Date of Thesis

9-16-2015

Thesis Type

Masters Thesis (Bucknell Access Only)

Degree Type

Master of Science in Mechanical Engineering

First Advisor

Craig Beal

Abstract

The improvement of data collection and computing capabilities on modern vehicles has made possible the creation of innovative vehicle control systems. These systems improve vehicle handling and general passenger safety, reducing the number of casualties and cost related to passenger vehicle crashes. Such systems utilize the measurements produced by sensors in modern vehicles that provide meaningful information about the state of different components of the vehicle, such as traction torque and rotational velocity of the individual tires. Although this information has proven useful, significant information about the vehicle's interaction with the road and its operating conditions, such as the maximum lateral force that can be produced by the tires, is still largely unavailable. Stability control systems help stabilize passenger vehicles by controlling the braking of the individual tires, which changes the forces they produce. Because these systems are constrained by the adhesion limit given by the friction coefficient, knowledge of the road conditions is an important factor in their performance. This project analyzes an estimation structure which allows the estimation of real time tire friction coefficients with the road using measurements of longitudinal, lateral, and rotational velocities and steering torque applied on the wheels. More specifically, the contributions of the work presented in this thesis are: An investigation of the theory behind the existing algorithms to estimate friction coefficient values is conducted. A friction estimating algorithm based on pneumatic trail measurements introduced by Hsu, Laws, and Herdes is re-based on a pneumatic trail model introduced by Ren and Guo, improving the estimation accuracy at low slip angles. An analysis of the sensitivity of the algorithm to the expected noise involved in the measurement of the input variables is presented. A data filtering process based on a Kalman Filter design is introduced. This filter utilizes the calculated uncertainty of the estimated friction values and multiple redundant estimations from each tire of the vehicle to increase the accuracy and performance of the proposed friction estimation method. A computational algorithm that recognizes "real" changes in the estimated friction coefficients, which is based in a Limit Checking Detector algorithm, is also presented. It is important to note that the actual friction coefficient of passenger vehicle tires might rapidly fluctuate as the vehicle moves. The main purpose of this change detection algorithm is to identify sudden significant changes in the average friction coefficient, limiting the influence of rapid erratic fluctuations on the real-time data being displayed. The implementation of these algorithms has the potential of being a significant advance in the automobile safety system field, as it would provide a more accurate method to obtain important information about the vehicle's interaction with the road compared to other proposed methods. Moreover, this algorithm utilizes instrumentation already available on regular passenger vehicles, reducing its implementation cost. The resultant estimation can be displayed to the user and work as an input to other stability control systems currently being developed.

Share

COinS