Method and a system for merging health indicators of a device

• US 9,607,451 B2
• Filed: 07/08/2015
• Issued: 03/28/2017
• Est. Priority Date: 07/10/2014
• Status: Active Grant

• Current Assignee: Airbus Helicopters
• Original Assignee: Airbus Helicopters
• Inventors: Mouterde, Joel
• Primary Examiner(s): Tarcza, Thomas
• Assistant Examiner(s): BERNS, MICHAEL ANDREW
• Application Number:
  US14/793,957
• Publication Number:
  US 20160012651A1
• Time in Patent Office: 629 Days
• US Class Current: 1/1
• CPC Class Codes: G05B 23/0235 based on a comparison with predetermined threshold or range, e.g. "classical methods", carried out during normal operation; threshold adaptation or choice; when or how to compare with the threshold
• CPC Class Codes: G05B 23/024 Quantitative history assessment, e.g. mathematical relationships between available data; Functions therefor; Principal component analysis [PCA]; Partial least square [PLS]; Statistical classifiers, e.g. Bayesian networks, linear regression or correlation analysis; Neural networks
• CPC Class Codes: G07C 5/08 Registering or indicating performance data other than driving, working, idle, or waiting time, with or without registering driving, working, idle or waiting time

Abstract

A method and system for merging health indicators of a device. During the method, health indicators Ind_i are firstly distributed in at least one indicator group G, each health indicator Ind_i being associated with a respective detection threshold Sd_i, it being possible to determine a characteristic operating point of the device by using the health indicators Ind_i of an indicator group G. An alarm threshold Sa is determined using training health indicators. Then, a merged health indicator MHI is determined by taking into account both of the alarm threshold Sa and also of the detection thresholds Sd_i of the health indicators Ind_i of the group G. Finally, the merged health indicator MHI is compared to the alarm threshold Sa and it is deduced whether or not the device presents a fault or a risk of such a fault appearing.

Claims

1. A method for monitoring the operation of a device, wherein measurement sensors are positioned on the device and/or in close proximity to the device to measure parameters of the device while the device is in operation, measurements of at least one of the sensors making it possible to determine at least one health indicator Ind_i, each health indicator Ind_i being associated with a detection threshold Sd_i, the method comprising:

during a preliminary step, distributing the health indicators Ind_i of the device in at least one indicator group G and applying an algorithm g to the health indicators Ind_i of an indicator group G in order to determine a characteristic operating point Pf of the device for the indicator group G;

during a training step when the device is considered healthy, operating the device during a training period, measuring the parameters of the device using the sensors while the device is operating during the training period, using the measurements of the sensors to determine training health indicators of the indicator group G, determining training operating points of the device by applying the algorithm g to the training health indicators, and determining a healthy-state model Mr for the device using the training operating points, a belonging function f indicating whether a characteristic operating point Pf of the device belongs to the healthy-state model Mr representing the device in a healthy state for the indicator group G using the training operating points, and an alarm threshold Sa associated with the belonging function f based on the training operating points;

during a step of taking account of the detection thresholds Sd_i, determining a transformation T of the belonging function f so as to take into account both of the alarm threshold Sa of the healthy-state model Mr and the detection thresholds Sd_i of the health indicators Ind_i of the group G; and

during a monitoring step the following steps are performed;

operating the device;

measuring the parameters of the device using the sensors while the device is operating;

using the measurements of the sensors to determine the health indicators Ind_i of an indicator group G;

determining a merged health indicator MHI by applying the transformation T to the health indicators Ind_i of an indicator group G;

comparing the merged health indicator MHI with the alarm threshold Sa;

deducing an operating state of the device based on the comparison; and

providing an alarm to an operator upon the operating state of the device being deduced from the comparison to be faulty.

2. The method according to claim 1, wherein:

the alarm threshold Sa is equal to a belonging criterion of a limit operating point located on a boundary of the healthy-state model Mr, the belonging criterion being determined by applying the belonging function f to the health indicators Ind_i corresponding to the limit characteristic operating point for the indicator group G.

3. The method according to claim 1, wherein, in order to determine a health indicator Ind_i of the device, firstly, at least one measurement is acquired from the sensors, and then the health indicator Ind_i is calculated as a function of the measurements.

4. The method according to claim 1, wherein the transformation T of the belonging function f is continuous when the belonging function f is continuous.

5. The method according to claim 1, wherein each partial derivative of order q of the transformation T is bounded over its definition domain, q being a positive integer greater than or equal to 1.

6. The method according to claim 1, wherein if the belonging function f is of class C^q, then the transformation T is at least of class C^m, m, and q being positive integers greater than or equal to 0 with m less than or equal to q, a function being of class C^q when the function can be differentiated q times and its derivatives are continuous.

7. The method according to claim 1, wherein the transformation T is determined by minimizing a difference between the transformation T as applied to the belonging function f and the belonging function f.

8. The method according to claim 1, wherein the healthy-state model for the device is constituted by a multidimensional Gaussian function.

