Nonlinear Robust Control of Torque Converter Clutch Slip System for Passenger Vehicles Using Advanced Torque Estimation Algorithms

In this paper, a torque-estimation-based robust controller for passenger car torque converter clutch slip system is presented. The proposed robust controller uses only the measurements available from inexpensive sensors that are installed in current passenger vehicles for torque estimation and feedback control. A conventional full state observer along with a neural-network-based open-loop hydraulic actuator observer is designed to estimate the unknown driving load, and a neural-network-based turbine torque estimator considering the temperature of oil circulating the torque converter is developed for improved turbine torque estimation accuracy. The stability of the internal dynamics is also investigated, and the performance and robustness of the robust controller is validated by simulation studies.     

Model-based fault detection and isolation in automotive yaw moment control system

This paper presents a model-based fault detection and isolation technique for automotive yaw moment control system. For this purpose, a novel fault detection and isolation algorithm for a class of actuator-plant systems is proposed. Compared with the existing fault detection and isolation techniques that can only isolate a target fault or require multiple observers to isolate multiple faults, a unique strength of the proposed algorithm is its ability to isolate faults at the component level solely based on the residuals generated by a single observer. The validity of the proposed algorithm, applied to automotive yaw moment control system, is investigated via a simulation study based on a realistic vehicle dynamics model. The results suggest that the proposed algorithm can isolate the component subject to fault while effectively handling two perennial nuisances: sensitivity to disturbances and false alarms due to uncertainties.     

Bayesian vehicle class recognition using 3-d probe

A new approach is presented to vehicle-class recognition from video clips. Two new concepts introduced are: probes consisting of local 3-d curve-groups which when projected into video frames are features for recognizing vehicle classes in video clips; and Bayesian recognition based on class probability densities for groups of 3-d distances between pairs of 3-d probes. A full Bayesian recognizer is realized via Monte Carlo simulation method. Also, a sub-optimal but robust camera calibration method is employed and tested extensively.