Sensor fault detection, isolation and accommodation

Details Sensor fault detection, isolation and accommodation Sensor fault detection, isolation and accommodation

2000-08-21

2003-07-22

Ton, David (Department: 2133)

Error detection/correction and fault detection/recovery

Pulse or data error handling

Digital logic testing

C714S025000, C361S088000, C361S093100

Reexamination Certificate

active

06598195

ABSTRACT:

BACKGROUND OF THE INVENTION
The present invention relates to sensor fault detection, isolation and accommodation for all types of engineering systems.
Most engineering systems use sensors to control and monitor the operation of the system. In the case of gas turbine engines, these sensors are used to measure process variables such as rotor speeds, temperatures, pressures, and actuator position feedbacks. The measured variable is then used to ensure that the system is being operated at the desired condition, that safety bounds are being observed, and that performance is being optimized.
Although sensors can be designed to be robust, sensor failure has been addressed by using redundant sensors and backup schemes. More recently, with the advent of digital controllers, analytical schemes for sensor failure detection, isolation, and accommodation (FDIA) have been developed. However, most sensor FDIA schemes are limited to simple range and rate tests. In such tests, sensor values are compared to expected minimum and maximum values and/or rates of change of values. The sensor is declared as faulted if it exceeds its limits. Such methods work for large failures that are very rapid, i.e., “hard” failures. However, in-range failures and slow drift failures, i.e., “soft” failures, are not addressed by such methods.
More sophisticated schemes use an analytical “model” of the sensor, which involves estimating the sensor values based on other inputs, usually other sensors or operating conditions. One such model is a “map model”, in which a sensor model of the form S
c
=f(S
a
,S
b
) is used. That is, the value of sensor “c” is assumed to be some reasonably simple function of sensors “a” and “b” . Depending on the application, each sensor model can be a function of one or more other sensors. For example, compressor inlet temperature can be modeled as a function of sensed fan inlet temperature and fan speed. Then, a sensor is declared as faulted whenever the difference between the sensor value and its model value exceeds a predetermined threshold.
Regardless of the technique used, the choice of a threshold is meaningful. Too tight a threshold leads to a large number of false alarms (false positives), whereas too large a threshold leads to fewer faults being detected (false negatives). Also, it is generally understood that detecting a fault is easier than isolating it to a specific sensor. Detecting a fault but ascribing it to the incorrect sensor is misclassification.
It would be desirable, then, to provide a method for detection, isolation and accommodation of sensor faults that is aimed at reducing the detection threshold as compared to current methods. It would further be desirable to achieve sensor FDIA while maintaining low rates of false positives, false negatives and misclassifications.
BRIEF SUMMARY OF THE INVENTION
To detect, isolate and accommodate sensor faults, a method is proposed that is based on the use of sensor-consistency models, hypothesis testing, and maximum-wins strategy. This method maximizes the number of correct isolations and minimizes the number of false positives.
Accordingly, the present invention provides a method for sensor fault detection, isolation and accommodation with a reduced detection threshold.


REFERENCES:
patent: 5436826 (1995-07-01), O'Flarity
patent: 5469364 (1995-11-01), Hughey et al.
patent: 5864773 (1999-01-01), Barna et al.
patent: 6208917 (2001-03-01), McKissick, Jr. et al.
patent: 6253130 (2001-06-01), Mergenthaler et al.

Affiliated with

Also associated with

Rating

Sensor fault detection, isolation and accommodation does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Sensor fault detection, isolation and accommodation, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Sensor fault detection, isolation and accommodation will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-3088455