System and method for equipment remaining life estimation

• US 7,725,293 B2
• Filed: 12/07/2006
• Issued: 05/25/2010
• Est. Priority Date: 12/07/2006
• Status: Active Grant

• Current Assignee: General Electric Company
• Original Assignee: General Electric Company
• Inventors: Bonissone, Piero Patrone, Yan, Weizhong, Goebel, Kai Frank, Eklund, Neil Holger White, Xue, Feng, Varma, Anil
• Primary Examiner(s): Feliciano; Eliseo Ramos
• Assistant Examiner(s): Suarez; Felix E
• Application Number:
  US11/608,058
• Publication Number:
  US 20080140361A1
• Time in Patent Office: 1,265 Days
• Field of Search: 702/113, 702/117, 702/178, 702/182, 702/184, 702/185, 702/179, 702/180, 702/183, 324/727, 700/99, 704/245, 706/45
• US Class Current: 702/183
• CPC Class Codes: G05B 23/0283 Predictive maintenance, e.g. involving the monitoring of a system and, based on the monitoring results, taking decisions on the maintenance schedule of the monitored system; Estimating remaining useful life [RUL] preventive maintenance, i.e. planning maintenance according to the available resources without monitoring the system G06Q10/06
• CPC Class Codes: G06N 5/04 Inference or reasoning models

Abstract

A method to predict remaining life of a target is disclosed. The method includes receiving information regarding a behavior of the target, and identifying from a database at least one piece of equipment having similarities to the target. The method further includes retrieving from the database data prior to an end of the equipment useful life, the data having a relationship to the behavior, evaluating a similarity of the relationship, predicting the remaining life of the target based upon the similarity, and generating a signal corresponding to the predicted remaining equipment life.

Claims

1. A method to predict remaining life of a target equipment, the method comprising:

receiving target information regarding a behavior of the target equipment, said target information including at least a target operational history;

identifying from a database at least one piece of peer equipment having similarities to the target equipment based upon said behavior of said target equipment;

retrieving from the database peer data of said peer equipment having a relationship to the behavior of said target equipment, wherein said peer data includes at least peer operational data;

evaluating a similarity of the relationship between the peer equipment and the target equipment using said peer data and said target information;

creating at least one model based upon the similarity;

predicting the remaining life of the target equipment based upon the models using an evolutionary search of said model for at least one iteration; and

generating a signal corresponding to the predicted remaining life of the target equipment.

2. The method of claim 1, further comprising:

using an evolutionary algorithm processing for of at least one of the identifying, retrieving, and evaluating.

3. The method of claim 2, wherein the evolutionary algorithm processing comprises:

encoding the parameters of at least one of the identifying, retrieving, and evaluating;

building a fuzzy model to output at least one predicted life estimation;

creating a performance metric to output at least one fitness function result; and

for each unit, comparing said fitness function result with the predicted life estimation to produce a predictive accuracy.

4. The method of claim 3, wherein the comparing comprises:

defining said prediction accuracy measured as an absolute value of a prediction error.

5. The method of claim 1, wherein the receiving target information comprises:

receiving target information regarding the behavior of a turbine engine as the target equipment.

6. The method of claim 1, further comprising:

applying a function to define a range for an attribute of the behavior;

wherein the identifying comprises determining if an attribute of the peer data is within the defined range.

7. The method of claim 6, wherein the applying comprises applying the function:

${\mathrm{TGBF}}_i<br><br>\left(x_i;<br><br>a_i,b_i,c_i\right)=\left\{\begin{array}{cc}{\left[1+{<br><br>\frac{x_i-c_i}{a_i}<br><br>}^{2<br><br>b_i}\right]}^{-1}& {\mathrm{if}<br><br>\left[1+{<br><br>\frac{x_i-c_i}{a_i}<br><br>}^{2<br><br>b_i}\right]}^{-1}><br><br>\mathrm{\varepsilon <br><br>}\\ 0& \mathrm{otherwise}\end{array}\right\}$

wherein;

TGBF_i(x_i;

a_i,b_i,c_i) represents the range;

a_i represents a width parameter;

b_i represents a slope parameter;

c_i represents a value of the attribute of the observed behavior;

ε

represents a truncation parameter; and

x_i represents a value of the attribute of the behavior.

8. The method of claim 1, further comprising:

applying a function to define a similarity coefficient, S_i,j of an attribute of the peer data;

wherein the evaluating comprises using the similarity coefficient S_i,j.

9. The method of claim 8, wherein the applying comprises applying the function:

$S_{i,j}=\mathrm{TGBF}<br><br>\left(x_{i,j};<br><br>a_i,b_i,x_{i,Q}\right)=\left\{{\left[1+{<br><br>\frac{x_{i,Q}-x_{i,j}}{a_i}<br><br>}^{2<br><br>b_i}\right]}^{-1}\right\}$

wherein;

S_i,j=TGBF(x_i,j;

a_i,b_i,x_i,Q) represents the similarity coefficient;

a_i represents a width parameter;

b_i represents a slope parameter;

x_i,Q represents a value of an attribute of the behavior; and

x_i,j represents a value of the attribute of the operational data of the peer equipment.

10. The method of claim 9, wherein the peer data comprises a plurality of attributes, each attribute having the similarity coefficient S_i,j, the method further comprising:

defining an equipment similarity coefficient, S_j as a minimum of the plurality of similarity coefficients S_i,j.

11. The method of claim 10, wherein the defining the equipment similarity coefficient S_j comprises:

applying a weighting value to each similarity coefficient S_i,j of the plurality of similarity coefficients S_i,j.

12. The method of claim 11, further comprising:

defining at least one of the weighting value, the width parameter, and the slope parameter with an evolutionary algorithm.

