Generating cumulative wear-based indicators for vehicular components

• US 9,530,256 B2
• Filed: 12/22/2015
• Issued: 12/27/2016
• Est. Priority Date: 09/26/2014
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Ahn, Hyung-il, Denesuk, Matthew, Hochstein, Axel, Leung, Ying Tat
• Primary Examiner(s): Jen, Ian
• Application Number:
  US14/978,438
• Publication Number:
  US 20160110933A1
• Time in Patent Office: 371 Days
• Field of Search: 701/36
• US Class Current: 1/1
• CPC Class Codes: G05B 23/00 Testing or monitoring of control systems or parts thereof monitoring of programme-control systems G05B19/048, G05B19/406
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 7/01 Probabilistic graphical models, e.g. probabilistic networks
• CPC Class Codes: G07C 5/006 Indicating maintenance
• CPC Class Codes: G07C 5/008 communicating information to a remotely located station transmission systems for measured values G08C
• CPC Class Codes: G07C 5/02 Registering or indicating driving, working, idle, or waiting time only apparatus forming part of taximeters G07B13/00
• 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
• CPC Class Codes: G07C 5/0808 Diagnosing performance data testing of vehicles G01M17/00; testing of electrical installation on vehicles G01R31/005

Abstract

Methods, systems, and computer program products for generating wear-based indicators are provided herein. A method includes assigning a failure label to each data point associated with a component associated with a failure-caused component replacement within a pre-specified number of runtime hours of the failure-caused component replacement; assigning a non-failure label to each data point associated with a failure-caused component replacement and not within the pre-specified number of runtime hours; assigning a non-failure label to each data point associated with a scheduled component replacement; assigning a non-failure label to each data point associated with an actively running instance of the component as yet to be replaced; estimating a failure probability for the component over a pre-specified future runtime; determining a cumulative hazard function for the component based on the failure probability; and generating a cumulative wear-based indicator for the component by executing a regression function based on the cumulative hazard function.

Claims

1. A method comprising the following steps:

assigning a failure class label to each data point, from a set of multiple data points derived from measurements associated with a vehicular component across a fleet of multiple vehicles, that (a) is associated with a failure-caused vehicular component replacement, and (b) is within a pre-specified number of runtime hours of the failure-caused vehicular component replacement;

assigning a non-failure class label to each data point, from the set of the multiple data points, that (a) is associated with a failure-caused vehicular component replacement, and (b) is not within the pre-specified number of runtime hours of the failure-caused vehicular component replacement;

assigning a non-failure class label to each data point, from the set of the multiple data points, that is associated with a scheduled vehicular component replacement;

assigning a non-failure class label to each data point, from the set of the multiple data points, that is associated with an actively running instance of the vehicular component as yet to be replaced;

estimating a failure probability for the vehicular component at each of the multiple data points over a pre-specified future runtime of the vehicular component based on the class label assigned to each of the multiple data points;

determining a cumulative hazard function for the vehicular component based on the failure probability, wherein said cumulative hazard function assesses the amount of accumulated risk that the vehicular component has faced from a given start time until the present time; and

generating a cumulative wear-based indicator for the vehicular component by executing a regression function at a given time based on (i) the cumulative hazard function, (ii) one or more selected parameters, and (iii) a determination as to whether the vehicular component (a) was previously replaced due to a failure, (b) was previously replaced due to a non-failure scheduled replacement, or (c) is actively running as yet to be replaced;

wherein the steps are carried out by at least one computing device.

2. The method of claim 1, comprising:

obtaining at least one input time series pertaining to the vehicular component across the fleet of multiple vehicles, wherein each input time series comprises data points from the set of multiple data points derived from the measurements associated with the vehicular component across the fleet of multiple vehicles.

3. The method of claim 1, wherein said failure probability (F_j(t)) for the vehicular component relates to said cumulative hazard function (H_j(t)) for the vehicular component such that F_j(t)=1−

S_j(t)=1−

exp(−

H_j(t)) wherein S_j(t) is a survival probability function for the vehicular component j at time t.

4. The method of claim 1, wherein said cumulative hazard function is based on an aggregate of multiple hazard function values of a given temporal interval across a given timeframe.

5. The method of claim 4, wherein said given temporal interval is one day, and wherein each of said multiple hazard functions of a one day temporal interval is defined as the total hazard (h_j(d)) during the daily runtime on day d for vehicular component j.

6. The method of claim 5, wherein the aggregate (H_j(t)) of multiple hazard functions of a one day temporal interval across the given timeframe comprises H_j(t)=Σ

<sub>all d in {d;

Meter(j,d)≦

t}</sub>h_j(d), wherein Meter(j,d) represents the accumulated runtime hours over days up to and including date d, and wherein the given timeframe comprises a given start date through time t.

7. The method of claim 1, wherein said given time comprises a given date, wherein said vehicular component was previously replaced due to a failure, and wherein said executing the regression function ({tilde over (h)}_j(d)) comprises calculating {tilde over (h)}_j(d)=(Meter(j, d)/Meter(j, T_F(j)))^α

wherein Meter(j,d) represents the total number of runtime hours up to and including date d, T_F(j) is the date that the vehicular component was previously replaced due to a failure, and α

≧

1.

8. The method of claim 1, wherein said given time comprises a given date, wherein said vehicular component was previously replaced due to a non-failure scheduled replacement, and wherein said executing the regression function ({tilde over (h)}_j(d)) comprises calculating {tilde over (h)}_j(d)=β

(Meter(j, d)/M_max)^α

, wherein M_max=max_i[Meter(i, T_F(i))], which represents the maximum total number of runtime hours over all vehicular components in the set of multiple data points, and β

represents a positive number between zero and one.

