74 14 PWM 1 PWM L a I a I a R a E a = U a 1 PID 2 3 T e = J D θ K D θ T M 2 fuzzy cerebellar model artculaton controller FCMAC E a = K e θ 3 T

Size: px

Start display at page:

Download "74 14 PWM 1 PWM L a I a I a R a E a = U a 1 PID 2 3 T e = J D θ K D θ T M 2 fuzzy cerebellar model artculaton controller FCMAC E a = K e θ 3 T"

纹地方
5 years ago
Views:

1 14 21 ELECTRI C MACHINES AND CONTROL Vol. 14 No. May PID FCMAC FCMAC PID BP FCMAC PID BP PID 1. 1 N. 1 N 1. % 3% 2%. 7 s. s FCMAC PID BP TP 273 A X Control and smulaton of electrc control loadng system of flght smulator SHI Xudong 1 2 ZHANG Fa 1 JING Tao 1 LI Dayong 2 1. Aeronautcal Automaton School Cvl Avaton Unversty of Chna Tanjn 33 Chna 2. MeasurementControl Tech & Communcatons Engneerng School Harbn Unversty of Scence and Technology Harbn 8 Chna Abstract There are problems such as nonlnearty dsturbance extraneous force and oscllaton of external rng n a flght smulator electrc control loadng system controlled by sngle PID. The parallel controller neural networks has the feature of robust fuzzy cerebellar model artculaton controller FCMAC can study any nonlnear map. Also BP neural networks compensator has the character of beng adaptve and the ablty to reduce the nfluence of extraneous force. Therefore a composte neural network control algorthm was proposed whch conssts of FCMAC neural networks and PID parallel controller and back propagaton BP neural networks. The control system was modeled and smulated based on the composte neural network. The results show that the devaton between model torque and output torque s nduced from 1. 1 N to. 1 N and the devaton n dsturbng sgnal stays the range of 1. %. Also the overshoot s nduced from 3% to 2% and stablzaton tme s shortened from. 7 s to. s. Key words flght smulator electrc control loadng system composte neural network FCMAC and PID parallel controller BP neural networks adaptve compensator MHR

2 74 14 PWM 1 PWM L a I a I a R a E a = U a 1 PID 2 3 T e = J D θ K D θ T M 2 fuzzy cerebellar model artculaton controller FCMAC E a = K e θ 3 T e = K t I a 4 FCMAC PID L a R a U a E a I a T e 4 BP Back Propagaton T M K t J D K D FCMAC PID BP PWM T PWM k m PWM U a = k m Ts 1 U c 1 T θ s = 1. 1 Js 2 Ds k 6 J D k PWM 1 F n A /D 2 f o F m F out F delt PWM PWM F out F f o k m1 k m2 a b y 驾驶杆 F n 位移信号速度信号力信号状态参数控制界面状态指令舵偏角高速微处理器传动机构 PWM 信号 PWM 驱动器直流力矩电机 Fg. 2 F n 2 Curves of dsplacement and force relatonshp 3 角度兹 k 3 1/(Js 2 Dsk) s J DsK D k 1 位移 y k e F out k m/(ts1) K t/(l asr a) 1/k 3 Fg Block dagram of electrc control loadng system Fg. 3 Block dagram of system transfer functon

3 7 2 6 PWM 3 x PID 4 F n 角度兹位移 y k 3 1/(Js 2 Dsk) k 1 PID Fg. 4 FCMAC PID u u u c 补偿器 s k m/(ts1) k e J DsK D K t/(l asr a) 1/k 3 F out 4 FCMAC N mn N max Block dagram of FCMAC compound control 2. 1 FCMAC PID F = N max N mn x x mn N x max x mn N mn 8 mn JoonHo Park x FCMAC x 6 mn x max PID N mn N max 7 FCMAC PID 2 PID FCMAC PID 8 F m NB NM NS ZO PS PM PB 7 A PID PID 1 A 2 A FCMAC 7 x A μ x = exp [ x c 2 ] FCMAC u σ 2 n u c σ FCMAC 3 Fg. r FCMAC 控制器 u n PID 控制器 u p u 飞行模拟机电动操纵负荷模型 FCMAC PID Block dagram of FCMAC and PID parallel control y CMAC 4 9 FCMAC FCMAC FCMAC p PID w p u k = u n k u p k 7 u p = a p w p 11 FCMAC FCMAC CMAC FCMAC Fg. 6 6 移输出加法器 FCMAC Structure of FCMAC neural networks 1 FCMAC x F m FCMAC 7 a = μ FCMAC