13. The method of claim 10, wherein the predicting comprises:

obtaining a mathematical model of a remaining life estimate of the peer equipment; and

applying the equipment similarity coefficient S_j to the model to predict the remaining life of the target equipment.

14. The method of claim 13, wherein the predicting comprises solving an equation:

$y_Q=\frac{\underset{j=1}{\overset{m}{\sum <br><br>}}<br><br>S_j\times <br><br>y_j}{\underset{j=1}{\overset{m}{\sum <br><br>}}<br><br>S_j}$

wherein;

y_Q represents the predicted remaining life of the target equipment;

j represents an index relating to the peer equipment;

m represents a total number of peer equipment included;

S_j represents the equipment similarity coefficient; and

y_j represents the remaining life estimate of the peer equipment.

15. A program storage device readable by a computer, the device embodying a program or instructions executable by the computer to perform the method of claim 1.

16. A system to predict remaining useful life of a target equipment, the system comprising:

a database comprising peer data for at least one peer equipment and target data for said target equipment;

a processor in signal communication with the database;

a computational model application for executing on the processor, the computational model performing a method, comprising;

receiving information regarding a behavior of the target equipment, said behavior including at least operational history;

identifying from the database at least one piece of peer equipment having similarities to the target equipment;

retrieving from the database peer data having a relationship to the behavior;

evaluating a similarity of the relationship;

creating at least one model based upon the similarity; and

predicting the remaining useful life of the target equipment based upon the similarity using an evolutionary search of said model for at least one iteration;

wherein the processor is responsive to the computational model application to generate a signal corresponding to the predicted remaining useful life.

17. The system of claim 16, wherein the computational model application further performs:

using an evolutionary algorithm processing for at least one of the identifying, retrieving, and evaluating.

18. The system of claim 17, wherein the evolutionary algorithm processing comprises;

encoding the parameters of at least one of the identifying, retrieving, and evaluating;

building a fuzzy model to output at least one predicted life estimation;

creating a performance metric to output at least one fitness function result; and

for each unit, comparing said fitness function result with the predicted life estimation to produce a predictive accuracy.

19. The system of claim 18, wherein the computational model application performs creating the performance metric, the creating the performance metric comprising:

defining said prediction accuracy measured as an absolute value of a prediction error.

20. The system of claim 16, wherein:

the target equipment comprises a turbine engine.

21. The system of claim 16, wherein the computational model application further performs:

applying a function to define a range for an attribute of the behavior;

wherein the identifying comprises determining if an attribute of the peer data is within the defined range.

22. The system of claim 21, wherein the applying the function by the computational model application comprises applying the function:

${\mathrm{TGBF}}_i<br><br>\left(x_i;<br><br>a_i,b_i,c_i\right)=\left\{\begin{array}{cc}{\left[1+{<br><br>\frac{x_i-c_i}{a_i}<br><br>}^{2<br><br>b_i}\right]}^{-1}& {\mathrm{if}<br><br>\left[1+{<br><br>\frac{x_i-c_i}{a_i}<br><br>}^{2<br><br>b_i}\right]}^{-1}><br><br>\mathrm{\varepsilon <br><br>}\\ 0& \mathrm{otherwise}\end{array}\right\}$

wherein;

TGBF(x_i;

a_i,b_i,c_i) represents the range;

a_i represents a width parameter;

b_i represents a slope parameter;

c_i represents a value of the attribute of the observed behavior;

ε

represents a truncation parameter; and

x_i represents a value of the attribute of the behavior.

23. The system of claim 16, wherein the computational model application further performs:

applying a function to define a similarity coefficient, S_i,j of an attribute of the peer data;

wherein the evaluating comprises using the similarity coefficient S_i,j.

24. The system of claim 23, wherein the applying the function by the computational model application comprises applying the function:

$S_{i,j}=\mathrm{TGBF}<br><br>\left(x_{i,j};<br><br>a_i,b_i,x_{i,Q}\right)=\left\{{\left[1+{<br><br>\frac{x_{i,Q}-x_{i,j}}{a_i}<br><br>}^{2<br><br>b_i}\right]}^{-1}\right\}$

wherein;

S_i,j, TGBF(x_i,j;

a_i,b_i,x_i,Q) represents the similarity coefficient;

a_i represents a width parameter;

b_i represents a slope parameter;

x_i,Q represents a value of an attribute of the behavior; and

x_i,j represents a value of the attribute of the peer data.

25. The system of claim 23, wherein the peer data comprises a plurality of attributes, each attribute having the similarity coefficient S_i,j, wherein the computational model application further performs:

defining an equipment similarity coefficient, S_j as a minimum of the plurality of similarity coefficients S_i,j.

26. The system of claim 25, wherein the computational model application performs defining an equipment similarity coefficient S_j, the defining comprising:

applying a weighting value to each similarity coefficient S_i,j of the plurality of similarity coefficients S_i,j.

27. The system of claim 25, wherein the predicting by the computational model application comprises:

obtaining a mathematical model of a remaining life estimate of the peer equipment; and

applying the equipment similarity coefficient S_j to the model to predict the remaining life of the target equipment.

28. The system of claim 27, wherein the predicting by the computational model application comprises solving an equation:

$y_Q=\frac{\underset{j=1}{\overset{m}{\sum <br><br>}}<br><br>S_j\times <br><br>y_j}{\underset{j=1}{\overset{m}{\sum <br><br>}}<br><br>S_j}$

wherein;

y_Q represents the predicted remaining life of the target equipment;

j represents an index relating to the peer equipment;

m represents a total number of peer equipment included;

S_j represents the equipment similarity coefficient; and

y_j represents the remaining life estimate of the peer equipment.