Data-centric monitoring method
View Patent ↗Health management of machines and/or equipment, such as gas turbine engines, airplanes, and industrial equipment using a model centric method.
1. A data-centric monitoring method for detecting, isolating, and predicting abnormal conditions in a system comprising the steps of:
a) acquiring measured data relating to at least one part or piece of the system;
b) filtering bad or corrupt data;
c) deriving a baseline for each monitored variable;
d) calculating a residual from the baseline for each data point;
e) calculating a trend line based on the residuals of each monitored variable;
f) detecting data points whose residuals fall outside a normal operating limit for each monitored variable;
g) detecting data points whose residuals violate one or several rules for abnormal conditions for each monitored variable;
h) detecting any rapidly changing trend line slope or shape;
i) issuing one or multiple alerts or warnings for any violations of the above; and
j) consolidating multiple alerts or warning that correspond to the same cause into a single alert or warning of fault or faults;
k) estimating the severity of each fault;
l) estimating the effect of each fault on each system capability;
m) analyzing each system capability variation of trend;
n) extrapolating data along the trend line;
o) detecting when data points will be outside operating limits;
p) detecting when data points will violate abnormal condition rules;
q) issuing one or multiple alerts or warnings for any detected violations of steps “o” and “p”; and
r) consolidating multiple alerts or warning that correspond to the same cause into a single alert or warning;
s) estimating fault severity and system capability over a future time window of interest.
2. The method according to claim 1 wherein the step of filter bad or corrupt data comprises the step of reducing variability of data points by dividing or multiplying the data points by one or more correction factors.
3. The method according to claim 2 further comprising the steps of extrapolating along the trend line, detecting what and when data will be outside control limits or control rules, and issuing alerts or warnings for any variables that deviated from control limits or control rules.
4. The method according to claim 1 wherein the step of filter bad or corrupt data comprises the step of reducing variability of data points by referring the data points to a selected reference point.
5. The method according to claim 4 further comprising the steps of extrapolating along the trend line, detecting what and when data will be outside control limits or control rules, and issuing alerts or warnings for any variables that deviated from control limits or control rules.
6. The method according to claim 1 wherein the step of filter bad or corrupt data comprises the step of reducing variability of data points by transforming the data points through a functional mapping.
7. The method according to claim 6 further comprising the steps of extrapolating along the trend line, detecting what and when data will be outside control limits or control rules, and issuing alerts or warnings for any variables that deviated from control limits or control rules.
8. The method according to claim 1 wherein the step of filter bad or corrupt data comprises the step of reducing variability of data points by combining the data points from several monitored variables.
9. The method according to claim 8 further comprising the steps of extrapolating along the trend line, detecting what and when data will be outside control limits or control rules, and issuing alerts or warnings for any variables that deviated from control limits or control rules.
10. The method according to claim 1 wherein the step of filter bad or corrupt data comprises the step of removing data points that correspond to unacceptable operating conditions.
11. The method according to claim 10 further comprising the steps of extrapolating along the trend line, detecting what and when data will be outside control limits or control rules, and issuing alerts or warnings for any variables that deviated from control limits or control rules.
12. The method according to claim 1 wherein the step of filter bad or corrupt data comprises the step of removing data points that are corrupted by measurement noise.
13. The method according to claim 12 further comprising the steps of extrapolating along the trend line, detecting what and when data will be outside control limits or control rules, and issuing alerts or warnings for any variables that deviated from control limits or control rules.
14. The method according to claim 1 wherein the step of filter bad or corrupt data comprises the step of removing data points that have errors in transmission or conversion.
15. The method according to claim 14 further comprising the steps of extrapolating along the trend line, detecting what and when data will be outside control limits or control rules, and issuing alerts or warnings for any variables that deviated from control limits or control rules.
16. The method according to claim 1 wherein the step of filter bad or corrupt data comprises two or more of the following steps: reducing variability of data points by dividing or multiplying the data points by one or more correction factors, reducing variability of data points by referring the data points to a selected reference point, reducing variability of data points by transforming the data points through a functional mapping, reducing variability of data points by combining the data points from several monitored variables, removing data points that correspond to unacceptable operating conditions, removing data points that are corrupted by measurement noise, and removing data points that correspond to unacceptable operating conditions.
17. The method according to claim 16 further comprising the steps of extrapolating along the trend line, detecting what and when data will be outside control limits or control rules, and issuing alerts or warnings for any variables that deviated from control limits or control rules.
18. The method according to claim 1 further comprising the steps of extrapolating along the trend line, detecting what and when data will be outside control limits or control rules, and issuing alerts or warnings for any variables that deviated from control limits or control rules.
19. The method according to claim 18 further comprising the step of associating the alerts or warnings with root-cause faults or failure by matching the alerts with similar alerts from previously confirmed faults or matching the residual pattern with probable fault/failure patterns, where a “pattern” is a collection of the magnitude and the sign of the residuals of a subset (or the complete set) of monitored variables.
20. The method according to claim 1 further wherein the trend line is linear.
21. The method according to claim 1 wherein the trend line is a higher order line, such as a quadratic or cubic curve.