié)構(gòu)自重減少20.64%,驗(yàn)證了此方法的可行性,為地坑式架車(chē)機(jī)的結(jié)構(gòu)設(shè)計(jì)和改進(jìn)提供參考依據(jù),具有較強(qiáng)的理論與實(shí)際意義.
[1] 丁輝,王明海. CHR動(dòng)車(chē)組地坑式架車(chē)機(jī)的研究與設(shè)計(jì)[J]. 電力機(jī)車(chē)與城軌車(chē)輛,2011,34(2):19-21. DING Hui, WANG Ming-hai. Research and design of CRH EMU underfloor lifting system [J]. Electric Locomotives and Urban Rail Vehicle,2011,34(2): 19-21.
[2] 喻貴忠,劉廣丹,呂安庶,等. 國(guó)產(chǎn)動(dòng)車(chē)組新穎架車(chē)機(jī)[J]. 機(jī)車(chē)電傳動(dòng),2010(3):37-41. YU Gui-zhong, LIU Guang-dan, Lü An-shu, et al. Domestic EMU new lifting system[J]. Electric Drive for Locomotives,2010(3): 37-41.
[3] 繆東. 固定式架車(chē)機(jī)在地鐵車(chē)輛段中應(yīng)用實(shí)踐[J]. 鐵道工程學(xué)報(bào),2008(10):92-95. MIAO Dong. The application and practice of stationary frame machine in metro car depot[J]. Journal of Railway Engineering Society, 2008(10): 92-96.
[4] 黎英豪,劉廣丹,喻貴忠. 高速動(dòng)車(chē)組的新穎地坑式架車(chē)機(jī)[J].鐵道機(jī)車(chē)車(chē)輛,2010,30(4): 72-76. LI Ying-hao, LIU Guang-dan, YU Gui-zhong.High speed CRH novel pit cars transferring machine[J]. Railway Locomotive & Car, 2010, 30(4): 72-76.
[5] 張樹(shù)申,劉廣丹,喻貴忠,等. 軌道交通車(chē)組檢修用新穎地坑式架車(chē)機(jī)[J]. 現(xiàn)代城市軌道交通,2010(5):5-8. ZHANG Shu-shen, LIU Guang-dan, YU Gui-zhong, et al. Novel underfloor lifting system for rail traffic vehicle maintenance[J]. Modern Urban Transit, 2010(5):5-8.
[6] 丁輝,邢曉東. 整列地坑式架車(chē)機(jī)在我國(guó)高速列車(chē)檢修中的運(yùn)用[J]. 機(jī)車(chē)電傳動(dòng),2011 (2):31-33. DING Hui, XING Xiao-dong. The entire column underfloor lifting system in the maintenance of high-speed trains in China[J]. Electric Drive for Locomotives, 2011 (2): 31-33.
[7] 王金諾,于蘭峰. 起重運(yùn)輸機(jī)械金屬結(jié)構(gòu)[M]. 北京:中國(guó)鐵道出版社,2002:43-47. WANG Jin-nuo, YU Lan-feng. Hoisting and conveying machinery metal structure [M]. Beijing: China Railway Press, 2002:43-47.
[8] 賴(lài)宇陽(yáng). Isight 參數(shù)優(yōu)化理論與實(shí)例詳解[M].北京:北京航空航天大學(xué)出版社,2012:3-7. LAI Yu-yang. Parameter optimization theory and examples of Isight [M]. Beijing: Beihang University Press, 2012:3-7.
[9] 聶勇軍,廖啟征. 基于ISIGHT的桁架結(jié)構(gòu)優(yōu)化設(shè)計(jì)[J]. 煤礦機(jī)械,2011,32(2):32-34. NIE Yong-jun, LIAO Qi-zheng. Optimization design of truss structure based on ISIGHT[J]. Coal Mine Machinery, 2011, 32(2):32-34.
[10] 文桂林,崔中,彭克立. 基于近似模型的高速磨床零部件結(jié)構(gòu)優(yōu)化設(shè)計(jì)研究[J]. 中國(guó)機(jī)械工程,2009, 20(8):906-910. WEN Gui-lin, CUI Zhong, PENG Ke-li. Research on structure optimization of high speed grinding machine based on approximate model[J]. China Mechanical Engineering, 2009, 20(8): 906-910.
[11] 王志龍. 桁架式橋梁檢測(cè)車(chē)臂架系統(tǒng)結(jié)構(gòu)分析及優(yōu)化設(shè)計(jì)[D]. 成都:西南交通大學(xué)機(jī)械工程學(xué)院,2011:53-54. WANG Zhi-long. Structural analysis and optimum design of truss structure bridge inspecting vehicle[D]. Chengdu: Southwest Jiaotong University, College of Mechanical Engineering, 2011:53-54.