4 76 14 u n k = 7 u p 12 p = 1 F = G 2 s y 17 6 FCMAC FCMAC BP Δw k = η E k w = η u k u n k a w k = w k 1 Δw k α w k w k 1 13 E k = u n k u k 2 /2 η α c σ Δc = η E k = η u c n k u k w μ x x c σ 2 c k = c k 1 Δc α c k 1 c k 2 Δσ = η E k = 2η u k σ u n k μ x x c 2 σ FCMAC w j w o 4 u FCMAC FCMAC PID u PWM u c d 2. 2 E k = 1 2 F m k F o k 2 22 F o = G 1 s F m G 2 s y 7 16 位移 y 1/k 1 s J DsK D FCMAC k e 输出力 F o PID k m/(ts1) K t/(l asr a) 1/k 3 Fg. 7 7 Block dagram of poston dsturbance system 4 1 a O 1 k = X k = x1 k x2 k x3 k x4 k 18 k k 1 k 2 1 x1 k = y k x2 k = y k 1 x3 k = y k 2 x4 k = 1 b net 2 k = 4 O 2 j = 1 w j x j k = f net 2 k = Sgmod σ k = σ k 1 Δσ α σ k 1 σ k 2 f x = ex e x e x e x c O 3 k = = 1 2 w o O 2 k 21 w k = η 1 E k w α 1w k 1 23 η 1 α 1 E k w = F m k F o k sgn ( F o k O 3 k ) O 2 k 24

5 77 E k = σ 2 w w o f net 2 k O 1 j k j 1. 1N. 1N PID σ 2 = F m k F o k sgn ( F o k O 3 k ) 3% 2%. 7 s. s k = y k 1 y k PID 8 F n Fg. 8 k 3k 1/(Js 2 Dsk) PID 控制器系统内环 8 位移 y F out Block dagram of external rng s optmzaton desgn 3 (b) 外环加入 PID 控制 L a = 7. 1 H R a = 7. 9 Ω 输出力 F 2 out 6 K e =. V mn/r 模型力 F m K t = J D =. 12 kg m 2 K D =. 7 PWM K m = 6 PWM T =. 2 s J = kg m 2 D = 1 k 3 =. 13 k 1 = 13 9 k = Fg. 9 Curves of response to step sgnal f = 4 N k m1 = 1. 2 k m2 =. 99 FCMAC N mn = 12 N max = 12 η =. 1 α =. BP η 1 =. 3 α 1 =. 9 PID P 1 = I 1 =. 2 D 1 =. 3 PID P = I =. 2 D = 输出力 F out N 1. % 9 a PID PID FCMAC PID PID PID PID PID 9 b 9 a PID PID 8 BP PID 9 c BP (a) PID 控制 F m F out (c) 复合神经网络控制 F n = 2sn 2. πt t = s PID

6 输出力 F out 模型力 F m (a) PID 控制.4 输出力 F out Fg (b) FCMAC 复合控制 Curves of response to dsturbance PID 7 FCMAC PID HE Xaox. Fuzzy neural network control of elec FCMAC ZHONG Wefeng BP trohydraulc poston servo system J trol J IPC 61H system J 7. learnng method for functon approxmaton 2. J SHEN Dongka HUA Qng WANG Zhanln. Motordrven load system based on neural networks J. ACTA Aeronautc ET Astronautc Sntca J WANG Lngwen LI Yuhua A study of suppress strategy for extra force on control loadng hydraulc servo system machne J. Tool & Hydraulcs CMAC J HOU Hajun LEI Yong YE Xaoyong. Nonlnear adaptve nverse control based on fuzzy CMAC network J. Journal of System Smulaton J LU Yng WANG Yonglang WANG Zhanln et al. Forcefeelng smulaton system of longtudnal control based on the electromotve force servo J. Journal of System Smulaton PARK JoonHo KIM TaeKue YOON TaeSung. Systematc control parameter tunng for actuator n control loadng system C IEEE Internatonal Conference on Industral Technology December Bhubaneswar Inda J Electrc Machnes and Con. CMACPID ZHU Zhaoyan YUE Dongha WANG Xancheng. The applcaton of CMACPID parallel control algorthm n electrohydraulc servo. Chnese Hydraulcs & Pneumatcs LAI Hungren WONG Chngchang. A fuzzy CMAC structure and C The 1th IEEE Internatonal Conference on Fuzzy Systems December 2 21 Melbourne Australa M.

34 7 S R θ Z θ Z R A B C D PTP θ t 0 = θ 0 θ t 0 = 0 θ t 0 = 0 θ t = θ θ t = 0 θ t = 0 θ t V max θ t a max 3 θ t A θ t t 0 t / V max a max A = 3 4 S S

34 7 S R θ Z θ Z R A B C D PTP θ t 0 = θ 0 θ t 0 = 0 θ t 0 = 0 θ t = θ θ t = 0 θ t = 0 θ t V max θ t a max 3 θ t A θ t t 0 t / V max a max A = 3 4 S S 7 0 7 No. 7 Modular Machne Tool & Automatc Manufacturng Technque Jul. 0 00-65 0 07-0033 - 06 *. 0040. 04 PTP s TH6 TP4. A The Moton plannng of Handlng Robot Based on Tme Optmal CAO Bo CAO Q-xn TONG Shang-gao