9. The method of claim 1, wherein said given time comprises a given date, wherein said vehicular component is actively running as yet to be replaced, and wherein said executing the regression function ({tilde over (h)}_j(d)) comprises calculating {tilde over (h)}_j(d)=β

(Meter(j, d)/M_max)^α

, wherein M_max=max_i[Meter(i, T_F (i))], which represents the maximum total number of runtime hours over all vehicular components in the set of multiple data points, and represents a positive number between zero and one.

10. The method of claim 1, wherein the one or more selected parameters comprise (i) the accumulated fuel consumption over a given time period, (ii) the accumulated amount of work over the given time period, and/or (iii) the accumulated number of sensor events for a target component type that encompasses the vehicular component over the given time period.

11. The method of claim 1, comprising:

generating multiple cumulative wear-based indicators for the vehicular component by executing multiple regression functions, wherein each of the multiple regression functions comprises a distinct combination of the one or more selected parameters; and

identifying the cumulative wear-based indicator from the multiple cumulative wear-based indicators that optimizes a given criterion.

12. A computer program product, the computer program product comprising a non-transitory computer readable storage medium having program instructions embodied therewith, the program instructions executable by a computing device to cause the computing device to:

assign a failure class label to each data point, from a set of multiple data points derived from measurements associated with a vehicular component across a fleet of multiple vehicles, that (a) is associated with a failure-caused vehicular component replacement, and (b) is within a pre-specified number of runtime hours of the failure-caused vehicular component replacement;

assign a non-failure class label to each data point, from the set of the multiple data points, that (a) is associated with a failure-caused vehicular component replacement, and (b) is not within the pre-specified number of runtime hours of the failure-caused vehicular component replacement;

assign a non-failure class label to each data point, from the set of the multiple data points, that is associated with a scheduled vehicular component replacement;

assign a non-failure class label to each data point, from the set of the multiple data points, that is associated with an actively running instance of the vehicular component as yet to be replaced;

estimate a failure probability for the vehicular component at each of the multiple data points over a pre-specified future runtime of the vehicular component based on the class label assigned to each of the multiple data points;

determine a cumulative hazard function for the vehicular component based on the failure probability, wherein said cumulative hazard function assesses the amount of accumulated risk that the vehicular component has faced from a given start time until the present time; and

generate a cumulative wear-based indicator for the vehicular component by executing a regression function at a given time based on (i) the cumulative hazard function, (ii) one or more selected parameters, and (iii) a determination as to whether the vehicular component (a) was previously replaced due to a failure, (b) was previously replaced due to a non-failure scheduled replacement, or (c) is actively running as yet to be replaced.

13. The computer program product of claim 12, wherein said failure probability (F_j(t)) for the vehicular component relates to said cumulative hazard function (H_j(t)) for the vehicular component such that F_j(t)=1−

S_j (t)=1−

exp(−

H_j(t)) wherein S_j(t) is a survival probability function for the vehicular component j at time t.

14. The computer program product of claim 12, wherein said cumulative hazard function is based on an aggregate of multiple hazard function values of a given temporal interval across a given timeframe.

15. The computer program product of claim 14, wherein said given temporal interval is one day, and wherein each of said multiple hazard functions of a one day temporal interval is defined as the total hazard (h_j(d)) during the daily runtime on day d for vehicular component j.

16. The computer program product of claim 15, wherein the aggregate (H_j(t)) of multiple hazard functions of a one day temporal interval across the given timeframe comprises H_j(t)=Σ

<sub>all d in {d;

Meter(j,d)≦

t}</sub>h_j(d), wherein Meter(j,d) represents the accumulated runtime hours over days up to and including date d, and wherein the given timeframe comprises a given start date through time t.

17. The computer program product of claim 12, wherein said given time comprises a given date, wherein said vehicular component was previously replaced due to a failure, and wherein said executing the regression function ({tilde over (h)}_j(d)) comprises calculating {tilde over (h)}_j(d)=(Meter(j, d)/Meter(j, T_F(j)))^α

wherein Meter(j,d) represents the total number of runtime hours up to and including date d, T_F(j) is the date that the vehicular component was previously replaced due to a failure, and α

≧

1.

18. The computer program product of claim 12, wherein said given time comprises a given date, wherein said vehicular component was previously replaced due to a non-failure scheduled replacement, and wherein said executing the regression function ({tilde over (h)}_j(d)) comprises calculating {tilde over (h)}_j(d)=β

(Meter(j,d)/M_max)^α

wherein M_max=max_i[Meter(i, T_F(i))], which represents the maximum total number of runtime hours over all vehicular components in the set of multiple data points, and β

represents a positive number between zero and one.

19. The computer program product of claim 12, wherein said given time comprises a given date, wherein said vehicular component is actively running as yet to be replaced, and wherein said executing the regression function ({tilde over (h)}_j(d)) comprises calculating {tilde over (h)}_j(d)=β

(Meter(j, d)/M_max)^α

, wherein M_max=max_i[Meter(i,T_F(i))], which represents the maximum total number of runtime hours over all vehicular components in the set of multiple data points, and β

represents a positive number between zero and one.

20. The computer program product of claim 12, wherein the one or more selected parameters comprise (i) the accumulated fuel consumption over a given time period, (ii) the accumulated amount of work over the given time period, and/or (iii) the accumulated number of sensor events for a target component type that encompasses the vehicular component over the given time period.