Researching on YH100 Numerical Control Servo Press Hydraulic Control System and Control Algorithm

Transcription

1 Sensors & Transducers, Vol. 79, Issue 9, September 04, pp Sensors & Transducers 04 by IFSA Publising, S. L. ttp:// Researcing on YH00 Numerical Control Servo Press Hydraulic Control System and Control Algoritm Kai LI, Lingfeng TANG, Youmin WANG, Cun ZHAO Scool of Mecanical and Automotive Engineering, Anui Polytecnic University, Wuu, Anui province, 4000, Cina Tel.: Received: 3 June 04 /Accepted: 9 August 04 /Publised: 30 September 04 Abstract: In order to study te numerical control (NC) servo press ydraulic control system and its control algoritm. Te numerical control servo press performance and control principle of ydraulic control system are analyzed. According to te flow equation of te ydraulic control valve, ydraulic cylinder flow continuity equation and te force balance equation of te ydraulic cylinder wit load press, te matematical model of ydraulic control system is establised. And te servo press ydraulic system transfer function is deduced. Introducing te suitable immune particle swarm control algoritm for servo press ydraulic system, and te control system block diagram is establised. Immune algoritm is used to optimize new control parameters of te system and adopt te new optimization results to optimize te system simulation. Te simulation result sows tat te ydraulic system s transition time controlled by te immune particle swarm algoritm is sorter tan traditional ones, and te control performance is obviously improved. Finally it can be concluded tat immune particle swarm PID control ave tese caracteristics suc as quickness, stability and accuracy. Applying tis principle into application, te obtained YH00 numerical control servo press ydraulic control system meets te requirement. Copyrigt 04 IFSA Publising, S. L. Keywords: Numerical control servo press, Hydraulic system, PID control, Immune particle swarm optimization algoritm, Simulation.. Introduction Servo press usually refers to te use of servo motor drive control of te press. Servo press can be divided into mecanical drive servo press and ydraulic servo press. Mecanical drive servo press converts te rotary motion of te motor into linear motion of te slider on te basis of discarding te traditional mecanical press s flyweel and clutc and some oter energy consumption parts, depending on pressing force [] provided by te servo motor instantaneous torque, troug te screw, crank connecting rod, toggle actuator. Tis can realize te slider s movement being controlled and meet te requirement of intelligent stamping macining flexibility. Hydraulic drive servo press is consisted of controller, servo valve, all kinds of sensors, te oil cylinder and oil circuit. Adopt servo motor drive main transmission oil pump, calm down control valve circuit and control ydraulic macine slider. Hydraulic system is a key part of te servo press, wic will affect te quality of te design of servo press. At present te system commonly uses control metods of PID control, optimal control, robust ttp:// 49

2 Sensors & Transducers, Vol. 79, Issue 9, September 04, pp control, adaptive control, etc. PID control wic is conducted on te basis of te simplified model becomes one of te main industrial control tecnologies because of its simple structure, good stability, reliability and easiness to adjust. However, te traditional PID controller using a linear combination of metods is difficult to reconcile te contradiction between speed and stability. In te case wit parameter variations and external disturbances, it is difficult to ensure tat te system as good robustness. And te optimal control requires accurate matematical model. Accurate matematical model of te servo press is too complex to apply to practical application. Robust control and adaptive control work well. But te control performance would be impacted due to te interference of uncertain factors []. Te immune particle swarm control as suc advantages as a nonlinear mapping, self-study, selforganization, adaptive, etc. Tis paper applies te immune particle swarm control metod to servo press ydraulic control system, and compare wit PID control metod. Simulation results sow tat te immune particle swarm algoritm as fast response speed, small peak and good stability.. Control System Modeling.. Scematic Diagram of Control System Fig. sows te control diagram of ydraulic control system of te servo press: x is te command signal; U i is te command voltage; U e is te bias voltage; i is te current; P L is te flow; X p is te valve displacement; U f is te feedback voltage. Fig.. Servo press ydraulic control principle diagram. Its principle is: After te computer sends command signals, digital to analog converses te command voltage is compared wit te feedback voltage [3], ten voltage deviation is obtained. Troug te amplifier to drive electro-ydraulic servo valve, so as to drive te ydraulic cylinder... Modeling of Hydraulic Control System In order to deduct te transfer function of te ydraulic dynamic element, te basic equations sould be listed first [4], tat is te flow equation of te ydraulic control valve is defined as Equation (), ydraulic cylinder flow continuity equation is defined as Equation () and te force balance equation of te ydraulic cylinder wit load is defined as Equation (3). ql = KqXv KcpL, () dxp Vt dpl ql = Ap + CtppL +, dt 4β dt () e stiffness load; F L is active at any external load force on te piston. Equations ()-(3) are te tree basic equations of valve controlled ydraulic cylinder, tey completely describe te dynamic caracteristics of valve controlled ydraulic cylinder. Te Laplace transform of tem are Equations (4)-(6). QL = Kqxv KcPL, (4) Vt QL = ApsX p + CtpPL + spl, (5) 4β AP = msx + BsX + KX + F, (6) p L t p p p p L By te tree basic equations, a linear transfer function block diagram can be obtained, wic is sown as Fig.. e d xp dxp p L t p p L Ap = m B Kx F dt + dt + +, (3) were C tp is te total leakage coefficient of ydraulic cylinder, C = + ep Ctp C ip ; m t is te total mass of te piston and its load converted on; B p is te viscous damping of te piston and te load; K is te spring Fig.. System transfer function block diagram. Te system open-loop transfer function can be obtained from Fig., wic is sown as Equation (7). 50

