HYDRAULIC ACTIVE GUIDE ROLLER SYSTEM FOR HIGH-SPEED ELEVATOR BASED ON FUZZY CONTROLLER (2).doc

chinese journal of mechanical engineering68*vol.20, no. 5, 2007feng yonghui zhang jianwuschool of mechanical engineering,shanghai jiaotong university,shanghai 200240, chinahydraulic active guide roller system for high-speed elevator based on fuzzy controllerabstract: increase of elevator speed brings about amplified vibrations of high-speed elevator. in order to reduce the horizontal vibrations of high-speed elevator, a new type of hydraulic active guide roller system based on fuzzy logic controller is developed. first the working principle of the hydraulic guide system is introduced, then the dynamic model of the horizontal vibrations for elevator cage with active guide roller system and the mathematical model of the hydraulic system are given. a fuzzy logic controller for the hydraulic system is designed to control the hydraulic actuator. to improve the control performance, preview compensation for the controller is provided. finally, simulation and experiments are executed to verify the hydraulic active guide roller system and the control strategy. both the simulation and experimental results indicate that the hydraulic active guide roller system can reduce the horizontal vibrations of the elevator effectively and has better effects than the passive one, and the fuzzy logic controller with preview compensation can give superior control performance. key words: high-speed elevator horizontal vibmions hydraulic /mt glide roller system fuzzy logic control0 introductionhorizontal vibration is ens of the main problems affecting ride comfort in elevators. there are many reasons contributing to the horizontal vibrations of elevators, such as the irregularity of the guide rails, the abnormity of the guide rollers, the running speed of elevators, the offset load of passengers, the wind buffeting, etc. research indicates that these horizontal vibrations are mainly generated by the irregularity of the guide rails1 and are in proportion to the running speed of elevators23 to meet the need of high and super-high buildings, elevator speed is now becoming higher and higher, which results in serious horizontal vibrations. suppression method of horizontal vibrations is necessary for high-speed elevator. the conventional horizontal vibration reduction techniques include structure changes and passive guide roller system using stiff spring. these passive suppression methods need no power supply and have the advantages of simple structure, low price and high reliability and can work well when the elevator speed is low. but they cannot do with the increasing elevator speed because of such disadvantages as limited frequency band width and low working force. active vibration reduction techniques can flexibly cope with the new situations and have good effects and have been used widely. accordingly, studies on the active suppression method of horizontal vibrations for high-speed elevators have been advanced. now mitsubishi co. and others have already developed their active guide roller device to alleviate horizontal vibrations of high-speed elevators based on magnetic effect451. and they have acquried patent protection for that61. accordingly, we propose in this paper a new type of active guide roller system based on hydraulic actuator which can reduce the horizontal vibrations of high-speed elevators. first, the structure and the working principle of the hydraulic active guide roller system are analyzed. then the mathematical model of the system is given. and the fuzzy logic controller with preview compensation is designed to control the active guide system. last, computer simulation and experimental tests are performed to verify the effectiveness of the hydraulic device and the controller.1 active control model of horizontal vibrations for elevator cagethe elevator system is a kind of complex multi-body structure. elevator cage is the part loading the passengers. fig. 1 is theconfiguration of a.i elevator cage. it includes a car body, a frame and four passive guide roller systems. the car body is supported by the frame and the guide system is fixed on the frame. the three rollers of the conventional guide systems are contacted with the t type rails tightly by the stiff spring damper. the elevator goes up and down when the guide roller runs along the rail. the unevenness of the guide rails is transmitted to the elevator cage through the guide roller system and brings about horizontal vibrations of the elevator. this problem can be overcome by the active guide roller system.guide rollerfel-mffig. 