This content is not included in your SAE MOBILUS subscription, or you are not logged in.
Fault-Tolerant Filtering in Active Vehicle Suspensions
Published May 23, 2004 by Society of Automotive Engineers of Korea in South Korea
The problem considered in this paper is the design and analysis of a sensor fault-tolerant filtering method for active vehicle suspensions. As the suspension system of an automotive vehicle influences driving comfort and safety, a reliable operation of the system is required.
Components, sensors and actuators in physical systems are often subject to unexpected and not permitted deviations from acceptable/usual/standard conditions. These deviations are called faults. Faults can cause the loss of the overall performance of a physical system, which may present hazards to personnel or lead to unacceptable economic loss. The aim of the fault-tolerant control is to adjust or to modify on-line the nominal control laws in order to maintain the safety of the operators and the reliability of the process.
This paper aims at investigating the design of a sensor fault- tolerant filtering method mainly because: (1) sensors are critical components in almost all modern engineering system; (2) a sensor fault- tolerant filter can, in principle, indirectly solve the problem of sensor fault-tolerant control. In fact, from the control point of view, sensor fault-tolerant control does not require any modification of the nominal control law; the only requirement is that the "estimator" provides an accurate estimate of the system state after an instrument fault occurs.
In this paper, the attention is focused only on the design of the best possible estimator in case of single or multiple sensor fault. The information about the time occurrence and location of the sensor fault is supposed to be given by an external fault detection and identification (FDI) system. Starting from a physical mathematical model of the considered active vehicle suspension implemented in a quarter-car test rig, a Kalman filter has been designed for the fault- free case and its performance evaluated with test rig data. Then, an observability test for detecting critical sensor(s) has been performed. A critical sensor is a point of failure of the system, since when its failure occurs no estimation can be provided at all. Finally, the design of a bank of Kalman filters, one for each possible sensor failure configuration which is not critical, has been carried out and the variation of the level of accuracy of the estimation in case of failure has been analyzed. This information is useful for possible controller parameters reconfiguration. The final validation with test rig data showed good results.