Improving Heavy Vehicle Emergency Braking Systems. Jonathan Miller and David Cebon Cambridge University, UK

Transcription

1 Improving Heavy Vehicle Emergency Braking Systems Jonathan Miller and David Cebon Cambridge University, UK

2 Presentation Overview Introduction Sliding Mode Slip Control Friction and Brake Gain Estimation Novel Methods of Actuation Conclusions

3 Acknowledgements - CVDC Haldex Brake Products Camcon Volvo Trucks Arvin Meritor Goodyear Tyres Firestone Industrial Products Qine?q Tinsley Bridge Limited Shell Oil Fluid Power Design Mektronika Systems MIRA Ltd FM Engineering Denby Transport Ltd

4 Key M s Field Tests with an Instrumented Vehicle M u Validated Vehicle Simulations

5 Simulation Validation Full-Scale Experimental Stop Speed, m/s Pressure, Bar Simulated Stop Time, s Speed, m/s Pressure, Bar Time, s

6 The Slip Control Problem Unladen Trailer, µ = 0.4, ABS Braking Speed, m/s Pressure, bar Vehicle Speed Wheel Speed Brake Chamber Pressure Time, s Slow cycling and deep deviations in slip Long stopping distances Large air consumption Braking Force, F x free rolling Optimal Braking Typical ABS slip variation Slip, λ asphalt ice wheel locked

7 Key Features of New ABS System 1. Slip Control 2. Accurate, low- cost iner?al measurement of vehicle speed 3. Con?nuous es?ma?on of op?mal slip point (brake gain) 4. Completely different form of pneuma?c actua?on with much faster valves

8 The Slip Control Problem λ d + Slip Actuator - Controller Dynamics λ P d T b Vehicle F x a x ω w a z ω y Road Friction Estimator Velocity Estimator

9 The Slip Control Problem λ d + Slip Actuator - Controller Dynamics λ P d T b Vehicle F x a x ω w a z ω y Road Friction Estimator Velocity Estimator

10 Slip Controller F x / F z, unitless 5 kph 100 kph Plant is unstable beyond the peak Rough roads Uncertain friction conditions Slip, unitless Variable Structure (Sliding Mode) Controller:

11 Sliding Mode Slip Control Conventional ABS Simulation on Asphalt Slip, unitless F x / F z, unitless Time, s Slip, unitless Sliding Mode Control Simulation on Asphalt Slip, unitless F x / F z, unitless Time, s Need to know road friction Slip, unitless

12 The Slip Control Problem λ d + Slip Actuator - Controller Dynamics λ P d T b Vehicle F x a x ω w a z ω y Road Friction Estimator Velocity Estimator

13 Tyre Force Observer Sliding mode observer to estimate tractive force Measure brake torque, wheel speed, longitudinal accn ABS Stop on Wet Basalt Tile Observer Performance during Post-processing Speed, m/s Fx / Fz, [-] Nominal Value Time, s Slip, unitless

14 The Slip Control Problem λ d + Slip Actuator - Controller Dynamics λ P d T b Vehicle F x a x ω w a z ω y Road Friction Estimator Velocity Estimator

15 Brake Gain Measurement Pulse brake on one wheel for 0.3 s during free rolling Use Recursive Least Squares algorithm to estimate the brake gain Speed, m/s Experimental results on dry asphalt Brake pulses Time, s Brake Gain, N-m/Pa Wheel 4 Time, s Nominal Range

16 Road Friction Estimation Plot combinations of observed F x and λ Nonlinear least squares regression for the tyre curve Optimal slip point used for slip control Sample at 500 Hz, regress at 5 Hz µ = 0.9 change to 0.2 at 1.15 s; rough road Fx / Fz, Unitless x x x x x x x x Toss out x Estimated Slip Curve True Slip Curve First 100 Data Points Current 100 Data Points Keep first Large set difference Keep of points throughout stop Slip, unitless Slip, unitless Time, s

17 The Slip Control Problem λ d + Slip Actuator - Controller Dynamics λ P d T b Vehicle F x a x ω w a z ω y Road Friction Estimator Velocity Estimator

18 Conventional Actuators Air from valve at centre of vehicle Piston Pads Lever Disk Pneumatic delays reduce bandwidth and controllability New, high speed valves mounted directly on the brake chamber

19 Performance Benefits Prototype Valve Conventional ABS Pneumatic Time Delay, ms SMC Conflict Plot, Low Friction Road Stops 30% shorter Air Usage, kg Rising Gain x Baseline ABS Stop Uses 70% less air Stopping Distance, m

20 Novel Pneumatic Valve Magnet Pole Piece Coil Flexure

21 Prototype Valve

22 Conclusions 1. New sliding mode slip controller for a air- braked vehicles 2. New Fric<on and brake gain es<ma<on algorithms performed well in full- scale experiments 3. Novel actuator. providing tenfold reduc<ons in delay 4. Slip control combined with novel actua<on can reduce stopping distances by 30% and air usage by 70%

23 Questions/Comments? Employment Prospects? Jonathan Miller Gates Cambridge Scholar CV at:

24 Simulating the Vehicle ¼ Car Model x UMTRI Tyre Curves Z s Z u M s M u ω + Fx / Fz, Unitless Slip, unitless IRI / ISO 8608:1995 road roughness Validated chamber model

25 Simulating the Pneumatics Unsteady flow through an open system Piston hysteresis model Open Loop Charging Closed Loop Pressure Control Pressure, Bar Pressure, Bar Time, s Time, s

26 Brake Gain Estimation Simulated estimation run on a rough, wet road Speed, m/s Brake pulse Brake Gain, N-m/Pa True Value Time, s Time, s

27 Velocity Estimation a z a x ω g θ Wheels speeds corrupted during braking: need to estimate vehicle speed Measure longitudinal acceleration, vertical acceleration, and pitch rate Gravity significantly affects the sensor signals

28 Simulation Results Vehicle Speed Estimated Speed Speed, kph Kalman filter removes sensor biases using wheel speeds Dead reckoning by integration of inertial sensors during braking Time, s Accuracy: 0.2 m/s after 5 s braking