1 configuration of elevator cage(1)considering the horizontal vibrations in the direction parallel to the elevator door, the dynamic model of the horizontal vibrations for elevator cage with active guide roller system is developed as shown in fig. 2. here the car body and the frame are supposed to be a whole rigid symmetrical body, the rollers are modeled as mass-spring-damper units with same structure and parameters71, the elevator runs at the constant velocity v, and only the rail disturbance is taken into account. the motion equations of the elevator cage in the horizontal direction are given asreceived september 12, 2006; accepted june 5, 2007received in revised form may 15, 2007;jo(2) 1994-2007 china academic journal electronic publishing house. all rights reserved, chinese journal of mechanical engineering69*cylindercylinder bracket. roller level -ohsv. accumulator -oil boxpressure sensor controllerfig. 3 configuration of the hydraulic active guide roller2.2 mathematical modelsthe equations of pressure and fluid flow of the hydraulic cylinder are shown as(4) (5)al0+xc-x,)p+ a(x, - ir) = a?kpiston rod surface area-initial displacement of the piston -displacement of the roller brackets*rp k-aq-9i-qo-1*-displacement of the guide roller -hydraulic pressure in the cylinder -oil bulk modulus -fluid flow of the cylinder -fluid flow of ihsv -fluid flow of ohsv-compressible fluid flow between the accumulator and the cylinder so the active control force produced by the hydraulic actuator can be expressed as followsf,=pa(6)where the sign in eq. (6) is positive when i is equal to 1 or 3, and is negative wheni is equal to 2 or 4.the fluid flow equations of the high-speed on/off valves are shown asp)2(p,(7)9i = cddri2(pp2) p(8) )p(10)where me-j-x-0 -mass of the elevator cage-moment of inertia of the elevator cage-displacement of the elevator cage-rotation angle of the elevator cage-vertical distance between the guide roller and thefrcenter of the elevator cage -active control force, produced by the actuator(act),=1,-, 4,4,*rltc,|act( m,mv.a0%|ewhere (3)f$hxw3jcr3xxr4*w4fig. 2 dynamic model of horizontal vibrations for sleviiior the motion equations of the guids rollers art given as xrl = -*r( - xt) + cr(xrt - -xw/) - f-,where m,mass of the guide rollerk,spring constant of the rollerc,damper constant of the rollerxtldisplacement of the rollerxwj-rail disturbance*wl(0=*w3(+o) *w2(0=*w4(+o)t-2luf+u 0uf+u 1(11)uf + up 0 u,+u9-(12)ut+up -luf+up0controllerwhere f is the output of flc and up is the output of pcfeedforward control-1.0 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1.0derivative of p pnb nm ns psze pm pb1-1.0 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1.0 output of flc f fig. 5 membership functions of fuzzy variablestvhle 1 fizz) control rulesdfrivativvof /oi! p.cssure of actuator ppnbnmnszepspmpbnbpbpbpmpszensnsnmpbpbpmpszensnmnspbpmpszensnmnmzepbpmpszensnmnbpspmpmpszensnmnbpmpmpszensnmnbnbpbpspszensnmnbnbpreview controllerfeedback controliconvenor3.2 preview controllerthe preview controller is a feedforward loop. it is just a compensation for the flc. as the rail disturbance varies at all time, the preview control up is assumed to befuzzy controllernp(0 = a*(r)xw(/ + r)dr(13)hsv_ hydraulic cylinderrollerfig. 4 block diagram of controller3.1 fuzzy logic controllerthe fuzzy logic controller is a feedback loop. it is composed of a fuzzification interface, fuzzy rule base, decision making logic and defuzzification interface. fuzzification is the process which transforms input signal into fuzzy sets. the input signal of the flc is the oil pressure of actuator p and its derivative p . the output signal uf means the duty ratio of pwm signal for hsv. when f is positive, ihsv will be on and the ohsv will be off. when wf is negative, the ihsv will be off and the ohsv will be on. when u( is zero, both valves will be off. the input signal p, p and the output signal uf vary in the interval 0, pm, -&pm, apj and -1, 1, respectively. their universe of discourse are defined as 0, 1, -1, 1 and -1, 1. the linguistic variables of the input and output signals are defined as pb, pm, ps, ze, ns, nm and nb. triangular and trapezoid membership functions are used for the fuzzy variables, which are shown in fig. 5. the objective of control is contained in the fuzzy rule base in the form of the linguistic variables using the fuzzy conditional statement. it is summarized in table 1. the construction rule is as follows: when the error between p and po is big, the error should be minimized as quickly as possible; when the error between p and po is small, the overshoot should be avoided