李小魁 黃全振 黃明明
摘 要: 為了解決傳統(tǒng)PID參數(shù)優(yōu)化方法易出現(xiàn)費(fèi)時(shí)、震蕩且不能保證所調(diào)參數(shù)最優(yōu)的問題,提出一種基于蜻蜓算法的PID控制參數(shù)優(yōu)化方法,該方法利用蜻蜓群體尋找食物的過程并以誤差性能指標(biāo)ITAE作為其適應(yīng)度函數(shù)實(shí)現(xiàn)PID控制參數(shù)的優(yōu)化。通過仿真實(shí)驗(yàn),并與粒子群優(yōu)化算法、人工蜂群算法、布谷鳥搜索算法等常用的PID參數(shù)整定方法進(jìn)行比較,結(jié)果表明,基于蜻蜓算法優(yōu)化的PID控制器具有更優(yōu)的控制性能。
關(guān)鍵詞: 蜻蜓算法; PID控制器; 參數(shù)優(yōu)化; ITAE; 控制性能; 反饋控制策略
中圖分類號(hào): TN876?34; TP18 文獻(xiàn)標(biāo)識(shí)碼: A 文章編號(hào): 1004?373X(2018)12?0102?06
Abstract: To solve the problems of time?consuming, vibration, and inability to guarantee the adjusted parameters optimal in the traditional PID parameter optimization method, a PID control parameter optimization method based on dragonfly algorithm is proposed. In the method, the PID control parameters are optimized by using the process of dragonfly population searching for food and taking the error performance indicator ITAE as the fitness function. The simulation experiment results show that, in comparison with the common PID parameter tuning methods such as particle swarm optimization algorithm, artificial bee colony algorithm, and cuckoo search algorithm, the PID controller based on dragonfly algorithm optimization has a better control performance.
Keywords: dragonfly algorithm; PID controller; parameter optimization; ITAE; control performance; feedback control strategy
0 引 言
PID控制器是最早提出的反饋控制器之一[1],在工業(yè)控制領(lǐng)域,有超過95%的反饋回路部件使用的是PID控制器,然而由于PID控制本質(zhì)是一種線性控制規(guī)律,在實(shí)際工業(yè)應(yīng)用中對(duì)于一些非線性、高階、時(shí)滯的復(fù)雜系統(tǒng),常規(guī)的PID控制效果并不能滿足生產(chǎn)要求。因此,PID控制器的參數(shù)優(yōu)化一直是控制理論研究的一個(gè)重要課題。
蜻蜓算法(Dragonfly Algorithm,DA)[2]源于自然界中蜻蜓捕食、遷徙和躲避外敵的群體行為,實(shí)現(xiàn)對(duì)目標(biāo)函數(shù)的優(yōu)化。該算法不僅具有粒子群算法的個(gè)人認(rèn)知和社會(huì)認(rèn)知能力,同時(shí)結(jié)合了布谷鳥算法[3]中[Le′vy]飛行行為,在算法尋優(yōu)過程中能夠有效避免陷入局部最優(yōu),提高算法的搜索性能。本文將蜻蜓算法用于PID控制器參數(shù)優(yōu)化,基本思路是將PID控制器待優(yōu)化的參數(shù)作為蜻蜓個(gè)體的位置信息,并以誤差性能指標(biāo)ITAE作為其適應(yīng)度函數(shù),利用蜻蜓群體尋找食物的過程實(shí)現(xiàn)對(duì)PID控制參數(shù)的優(yōu)化。并選取7類工業(yè)控制中的典型控制模型,通過Matlab對(duì)各系統(tǒng)進(jìn)行仿真實(shí)驗(yàn),并與粒子群算法[4?6]、人工蜂群算法、布谷鳥搜索算法等幾種常用的PID參數(shù)優(yōu)化方法進(jìn)行比較。實(shí)驗(yàn)結(jié)果表明,利用蜻蜓算法優(yōu)化的PID控制器對(duì)各控制模型均具有很好的控制性能。
1 PID控制器
PID控制器的規(guī)律為:
4.3.2 20次獨(dú)立運(yùn)行整定結(jié)果對(duì)比
考慮到智能優(yōu)化算法的隨機(jī)性,在上述相同的環(huán)境下,各算法針對(duì)7類不同控制模型分別進(jìn)行20次獨(dú)立仿真實(shí)驗(yàn),從整定得到的ITAE性能指標(biāo)結(jié)果的最優(yōu)值、最差值、平均值和方差進(jìn)行不同的算法分析,結(jié)果見表3。從表3可看出,對(duì)于[G1(s)],[G3(s)]和[G5(s)]被控對(duì)象,蜻蜓算法在最優(yōu)值、最差值、平均值和方差的仿真結(jié)果均優(yōu)于其他4種算法;對(duì)于[G2(s)],[G4(s)]和[G6(s)]這三個(gè)被控對(duì)象,蜻蜓算法的最優(yōu)值、最差值、平均值均優(yōu)于其他對(duì)比算法;而對(duì)于[G7(s)]非最小相位系統(tǒng),蜻蜓算法的最優(yōu)值略差于粒子群算法,但其在最差值、平均值和方差方面均優(yōu)于粒子群算法和其他對(duì)比算法。綜合表2和表3中數(shù)據(jù)及圖1~圖7中性能指標(biāo)進(jìn)化曲線的對(duì)比分析,從性能指標(biāo)、響應(yīng)時(shí)間及適用的廣泛性等角度來考慮,對(duì)于7類不同典型被控系統(tǒng),本文算法明顯優(yōu)于其他4種算法。
5 結(jié) 語
針對(duì)傳統(tǒng)PID整定方法過程中可移植性差、費(fèi)時(shí)且不能保證最佳控制性能的不足,提出一種基于蜻蜓算法優(yōu)化的PID控制器。利用蜻蜓算法的尋優(yōu)能力強(qiáng)的特點(diǎn),將蜻蜓的位置定義為PID控制器的三個(gè)參數(shù),將蜻蜓群體尋找獵物的過程視為實(shí)現(xiàn)PID控制參數(shù)的優(yōu)化。在Simulink環(huán)境下,PID控制系統(tǒng)模型的每組控制參數(shù)對(duì)應(yīng)的ITAE性能指標(biāo)值得到最優(yōu)的控制參數(shù)對(duì)被控系統(tǒng)進(jìn)行控制,具有較強(qiáng)的自適應(yīng)性和魯棒性。本文選取工業(yè)控制中常見的7類被控系統(tǒng)進(jìn)行仿真實(shí)驗(yàn),實(shí)驗(yàn)結(jié)果表明,基于蜻蜓算法的PID控制器具有更好的控制性能和魯棒性,因此,在工業(yè)控制中有較好的應(yīng)用前景。
注:本文通訊作者為黃全振。
參考文獻(xiàn)
[1] ABACHIZADEH M, YAZDI M R H, YOUSEFI?KOMA A. Optimal tuning of PID controllers using artificial bee colony algorithm [C]// Proceedings of IEEE/ASME International Conference on Advanced Intelligent Mechatronics. Montreal: IEEE, 2010: 379?384.
