Date of Award
Master of Science (MS)
Law, E. Harry
Wagner , John R
Ayalew , Beshah
This project explored the feasibility of using measured responses of a passenger car together with a fuzzy logic based control algorithm to sense the onset of under-steer (or loss of steering control) and mitigate or eliminate it. The controller is simple and robust and, unlike existing controllers, instead of comparing the vehicle response to that of an idealized model it makes decisions based solely upon the measured response of the car.
Simulations were conducted (using CarSim) of various vehicles executing the skid pad and the double lane change tests to characterize the vehicle behavior. Consistent and qualitatively similar patterns in vehicle response during the inception of and at limit under-steer were observed. A fuzzy logic routine was developed that analyzes real-time measurements of steering wheel angle (SWA) and lateral acceleration (Ay). Based on the relative `trends' of the signals, the control algorithm decides upon the presence and extent of under-steer in the vehicle. The degree of under-steer then defines the corrective action.
The fundamental concept is to measure a drop in the instantaneous lateral acceleration gain, i.e., Ay/SWA, indicating a lack of response. It is quantified as a normalized error and transformed into an under-steer number between 0 and 10 using a pair of fuzzy inference systems. Once incipient under-steer is detected, the brakes and engine throttle are managed to limit the lateral deviation from the travel lane. The controller also senses vehicle velocity, master cylinder brake pressure and normalized throttle input to improve controllability. This approach eliminates the need for either a simple or a complex vehicle model and the associated dependence on the model parameters.
Controller performance was validated using a braking-in-turn maneuver developed by the author and the standard double lane change maneuver. The results have shown clear improvement in the tracking ability of a vehicle. The simulations were conducted at different speeds with each of several vehicles and with different tire-to-ground friction values without any changes to the control algorithm. This has shown that the controller is robust across different conditions. The controller is successful in increasing the maximum safe speed for a negotiating a curve for all vehicles on various road conditions.
The last part of the controller was to combine it with an existing over-steer controller, developed at Clemson University, which also uses fuzzy logic. This was successfully completed to obtain a fully functional ESC system, independent of a vehicle model. Future work will include tuning the controller based on track data from real vehicles.
Pandit, Chinmay, "A Model-free Approach to Vehicle Stability Control" (2013). All Theses. 1637.