to keep the system stable. the most widely used center of gravity method is applied here to defuzzify the output.where a is a negative constant, t is the width of the preview time window which depends on the delay time of the hydraulic system, k is the weight of the disturbance displacement xw it increases when the distance between the sampling place and the roller location decreases and should satisfy the following equation(14)= 1f*(odias computer-aided control in reality is digital control, rewrite eq. (13) and eq. (14) in discrete form81(15) (16)po)=a*p(xo+oa/zm0=iwhere m is the number of preview steps, j represents the current sampling time and a/ is the sampling cycle. suppose*(/) = exp(-p-;)(17)where p is a positive constant. eq. (17) can be normalized as*(/)(18)u0 =i*(04 simulation studybased on the mathematical model in this paper, simulation study is carried out using matlab. table 2 shows part model94- china ca emic jnaeetoicsius at eh/enchinese journal of mechanical engineering71 30parameters. set the objective pressure value as 2.5 mpa. permitted error range of 0.1 mpa is allowed. take the measured rail disturbance as the input signal. fig. 6 shows the curve of the guide rail unevenness. fig. 7 shows the simulation results. fig. 7a is the acceleration response with passive guide roller. from the simulation results we can see that the maximal value is 0.278 7 m s2, the root mean square value is 0.090 1 m s2. the maximal value is head and shoulders above the boundary 0.15 m s2 specified by china elevator standards gb/t 10058-1997 and the ride comfort can be affected seriously. fig. 7b and fig. 7e are the acceleration curve and the pressure curve with the hydraulic active guide roller system based on the fuzzy logic controller, respectively. the acceleration response is reduced effectively with amplitude of 0.141 9m* s2and the root mean square value is 0.045 6 m s2, while the pressure value of the hydraulic actuator exceeds permitted error range. fig. 7c and fig. 7f are the acceleration curve and the pressure curve with the hydraulic active guide roller system based on the fuzzy logic controller with preview compensation. a fuzzy logic controller with preview compensation brings about ideal results. the maximal value of the acceleration response is 0.129 3 m s2 and the root mean square value is 0.032 4 m s2, and the hydraulic pressure are constrained between 2.4 mpa and 2.6 mpa. the preview compensation utilizes the future information in advance and can counteract the time delay of the hydraulic system in some degree. fig. 7d shows :ie power spectry.l unsity(psd) curves of the horizontal vibrations acceleration. the curve of the passive guide roller is displayed in the solid lines. the curve of the hydraulic active guide roller system based on the fuzzy logic controller is displayed in the dashed lines. the curve of the hydraulic active guide roller system based on the fuzzy logic controller with preview compensation is displayed in the dot-dashed lines. from fig. 7d we can conclude that the active guide roller system attenuates the vibrations in the low frequency area effectively, especially the vibrations of the main resonant frequencies. the fuzzy logic controller with preview compensation exhibits better control effects.table 2 simulation parametersvalueparametermass of the guide roller m,/kg0mass of the elevator m/kg2.5x103spring constant of the guide roller v(n m1)134damper constant of the guide roller c,/(kn s m1)671piston rod surface area a/m13.14x10oil bulk modulus avgpa2.55fluid flow discharge coefficient q0.72pressure of supply oil pi/mpa3pressure of the oil box /vmpa0.101oil density p/(kg m5)8605rs -5|l owvvayvaa_!iiiil_601020304050rail place z/m(a) left guide rail70510152025running time f/s(a) acceleration response with passive guide roller101520runr.js i.iie i/s(b) accelera-oi response: with active guide roller by fuzy ccnryjlei2505101520running time f/s(c) acceleration response with active guide roller by fuzzy controller with preview compensation0.012 0.010 0.008 0.006 0.004 0.002i3 4 5 6 7 1 frequency /hz(d) power spectral density of the acceleration2.70510152025running time f/s(e) pressure response with active guide roller by fuzzy controller3030302.6jiiiii_201060304050rail place z/m (b) right guide rail fig. 6 curve of the guide rail unevenness70101520running time f/s(f) pressure response with active guide roller by fuzzycontroller with preview compensationfig. 7 comparison of the simulation results 1994-2007 china academic journal electronic publishing house. all rights reserved, 72-feng yonghui, et al: hydraulic active guide roller system fo