3 Sensors & Transducers, Vol. 79, Issue 9, September 04, pp Kv G(s) H(s) =, s δ s( + s+ ) ω ω (7) Te control block diagram of servo press ydraulic system based on immune particle swarm optimization is sown in Fig. 3. Te parameters of te system transfer function assignment: K f is te displacement sensor gain, K f =00 V/m; K α is te servo amplifier gain, K α =.6 A/V; K sv is te proportional servo valve gain, K sv = m 3 /(s A); K v is te open-loop amplification factor; ω = 357 rad / s, δ = 0.. Te open-loop transfer function of te system can be expressed as Equation (8)..7 G(s) H(s) =. s 0. s( + s+ ) Researcing of Servo Press Hydraulic System Immune Control Algoritm 3.. Te Basic Principle of Immune Particle Swarm Optimization (8) Te immune particle swarm optimization is on te basis of framework of particle swarm algoritm, introducing te immune principle of life science into te particle swarm algoritm. On one and, use te immune memory and self regulation mecanism to maintain te fitness level of particles to keep a certain concentration, ensuring te diversity of population. On te oter and, introduce te vaccine and oter operations. Purposely and selectively guide te evolution process, wic can improve te searc performance of te algoritm [5]. Tis paper introduces immune particle swarm algoritm to overcome convergence of te particle swarm algoritm wic cause loss of diversity issues. Tis metod not only retains te briefness and easiness of particle swarm algoritm, but also overcomes te premature penomenon of te optimization process [6], wic makes te algoritm faster and stronger at totally searcing. 3.. PID Controller Design Based on Immune Algoritm Classical PID algoritm is widely used in industrial application. But because of te complicated system model as well as te operator's lacking of experience and oter reasons, it is difficult to determine control various signal process and evaluation index accurately. Using te immune particle swarm optimization to set te tree parameters of PID controller, it can optimize te appropriate parameters rapidly and accurately [7]. Fig. 3. Te immune particle swarm control system. In order to obtain satisfactory excessive dynamic caracteristics, te IAE performance index is used as te minimum objective function in parameter selection. In order to prevent too muc control energy, join te square of control input items in te objective function. Coose te Equation (9) as a parameter to select te optimal index [8]. ( ) ω 0 (t) ω (t) ω3 u, (9) J = e + u dt+ t were t u is te rising time; e(t) is te system error; u(t) is te output of immune particle swarm controllers, ω, ω, ω 3 are te weigt. In order to avoid oversoot penomenon [9], wic refers tat te current output of te controlled system is less tan te previous output. Put te difference before and after te output of te absolute value of te weigting system as te optimal index by using penalty function. Tis can solve te problem of oversoot effectively. Te optimal index is sown in Equation 0. if ey (t) < 0, ( ) ω 0 (t) ω (t) ω4 (t) ω3 u, J= e + u + ey dt+ t (0) were ey = y(t) y(t ), y (t) is te output of te controlled object, ω4 for weigts, and ω4 ω. Due to te fact tat te bigger te fitness is, te better te particle swarm algoritm [0] are. Select te fitness function is f =. Te smaller te J is, J te bigger te fitness f is. Parameter optimization process is sown in Fig System Simulation and Analysis 4.. Determine te Parameters of Immune Control Selecting te initialization number of te particle swarm is N=30, Learning factor C, C value is. 5