[12] 里超. 連續(xù)卸船機(jī)結(jié)構(gòu)系統(tǒng)可靠性分析及優(yōu)化設(shè)計(jì)[D]. 成都:西南交通大學(xué)機(jī)械工程學(xué)院,2011:12-13. LI Chao. Reliability analysis and optimum design of continuous ship unloading machine[D].Chengdu: Southwest Jiaotong University, College of Mechanical Engineering,2011:12-13.
[13] 唐輝,于蘭峰,嚴(yán)飛,等. L型門(mén)式起重機(jī)主梁非概率可靠性分析[J]. 礦山機(jī)械, 2015, 43(11):9-62. TANG Hui, YU Lan-feng, YAN Fei, et al. Non-probability reliability analysis on girder of L-typed gantry crane [J].Mining & Processing Equipment,2015, 43(11):59-62.
[14] 呂輝,于德介,謝展,等. 基于響應(yīng)面法的汽車(chē)盤(pán)式制動(dòng)器穩(wěn)定性?xún)?yōu)化設(shè)計(jì)[J].機(jī)械工程學(xué)報(bào),2013,49 (9):55-60. Lü Hui, YU De-jie, XIE Zhan, et al. Optimization design of vehicle disc brake stability based on response surface methodology[J]. Journal of Mechanical Engineering, 2013, 49 (9): 55-60.
[15] 廖興濤,張維剛,李青,等. 響應(yīng)表面法在薄壁構(gòu)件耐撞性?xún)?yōu)化設(shè)計(jì)中的應(yīng)用研究[J].工程設(shè)計(jì)學(xué)報(bào),2006,13(5):298-302. LIAO Xing-tao,ZHANG Wei-gang, LI Qing, et al. Application of response surface method in optimization design of thin-walled component[J]. Chinese Journal of Engineering Design, 2006, 13(5):298-302.
[16] 張俊紅,朱傳峰,畢鳳榮,等. 基于DOE和MIGA的消聲器優(yōu)化設(shè)計(jì)[J]. 機(jī)械科學(xué)與技術(shù), 2016, 35(2):296-302. ZHANG Jun-hong, ZHU Chuan-feng, BI Feng-rong, et al. Optimization design of muffler based on DOE and MIGA[J]. Mechanical Science and Technology for Aerospace Engineering, 2016, 35(2):296-302.
[17] 趙志國(guó),司傳勝. 基于多島遺傳算法的鉸接車(chē)輪邊減速器優(yōu)化設(shè)計(jì)[J]. 機(jī)械設(shè)計(jì)與制造, 2010(12):213-215. ZHAO Zhi-guo, SI Chuan-sheng. Optimization design of articulated wheel side reducer based on multi Island genetic algorithm[J]. Mechanical Design and Manufacturing, 2010(12):213-215.
[18] 朱小龍,于蘭峰,郭俊才,等. 基于ANSYS和ISIGHT的橋式起重機(jī)主梁結(jié)構(gòu)輕量化設(shè)計(jì)[J]. 礦山機(jī)械,2012,40 (7):52-55. ZHU Xiao-long, YU Lan-feng, GUO Jun-cai, et al. Lightweight design on girder of bridge crane based on ANSYS and ISIGHT[J]. Mining & Processing Equipment, 2012, 40(7):52-55.
Research on lightweight design of underfloor lifting system based on response surface method
CHENG Bing, YU Lan-feng, WU Yong-ming, FU Kang
(School of Mechanical Engineering, Southwest Jiaotong University, Chengdu 610031, China)
Traditional design methods of underfloor lifting system tend to be conservative. In order to take full advantages of the bearing capacity of its material, a structure weight optimization method was proposed and combined with the response surface approximate model method and optimization technique. The method was based on multidisciplinary optimization software ISIGHT and ANSYS. With the analysis of design variables on the sensitivity of the response by optimal Latin hypercube experimental design method, and the final design variables were obtained. Then, the approximate model was optimized by using the multi-island genetic algorithm after the response surface model was established. Results showed that the method could greatly improve the efficiency of optimization, and the structure weight of underfloor lifting system reduced 20.64%,and the lightweight effect was obvious. The research results can provide theoretical guidance for the design of underfloor lifting system.
underfloor lifting system; approximate model; optimal Latin hypercube design; multi-island genetic algorithm
2016-05-10.
本刊網(wǎng)址·在線期刊:http://www.zjujournals.com/gcsjxb
四川省國(guó)際合作研究計(jì)劃(2014HH0022).
程兵(1990—),男,湖北黃岡人,碩士生,從事結(jié)構(gòu)設(shè)計(jì)及優(yōu)化研究,E-mail:swjtu_chengbing@163.com. http://orcid.org//0000-0003-2391-1256 通信聯(lián)系人:于蘭峰(1964—),女,山東濰坊人,教授,博士,從事現(xiàn)代設(shè)計(jì)理論及方法研究,E-mail:jdlf2000@sina.com.
10.3785/j.issn. 1006-754X.2016.06.013
TH 21
A
1006-754X(2016)06-0606-06