基于nsga-ii算法的多目標(biāo)參數(shù)優(yōu)化的主動(dòng)隊(duì)列管理新策略(完整版)
【正文】 d RED[A] . Proceedings of the INFOCOM’ 99[C]. New York: IEEE Computer Society, 1999. 13461355.[4] Athuraliya S, Low S, Li V H, Yin Q H. REM: Active queue management[J]. IEEE Network, 2001,15(3):4853.[5] Misra V, Gong W B, Towsley D. Fluidbased Analysis of a Network of AQM Routers Supporting TCP Flows with an Application to RED [A]. Proc. ACM/SIGCOMM[C]. 2000, 151160.[6] Hollot C V, Misra V, owsley T D, et al. A Control Theoretic Analysis of RED[A] . Proc. IEEE INFOCOM[C]. Alaska, USA, 2001, 15101519.[7] Hollot C V, Misra V, Towsley D, et al. On Designing Improved Controllers for AQM Routers Supporting TCP Flows [A]. Proc. IEEE INFOCOM[C]. Alaska, USA, 2001, 17261734.[8] WU T B,LIU ZR,WANG J PID parameters based on improved chaos algorithm[J]. Journal of Electronic Measurement and Instrument,2007,21(4):5962.[9] LI L X,PENG H P,et parameter tuning based on chaotic ant swarm[J]. Chinese Journal of Scientific Instrument, 2006,27(9):11041106.[10] SUN J P,YAN of fuzzy PID controllers based on improved genetic algorithms[J]. Chinese Journal of Scientific Instrument, 2006,27(6 Suppl.):19911992.[11] HUANG Y, HAN P. PID controll based on BP neural network for superheated steam temperature sysytem[J]. Chinese Journal of Scientific Instrument, 2006,27(6 Suppl.):19801981.[12] DEB K,PRATA PA,AGARWAL S,et al. A fast and elitist multiobjective genetic algorithm: NSGAⅡ[j].IEEE Transactions on Evolutionary Computation, 2002,6(2):182197.[13] CANTUPAZ E. A summary of research on parallel genetic algorithms[R].IlliGAL Report No: 95007,1995.作者簡(jiǎn)介:陸錦軍（1964 ），男，教授，主要研究方向?yàn)榫W(wǎng)絡(luò)系統(tǒng)的擁塞控制，智能控制理論與應(yīng)用。圖9(a) RED隊(duì)列長(zhǎng)度（d=220ms） 圖9(b) PID（GA）隊(duì)列長(zhǎng)度（d=220ms）圖9(c) PID（SPSO）隊(duì)列長(zhǎng)度（d=220ms） 圖9(d) PID（QDPSO）隊(duì)列長(zhǎng)度（d=220ms）圖9(e) PID（NSGAII）隊(duì)列長(zhǎng)度（d=220ms）從實(shí)驗(yàn)結(jié)果可以看出，RED在大時(shí)滯中出現(xiàn)了持續(xù)震蕩，相比之下，基于GA、PSO、QDPSO、NSGAII優(yōu)化的PID算法響應(yīng)速度較快，但基于NSGAII 的優(yōu)化算法的響應(yīng)速度最快，動(dòng)靜態(tài)綜合性能最好。網(wǎng)絡(luò)拓?fù)浣Y(jié)構(gòu)如圖8所示，仿真實(shí)驗(yàn)結(jié)果與RED、PID(GA)、PID(SPSO), PID(QDPSO) 等算法進(jìn)行比較。偽并行NSGAII算法的參數(shù)設(shè)置如上文所述，式(1)中Gmin取2，Pmin取60，、的取值范圍為：[1，3]、[1，2]、[1，]。每個(gè)子種群采用NSGAII算法進(jìn)行遺傳操作，NSGAII參數(shù)設(shè)置為：圖7偽并行NSGAII算法流程圖① 選擇：聯(lián)賽選擇，選擇規(guī)模為2；② 重組：實(shí)值重組?！斑w移策略”是并行遺傳算法引入了一個(gè)新的算子，它是指在進(jìn)化過程中子群體間交換個(gè)體的過程，遷移可以加快較好個(gè)體在群體中的傳播，提高收斂速度和解的精度，與單種群相比可用較小的計(jì)算量達(dá)到
環(huán)評(píng)公示相關(guān)推薦
鄂ICP備17016276號(hào)-1