Noisy monitor detection and intermittent fault isolation

• US 8,386,849 B2
• Filed: 01/29/2010
• Issued: 02/26/2013
• Est. Priority Date: 01/29/2010
• Status: Expired due to Fees

• Current Assignee: Honeywell International Inc.
• Original Assignee: Honeywell International Inc.
• Inventors: Shao, Qingqiu Ginger, Magnuson, Randy R., Miller, Dave, Barton, Bradley John, Kolbet, David Michael
• Primary Examiner(s): TRUONG, LOAN
• Application Number:
  US12/696,877
• Publication Number:
  US 20110191635A1
• Time in Patent Office: 1,124 Days
• Field of Search: 714/25, 714/37, 714/48
• US Class Current: 714/25
• CPC Class Codes: G06F 11/0706 the processing taking place on a specific hardware platform or in a specific software environment
• CPC Class Codes: G06F 11/0751 Error or fault detection not based on redundancy power supply failures G06F1/30; network fault management H04L41/06

Abstract

A method of detecting and diagnosing system faults, includes detecting the noisy status of a monitor during operations and incorporating a quantified monitor uncertainty level to support fault isolation reasoning. A sequential probability ratio test is used to statistically test the noisy status of a monitor and Shannon'"'"'s entropy theory is used to quantify the uncertainty levels of the monitor to support the use of the monitor values in fault isolation.

Claims

1. A method of detecting and diagnosing system faults, comprising

a) collecting data;

b) performing a first assessment of a monitor of the data to determine whether the monitor is noisy;

c) if the monitor is not noisy, determining fault isolation;

d) if it cannot be determined that the monitor is noisy, collect additional data;

e) performing a second assessment whether the monitor is noisy or there is an intermittent fault;

f) if the second assessment cannot be performed, determining uncertainties.

2. The method of claim 1 wherein the determining step comprises:

using Shannon'"'"'s entropy theory.

3. The method of claim 1 wherein the steps b-e comprise:

using the sequential probability ratio statistical test.

4. The method of claim 3 wherein the determining step comprises:

using Shannon'"'"'s entropy theory.

5. The method of claim 1 further comprising:

converting the detection of the noisy status to a diagnostic monitor.

6. The method of claim 1 wherein steps b-e comprise:

distinguishing between a spurious fault due to noisy monitors, a transient fault due to unpredictable conditions, and an intermittent fault.

7. A method of detecting and diagnosing system faults, comprising:

a) continuously receiving a monitor values input (M=0 or

     1) at the monitor;

b) if M=0 and N_m=0 are satisfied, repeating step a);

c) if M=0 and N_m=0 are not satisfied, but M=1 and N_m=0 are satisfied;

d) initializing a database by setting N_m=1, N_s=1, M=1, set the monitor noisy status of a monitor to “

undecided”

, and update the monitor state and the monitor noise status;

e) sending a monitor updated notification to a failure mode isolation reasoner; and

f) returning to step a);

g) if neither (M=0 and N_m=0) nor (M=1 and N_m=0)are satisfied, updating the database by setting N_m=N_m+1 and, if N=1, setting N_s=N_s+1;

h) if N_s≦

r_low_m satisfied;

i) update the database to monitor “

not noisy, M=0”

;

j) send a monitor updated notification to the failure mode isolation reasoner; and

k) updating the database by N_m=0 and N_s=0;

l) if N_s≦

r_low_m is not satisfied and N_S≧

r_high_m is satisfied;

m) update the database to monitor “

not noisy, M=1”

; and

n) send a monitor updated notification to the failure mode isolation reasoner; and

o) updating the database by N_m=0 and N_s=0;

p) if neither N_s≦

r_low_m and N_s≧

r_high_m are satisfied, apply Shannon'"'"'s Entropy theory by;

q) calculating the uncertainty p=N_s/N_m, H(p)=−

p log(p,2)−

(1−

p)log((1−

p),

     2);

r) updating the database for a dynamic value of d=d*(1−

H(p)) and f=f*(1+H(p));

s) if N_m>

N_mT and H(p)<

0.9 is not satisfied or the monitor is noisy, repeat from a);

t) N_m>

N_mT and H(p)<

0.9 is satisfied and the monitor is not noisy, gather additional information including the steps of;

u) determining if there are any failure modes that only have this monitor for detection;

v) if step u) is not satisfied and if there are other monitors associated with the same failure modes to the current monitor that are noisy;

w) update the database by setting the monitor to noisy;

x) sending a monitor updated notification to a failure mode isolation reasoner; and

y) repeating from step a);

z) if step u) is satisfied;

aa) sending an alert to the user signifying there is a potential noisy sensor or a system intermittent fault; and

bb)) setting the monitor to noisy;

cc) sending a monitor updated notification to a failure mode isolation reasoner; and

dd) repeating from step a);

ee) in response to the monitor updated notification to the failure mode isolation reasoner of steps e), j), n), x), and cc), and if the monitor is not noisy and if any new failure modes are associated with the monitor;

ff) updating a failure mode coverage; and

gg) updating the failure mode isolation reasoning by excluding noisy monitors from being considered;

hh) in response to the monitor updated notification to the failure mode isolation reasoner of steps e), j), n), x), and cc), and if the monitor is not noisy and if no new failure modes are associated with the monitor;

ii) updating the failure mode isolation reasoning by excluding noisy monitors from being considered;

jj) in response to the monitor updated notification to the failure mode isolation reasoner of steps e), j), n), x), and cc), and if the monitor is noisy;

kk) updating the failure mode isolation reasoning by excluding noisy monitors from being considered.