Professional Documents
Culture Documents
based on Hossein Mirzaeinejad, Mehdi Mirzaei*, A novel method for non-linear control of wheel slip in anti-lock braking systems , Control Engineering Practice 18 (2010) 918926.
Presented by, Saurabh Gupta - 123014014 Darshan Bang - 123100036 Kajal Khan - 123100046
Contents
Anti-lock Braking System Schematic of ABS Modeling Control system design Control law without integral feedback Control law with integral feedback Simulation Conclusions
Objectives
Reduce stopping distances Improve stability Improve steer-ability during braking
Features
Prevent wheels from locking during heavy braking Modulates the brake line pressure independent of the pedal force
Schematic of ABS
(http://www.google.co.in/imgres?q=abs+system&sa=X&hl=en&biw=1366&bih=667&tbm=isch&tbnid=DVlx205u0VvitM:&imgrefurl=http://www.cvel.clemson.
edu/auto/systems/braking.html&docid=7RQ57SQgNULEmM&imgurl=http://www.cvel.clemson.edu/auto/systems/images/ABS-architecture.png&w=396&h=272 &ei=IzhoUeb-FIWGrAfgtoHgBA&zoom=1&ved=1t:3588,r:12,s:0,i:145&iact=rc&dur=1066&page=1&tbnh=174&tbnw=254&start=0&ndsp=15&tx=147&ty=35)
ABS - Modeling
Governing Equations
= ( ) = =
1 2 = + + V = longitudinal velocity of vehicle = rotational speed of wheel = longitudinal tire force Tb = braking torque = longitudinal slip ratio = friction coefficient R = wheel radius = total mass of the model = wheel mass ( ) + of sprung mass ( )
2 = 2 +
3 = 2
Purpose of controller is to maintain wheel slip x2 = and its integral 3 = , closed to their desired responses
1 =
2 =
= [ ] are the state vectors ( = , = ) = are the outputs of system = Reference model for wheel slip: = 1 Where, = 0.15, = 20
1 2 +
+ +
=2
Performance index
=0 = 1 1 1 + 0.252
Necessary Condition for optimality 1 + 0.252 2 + 0.5 3 + 1 + 0.252 2 2 With Integral feedback Without integral feedback
1 + 2 2 2 3 = 3 3
3 0 2
1 2 + 2 = 2 2
Error in estimating 2 is due to the error in estimation of friction force So if error on is bounded then error in 2 will also be bounded by a constant F>0 2 2 2 () where = Fh
So to control tracking error, h should be decreased, but with decreasing h, control energy becomes large and oscillatory
2 +
1 2 2
1 3 3
Where,
+ (2 2 )
3 and (2 2 ) have the same sign So integral variable error reduces the effect of model uncertainties So wheel sleep tracking error is much less the previous case
Error dynamics of the integral variable 3 : 3 = 2
1 2 2 = 3 1
1 3 2 (2 2 ) + 3 0 0
Steady state tracking error (2 = 3 = 0) 2 = 0, 3 Wheel slip tracking error in steady state will be zero Effect of model uncertainties is transferred to the tracking error of the integral variable
1 3
Simulation
Parameter Value R L hcg mw mvs 0.326 m 2.5 m 0.5 m 40 kg 415 kg
1.7 kgm-2
Flat dry road, =0.8 Velocity 20 m/s (72 km/hr) There is no modeling uncertainty
Simulation results during braking with and without control: (a) wheel and vehicle speed (uncontrolled), (b) wheel and vehicle speed (controlled), (c) wheel slip, and (d) wheel slip tracking error
Simulation Cont
Uncertainties in vehicle mass = 15 % Uncertainties in Coefficient of friction = 10 % Time of prediction (h) = 0.003 s
Comparison of performances of the controller with and without integral feedback: (a) wheel slip tracking error and (b) braking torque.
Conclusions
Controller with integral feedback control is more robust, can handle nonlinearity and uncertainty of the model in a better way Controller without integral feedback As time of prediction (h), error in tracking slip , energy input and torque input ~ Controller with integral feedback reduces the oscillation, with better tracking error
13
Appendix A
Appendix B