4 Sensors & Transducers, Vol. 79, Issue 9, September 04, pp Inertia weigt coefficient ω = Te number of iterations MaxDT=00, te searc space dimension D=0, crossover probability Pc=0.8. Mutation probability Pm=0.-[::N]*(0.0)/N, tat s to say te mutation probability is associated wit te size of te fitness. Te bigger te fitness is, te smaller te mutation probability is. Te range of K p is [0,0], te range of K i and K d are [0,0], weigts ω = 0.98, ω = 0.00, ω 3 =.5, ω 4 = 00. Te proposed immune particle swarm optimization algoritm is used for parameter optimization. Te optimization results are sown as follows: K p =9.0648, K i =0.7498, K d =0, optimal solution J=38.956, fitness f= System Simulation and Analysis After entering te unit step signal, its simulation results is sown in Fig. 6. Te time of Immune particle swarm optimization PID control and plateauing time of traditional PID control are 0.05 s and s respectively. Te simulation results sow tat te response speed of immune particle swarm optimization control is faster, te stable time is sort, and ave no oversoot response curve, te system is stable. Fig. 6. Te results of simulation diagram. Fig. 4. Optimization process diagram. 4.. Parameter Optimization Servo press ydraulic system is te system of te single-input and single-output. Te iterative optimization process of optimal index J is sown in Fig. 5. Fig. 5. Te iterative optimization process of te optimal index. 5. Experimental Verification 5.. Experimental Principle Tis experiment is completed on process integrated automation control system experimental platform. Te experiment uses te boiler liner as a controlled object, and te boiler liner water temperature is te controlled variable of te system. Te experiment requires water temperature in te boiler tank keeping to given value, and coose te boiler tank temperature TT signal detected by platinum resistance as te feedback signal. Use te difference of compared given value to control an output voltage of tree pase voltage regulator module (Te tree-pase voltage of te electric eating tube) to acieve te purpose of controlling te boiler water tank. Te experiment adopts te sceme of adding cooling water into te jacket of te boiler as te external signal interference to verify te effect of te 3 parameters proportional, integral, differential based on immune particle swarm algoritm on controlling te experiment process of boiler temperature. 5

5 Sensors & Transducers, Vol. 79, Issue 9, September 04, pp Experimental Content Storing enoug water in te water storage tank before te experiment. Start te instrument, select te PID control law, set te parameters according to te adjusted parameters of PID controller. Switc te regulator to "automatic" state wen te boiler liner water temperature is stable on te given value. After te water temperature is balance, increase (decrease) te set value to let it as a positive (or negative) step incremental cange (Tat is step interference, tis increase sould not be too large, usually 5 % to 5 % of te set value). So te boiler liner water temperature can leave te original equilibrium state, wic is sown in Fig. 7. Te water temperature stability is turned to a new setting value after a period of adjustment time. Record te setpoint of te intelligent instrument, te output value and instrument parameters. Te response curve process of te liner temperature is sown in Fig. 8. After entering controlling experimental parameters into te system interface, te curve grap of te boiler tank water temperature constant value control transition is sown in Fig. 8. Wen te boiler tank water temperature reaces to te specified temperature of 40 C, te system would inject a certain amount of cold water to te tank for cooling automatically. After te first external interference, toug te traditional PID control it can be seen tat te time wic system need to get back to te stable state is 5 min in te Fig. 9. If applying te same external disturbance again after te system being back to stable state, as it can be seen from Fig., te time wic system needs to get back to te stable state is 3 min. Compared wit te previous one, te oversoot is decreased. It can be concluded tat te immune particle swarm algoritm PID control can acieve te set temperature under te interference signals quickly and steadily, wic proves te immune particle swarm algoritm PID control is superior to te traditional PID control. At te same time it proves te rationality and te correctness of te algoritm. Fig. 7. PID control based on immune particle swarm subject to external interference. Fig. 9. Servo press ydraulic system control process curve. 6. Hydraulic System Pressure Test and Trial Operation Appyling te results into application, it produces te YH00 NC servo press ydraulic control system wic meets te requirement. Fig. 0 is a servo ydraulic system and press boarded experiment. Fig. 8. Te boiler liner water step response curves Experiment Parameter Input and Analysis of Experimental Results Set te temperature at 40. Entered te tree parameters based on engineering experience K p =3, K i =, K d = into te system control interface. Entered te tree parameters drew by immune particle swarm optimization K p =3, K i =0.5, K d = into te system control interface. Fig. 0. Te boarded of NC servo press ydraulic system. 53