9. The method according to claim 1, wherein the healthy-state model for the device is constituted by a superposition of multidimensional Gaussian functions.

10. The method according to claim 9, wherein each multidimensional Gaussian function is associated with a weighting factor.

11. The method according to claim 1, wherein the belonging function f is a probability density function of the healthy-state model Mr for the device suitable for characterizing the position of the characteristic operating point Pf of the device relative to a boundary of the healthy-state model.

12. The method according to claim 1, wherein the belonging function f is a distance between the characteristic operating point Pf and the center of the healthy-state model Mr for the device.

13. The method according to claim 12, wherein the distance is determined by the Mahalanobis distance.

14. The method according to claim 12, wherein for any characteristic operating point Pf situated on a straight line passing through the center of the healthy-state model Mr and a limit characteristic operating point situated on the boundary of the healthy-state model Mr that is equal to the alarm threshold Sa, if the belonging criterion corresponding to each detection threshold Sd_i is greater than the alarm threshold Sa, the merged health indicator MHI is equal to the belonging criterion, whereas if the belonging criterion corresponding to at least one detection threshold Sd_i is less than or equal to the alarm threshold Sa, the merged health indicator MHI is proportional to the belonging criterion, a multiplier coefficient being applied to the belonging criterion, which coefficient is equal to the ratio between the alarm threshold Sa and a belonging criterion corresponding to the detection threshold Sd_i.

15. The method according to claim 1, wherein, for a characteristic operating point Pf having the belonging criterion less than or equal to the alarm threshold Sa and the belonging criteria corresponding to the detection thresholds Sd_i in a neighborhood of the characteristic operating point Pf are greater than the alarm threshold Sa, the merged health indicator MHI is equal to the belonging criterion, whereas, if the belonging criterion corresponding to at least one of the detection thresholds Sd_i in the neighborhood of the characteristic operating point Pf is less than or equal to the alarm threshold Sa, the merged health indicator MHI is proportional to the belonging criterion, a multiplier coefficient being applied to the belonging criterion, which coefficient is equal to the ratio between the alarm threshold Sa and a belonging criterion corresponding to the detection threshold Sd_i.

16. The method of claim 1 wherein:

the device is part of a rotary wing aircraft and is used for operating the aircraft in flight; and

operating the device during the monitoring step includes operating the aircraft in flight.

17. The method of claim 16 wherein:

the device includes at least one of an engine, a transmission, and a rotor of the aircraft.

18. The method of claim 1 wherein:

providing an alarm includes providing an alarm to a pilot of the aircraft upon the operating state of the device being deduced from the comparison to be faulty.

19. A system for monitoring the operation of a device, wherein measuring sensors are positioned on the device and/or in close proximity to the device, measurements of at least one of the sensors making it possible to determine at least one health indicator Ind_i of the device, each health indicator Ind_i being associated with a detection threshold Sd_i, the system comprising:

centralization means provided with at least one central computer and at least one main memory;

an acquisition system for acquiring measurements from the sensors; and

at least one transfer means for transfer between the acquisition system and the centralization means;

wherein the merging system is capable of implementing the method according to claim 1.

20. A rotary wing aircraft comprising:

a device used for operating the aircraft in flight;

a plurality of measurement sensors positioned on the device and/or in close proximity to the device to measure parameters of the device while the aircraft is in flight, measurements of at least one of the sensors making it possible to determine at least one health indicator Ind_i, each health indicator Ind_i being associated with a detection threshold Sd_i; and

a computer system configured to;

during a preliminary step, distribute the health indicators Ind_i of the device in at least one indicator group G and apply an algorithm g to the health indicators Ind_i of an indicator group G in order to determine a characteristic operating point Pf of the device for the indicator group G;

during a training step when the device is considered healthy, measure the parameters of the device using the sensors while the aircraft is in flight during the training period, use the measurements of the sensors to determine training health indicators of the indicator group G, determine training operating points of the device by applying the algorithm g to the training health indicators, and determine a healthy-state model Mr for the device using the training operating points, a belonging function f indicating whether a characteristic operating point Pf of the device belongs to the healthy-state model Mr representing the device in a healthy state for the indicator group G using the training operating points, and an alarm threshold Sa associated with the belonging function f based on the training operating points;

during a step of taking account of the detection thresholds Sd_i, determine a transformation T of the belonging function f so as to take into account both of the alarm threshold Sa of the healthy-state model Mr and the detection thresholds Sd_i of the health indicators Ind_i of the group G; and

during a monitoring step;

measure the parameters of the device using the sensors while the aircraft is in flight;

use the measurements of the sensors to determine the health indicators Ind_i of an indicator group G;

determine a merged health indicator MHI by applying the transformation T to the health indicators Ind_i of an indicator group G;

compare the merged health indicator MHI with the alarm threshold Sa;

deduce an operating state of the device based on the comparison; and

provide an alarm to a pilot of the aircraft upon the operating state of the device being deduced from the comparison to be faulty.