11.07.2015 Views

Design and Simulation of Active Suspension System by Using Matlab

Design and Simulation of Active Suspension System by Using Matlab

Design and Simulation of Active Suspension System by Using Matlab

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

When we want to consider input W(s) only, we set F a (s)=0.Thus we get the transfer function G 2 (s) as the following:3 2X1(s)− X 2(s)− m1b2s− m1k2sG 2(s)==(7)W(s)∆We can put the above Transfer Function equations (6) <strong>and</strong>(7) into <strong>Matlab</strong> <strong>by</strong> defining the numerator <strong>and</strong> denominator<strong>of</strong> Transfer Functions in the form, nump/denp for actuatedforce input <strong>and</strong> nums/dens for disturbance input, <strong>of</strong> thest<strong>and</strong>ard transfer function G 1 (s) <strong>and</strong> G 2 (s):numpnumsG1 (s) = G2(s)=(8)denpdensOn the basis <strong>of</strong> above equations we have created an m-filewith the data that we have found. We can use <strong>Matlab</strong> todisplay how the original open-loop system performs(without any feedback control). Adding the comm<strong>and</strong> 'step(nump,denp)' into the m-file <strong>and</strong> running the file in the<strong>Matlab</strong> comm<strong>and</strong> window, we get the open-loop responseto unit step actuated force.Figure 3. tells us that, when the bus passes a 10 cm highbump on the road, the bus body will oscillate for anunacceptably long time (≈ 50 sec), <strong>and</strong> with a much largeramplitude than the initial impact. The passengers in that buswill not be comfortable with such an oscillation. The bigovershoot <strong>and</strong> the slow settling time will cause damage tothe suspension system. As we have already stated, thesolution to this problem is to add a feedback controller intothe system to improve the performance. We have opted forPID controller, for we have come to the conclusion that noother solution would meet the initial requirements in theright way. The schematic <strong>of</strong> the thus obtained close-loopsystem is the following:r=0 ++ControllerPlantx -x 1 2+ F-aFWFigure 4. Closed-Loop <strong>System</strong> Block DiagramMODELLING BY USING STATE-SPACEEQUATIONFigure 2. Open-Loop Response to Unit Step Actuated ForceFigure 2. tells us that the system is underdumped. Peoplesitting in the bus will feel a small amount <strong>of</strong> oscillation <strong>and</strong>the steady-state error is about 0.01 mm. However, the busneeds unacceptably long time to reach the steady state - thesettling time is rather long. The solution to this problem liesin including a feedback controller into the block diagram <strong>of</strong>the system. Adding the comm<strong>and</strong> 'step(0.1*nums,dens)'into the m-file we get the open-loop response to 0.1 m stepdisturbance.Figure 3. Open-Loop Response to 0.1 m Step DisturbanceThe representation <strong>of</strong> a system in the form <strong>of</strong> the statespaceequation is usually much easier to derive fromdifferential equations than Laplace transform method.Difficulties arise when the derivatives <strong>of</strong> the inputs appearin the differential equations, which leads to a situationrequiring much algebra. The dynamic equations whichinclude the mass both <strong>of</strong> the bus <strong>and</strong> suspension system, arethe same as with modelling <strong>by</strong> using transfer function -equations (1) <strong>and</strong> (2). To be a valid state-spacerepresentation, the derivative <strong>of</strong> all states must be in terms<strong>of</strong> inputs <strong>and</strong> the states themselves. The state variables may,but generally don't have to, be the system output at thesame time. Now, let's choose the states that we shall beusing. To begin, let's divide the equations (1) <strong>and</strong> (2) <strong>by</strong> m 1<strong>and</strong> m 2 , respectively <strong>and</strong> introduce the substitute Y 1 = x 1 -x 2 .Note that W appears in the equation for x 2 :b1k1F ax 1 = − Y1 − Y1+(9)m1m1m1b1k1b2x2= Y1 + Y1+ (W− x2)+m2m2m2(10)k 2Fa+ (W − x2)−m2m2The first state-space equation will be x 1 . Since noderivatives <strong>of</strong> the input appear in the equation for x 1 , wechoose x 1 for the second state. Then, we choose the thirdstate as the difference between x 1 <strong>and</strong> x 2 . After doing somealgebra, we will determine what the fourth state should be.So we subtract equation (10) from equation (9) to get anexpression for Y 1 :3

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!