[2] MIRJALILI S. Dragonfly algorithm: a new meta?heuristic optimization technique for solving single?objective, discrete, and multi?objective problems [J]. Neural computing & applications, 2016, 27(4): 1053?1073.
[3] 王慶喜,儲(chǔ)澤楠.基于動(dòng)態(tài)布谷鳥搜索算法的PID控制器參數(shù)優(yōu)化[J].計(jì)算機(jī)測(cè)量與控制,2015,23(4):1215?1217.
WANG Qingxi, CHU Zenan. Parameters optimization of PID controller based on dynamic cuckoo search algorithm [J]. Computer measurement & control, 2015, 23(4): 1215?1217.
[4] 楊智,陳志堂,范正平,等.基于改進(jìn)粒子群優(yōu)化算法的PID控制器整定[J].控制理論與應(yīng)用,2010,27(10):1345?1352.
YANG Zhi, CHEN Zhitang, FAN Zhengping, et al. Tuning of PID controller based on improved particle?swarm?optimization [J]. Control theory & applications, 2010, 27(10): 1345?1352.
[5] 胡偉,徐福緣.基于改進(jìn)粒子群算法的PID控制器參數(shù)自整定[J].計(jì)算機(jī)應(yīng)用研究,2012,29(5):1791?1794.
HU Wei, XU Fuyuan. Self?tuning of PID parameters based on improved particle swarm optimization [J]. Application research of computers, 2012, 29(5): 1791?1794.
[6] 熊偉麗,徐保國(guó),周其明.基于改進(jìn)粒子群算法的PID參數(shù)優(yōu)化方法研究[J].計(jì)算機(jī)工程,2005,31(24):41?43.
XIONG Weili, XU Baoguo, ZHOU Qiming. Study on Optimization of PID parameter based on improved PSO [J]. Computer engineering, 2005, 31(24): 41?43.
[7] 鄭坤明,張秋菊.基于彈性動(dòng)力學(xué)模型與遺傳算法的Delta機(jī)器人模糊PID控制[J].計(jì)算機(jī)集成制造系統(tǒng),2016,22(7):1707?1716.
ZHENG Kunming, ZHANG Qiuju. Fuzzy PID control of delta robot based on elastic dynamic model and genetic algorithm [J]. Computer integrated manufacturing systems, 2016, 22(7): 1707?1716.
[8] WIKELSKI M, MOSKOWITZ D, ADELMAN J S, et al. Simple rules guide dragonfly migration [J]. Biology letters, 2006, 2(3): 325?329.
[9] 李遠(yuǎn)梅,張宏立.基于改進(jìn)螢火蟲算法PID控制器參數(shù)優(yōu)化研究[J].計(jì)算機(jī)仿真,2015,32(9):356?359.
LI Yuanmei, ZHANG Hongli. Optimization of PID controller parameters based on improved glowworm swarm algorithm [J]. Computer simulation, 2015, 32(9): 356?359.
[10] 曾成,趙錫均.改進(jìn)量子遺傳算法在PID參數(shù)整定中應(yīng)用[J].電力自動(dòng)化設(shè)備,2009,29(10):125?127.
ZENG Cheng, ZHAO Xijun. Application of improved quantum genetic algorithm in PID parameter tuning [J]. Electric power automation equipment, 2009, 29(10): 125?127.
[11] 蔡超,周武能.人工蜂群算法整定PID控制器參數(shù)[J].自動(dòng)化儀表,2015,36(8):74?77.
CAI Chao, ZHOU Wuneng. Self?tuning PID parameters by using artificial bee colony algorithm [J]. Process automation instrumentation, 2015, 36(8): 74?77.
[12] HUYNH T H. A modified shuffled frog leaping algorithm for optimal tuning of multivariable PID controllers [C]// Proceedings of IEEE International Conference on Industrial Technology. Chengdu: IEEE, 2008: 1?6.