6 Sensors & Transducers, Vol. 79, Issue 9, September 04, pp Conclusions ) Troug te flow equation of te ydraulic control valve, ydraulic cylinder flow continuity equation and te force balance equation of te ydraulic cylinder wit load, te control system model is establised. Te stability of te system s transfer function analysis sows tat te system is stable. ) NC servo press ydraulic control system s immune particle swarm optimization algoritm is developed. Te simulation and experiment results sow tat immune control algoritm increases te system s convergence speed and global searcing capability and effectively and can avoid "premature" penomenon. Te immune particle swarm optimization algoritm PID controller is designed. Troug te parameter optimization, identify te output system of K p, K i and K d under te excitation signal. Te simulations reveal tat wen compared wit te traditional PID control of NC servo press ydraulic system, te indicators of te NC servo press ydraulic system based on immune particle swarm control are significantly improved and its control caracteristic is obvious. 3) Verify te accuracy and te advantages of te immune particle swarm optimization algoritm. Test sows tat te immune particle swarm algoritm PID control is superior to te traditional PID control. At te same time it proves te rationality and validity of tis algoritm. Applying te results into application, te YH00 NC servo press ydraulic control system is produced and it meets te requirements. Acknowledgments Te accomplisment of tis paper is supported by Science and Tecnology Researc Project of Anui Province (No ) Science and Tecnology Project of WuHu (No. 03cxy08), National Natural Science Foundation of Cina (Grant No ). References []. Jian Li, Servo press development and application, Cina First Heavy Industries Tecnology, No. 5, 00, pp. -5 (in Cinese). []. Byeong-Mook Cung, Yoon-Kyu Lim, and Gregory D. Buckner, A learning metod for multivariable PID control syntesis based on estimated plant Jacobian, International Journal of Control, Automation, and System, Vol. 7, Issue 5, 009, pp (in Cinese). [3]. Cun Zao, and Youmin Wang, Exploring servo pressure ydraulic system design and researc on its immune-particle swarm optimization control algoritm, Mecanical Science and Tecnology for Aerospace Engineering, Vol. 3, Issue, 0, pp (in Cinese). [4]. Tingting Zao, Hao Wang, and Minquan Jia, Solutions and designs of servo presses, Macine Tool and Hydraulics, Vol. 38, Issue 8, 00, pp. -3 (in Cinese). [5]. Anqiang Li, Liping Wang, Weimin Lin, and Cangming Ji, Application of immune particle swarm optimization algoritm to sort-term optimal dispatc of cascade ydropower stations, Journal of Hydraulic Engineering, Vol. 39, Issue 4, 008, pp (in Cinese). [6]. Fu Duan, and Tongfen Su, Modified immune particle swam optimization algoritm and its application, Journal of Computer Applications, Vol. 30, Issue 7, 00, pp (in Cinese). [7]. Yanwen Huang, Piceng Tung, and Cengyu Wu, Tuning PID control of an automatic arc welding system using a SMAW process, International Journal of Advanced Manufacturing Tecnology, Vol. 34, 007, pp (in Cinese). [8]. Yuqin Wang, Youmin Wang, and Kaiyu Bo, Researc for te veicle semi-active suspension system based on immune-particle swarm optimization algoritm, Journal of Jinggangsan University, Vol. 3, Issue 6, 00, pp (in Cinese). [9]. Jin Guan, Jianua Ceng, Lei Wu, Xingbo Liu, and Suai Gan, Te state switc tecnology of gimbaled INS, Sip Building of Cina, Vol. 49, Issue, 008, pp (in Cinese). [0]. Zixiong Liu, and Hua Liang, Parameter setting and experimental analysis of te random number in particle swarm optimization algoritm, Control Teory and Applications, Vol. 7, Issue, 00, pp (in Cinese). 04 Copyrigt, International Frequency Sensor Association (IFSA) Publising, S. L. All rigts reserved. (ttp:// 54