Unified approach to detection and isolation of parametric faults using a kalman filter residual-based approach

• US 9,002,678 B1
• Filed: 01/10/2014
• Issued: 04/07/2015
• Est. Priority Date: 01/10/2014
• Status: Active Grant

• Current Assignee: King Fahd University Of Petroleum And Minerals
• Original Assignee: King Fahd University Of Petroleum And Minerals
• Inventors: Cheded, Lahouari, Doraiswami, Rajamani
• Primary Examiner(s): Teixeira Moffat, Jonathan C
• Assistant Examiner(s): BETSCH, REGIS J
• Application Number:
  US14/152,064
• Time in Patent Office: 452 Days
• Field of Search: 702/183
• US Class Current: 702/183
• CPC Class Codes: G05B 23/0254 based on a quantitative model, e.g. mathematical relationships between inputs and outputs; functions: observer, Kalman filter, residual calculation, Neural Networks
• CPC Class Codes: G05B 23/0281 Quantitative, e.g. mathematical distance; Clustering; Neural networks; Statistical analysis
• CPC Class Codes: G06F 17/18 for evaluating statistical data , e.g. average values, frequency distributions, probability functions, regression analysis forecasting specially adapted for a specific administrative, business or logistic context G06Q10/04

Abstract

A unified method to detection and isolation of parametric faults in a physical system resulting from variations in the parameters of its constituting subsystems, termed herein as diagnostic parameters, uses Kalman filter residuals. Rather than using the feature vector made of the coefficients of the numerator and denominator of the system transfer function, which is known to be a non-linear function of the diagnostic parameter variations, the method first shows and then exploits, for fault detection purposes, the fact that the Kalman filter residual is a multi-linear function of the deviations in the diagnostic parameters, i.e. the residual is separately linear in each parameter. A fault is then isolated using a Bayesian multiple composite hypotheses testing approach. A reliable map relating the diagnostic parameters to the residual is obtained off-line using fault emulators. The unified fault detection and isolation method is successfully evaluated on both simulated data as well as on real data obtained from a benchmarked laboratory-scale coupled-tank system used to exemplify an industrial two-tank process.

Claims

1. A computer-implemented method of fault isolation and detection in a physical fluid control system comprising a controller, a pump having a pump flow, a first tank having a leakage flow and a second tank having an output flow, wherein the controller is in operable connection with the pump and the pump is in fluid connection with the first tank by a first flow, and the first tank is in fluid connection with the second tank, comprising:

carrying out unified detection and isolation of multiplicative faults of the fluid flows and control of the fluid control system with factors saved in a first memory of a computer and Kalman filtering residuals saved in a second memory of the computer;

mapping the Kalman filter residuals with diagnostic parameters; and

estimating, in the processor of the computer, influence vectors for the leakage flow, an actuator corresponding with the pump flow and sensor faults corresponding with the output flow by perturbing the diagnostic parameters one at a time during an off-line identification phase,

isolating, in a processor of the computer, the multiplicative faults with composite Bayesian hypothesis testing taking into account statistics of noise and disturbances affecting the Kalman filter residuals based on the square of the cosine of angle between the residual of the diagnostic parameters and corresponding estimated influences.

2. The computer-implemented method of claim 1, wherein the unified detection and isolation of multiplicative faults is performed in a simulated system with various signal to noise ratios and a physical system.

3. The computer-implemented method of claim 1, further comprising:

mapping the Kalman filter residuals and the diagnostic parameters to isolate a system fault.

4. The computer-implemented method of claim 1, wherein fault detection is carried out with a Kalman filter to isolate the multiplicative system faults.

5. The computer-implemented method of claim 1, further comprising applying a linear regression model of the Kalman filtering residuals to diagnostic parameter variations.

6. The computer-implemented method of claim 1, further comprising detecting the multiplicative system faults with the Kalman filter residuals as a multi-linear function of deviations in the diagnostic parameters.

7. The computer-implemented method of claim 1, wherein the Kalman filtering residuals are applied linearly separately in each of the diagnostic parameters.

8. The computer-implemented method of claim 1, wherein the mapping is carried out with a fault emulator.

9. The method of claim 1, wherein

a Bayes decision strategy of

10. The computer-implemented method of claim 1, further comprising providing an estimated fault size by a condition-based maintenance program operating on the computer.

11. A non-transitory computer readable medium having instructions stored therein that when executed by one or more processors cause the one or more processors to perform a method of unified detection and isolation of system faults in a physical fluid control system comprising a controller, a pump having a pump flow, a first tank having a leakage flow and a second tank having an output flow, wherein the controller is in operable connection with the pump and the pump is in fluid connection with the first tank by a first flow, and the first tank is in fluid connection with the second tank, the method comprising:

carrying out unified detection and isolation of multiplicative faults of the fluid flows and control of the fluid control system with factors saved in a first memory of a computer and Kalman filtering residuals saved in a second memory of the computer;

mapping the Kalman filter residuals with diagnostic parameters; and

isolating, in a processor of the computer, system faults with composite Bayesian hypothesis testing taking into account statistics of noise and disturbances affecting the Kalman filter residuals,

estimating, in the processor of the computer, influence vectors for the leakage flow, an actuator corresponding with the pump flow and sensor faults corresponding with the output flow by perturbing the diagnostic parameters one at a time during an off-line identification phase;

isolating, in the processor of the computer, the multiplicative faults with composite Bayesian hypothesis testing taking into account statistics of noise and disturbances affecting the Kalman filter residuals based on the square of the cosine of angle between the residual of the diagnostic parameters and corresponding estimated influences.

12. The computer readable medium of claim 11, wherein the processors further perform the unified detection and isolation of multiplicative faults is performed in a simulated system with various signal to noise ratios and a physical system.

13. The computer readable medium of claim 11, wherein the processors map the Kalman filter residuals system faults and the diagnostic parameters to isolate a system fault.

14. The computer readable medium of claim 11, wherein the processors further perform fault detection with a Kalman filter to isolate the multiplicative system faults.

15. The computer readable medium of claim 11, wherein the processors further apply a linear regression model of the Kalman filtering residuals to diagnostic parameter variations.

16. The computer readable medium of claim 11, wherein the processors further perform detecting the multiplicative system faults with the Kalman filter residuals as a multi-linear function of deviations in the diagnostic parameters.

17. The computer readable medium of claim 11, wherein the processors further perform applying the Kalman filter residuals linearly and separately in each parameter.

18. The computer readable medium of claim 11, wherein the processors further perform the mapping with fault emulators.

19. The computer readable medium of claim 11, wherein

the isolating is carried out with a Bayes decision strategy of

20. The computer readable medium of claim 11, wherein the processors further provide an estimated fault size by a condition-based maintenance program.