Multi-stage failure analysis and prediction

• US 9,569,298 B2
• Filed: 02/13/2016
• Issued: 02/14/2017
• Est. Priority Date: 03/14/2013
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Hampapur, Arun, Li, Hongfei, Li, Zhiguo, Zhu, Yada
• Primary Examiner(s): Mehrmanesh, Elmira
• Application Number:
  US15/043,532
• Publication Number:
  US 20160210183A1
• Time in Patent Office: 367 Days
• Field of Search: 714/26, 714/47.2, 714/47.3
• US Class Current: 1/1
• CPC Class Codes: G06F 11/0709 in a distributed system consisting of a plurality of standalone computer nodes, e.g. clusters, client-server systems
• CPC Class Codes: G06F 11/0751 Error or fault detection not based on redundancy power supply failures G06F1/30; network fault management H04L41/06
• CPC Class Codes: G06F 11/0787 Storage of error reports, e.g. persistent data storage, storage using memory protection
• CPC Class Codes: G06F 11/079 Root cause analysis, i.e. error or fault diagnosis in a hardware test environment G06F11/22; in a software test environment G06F11/36
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 7/01 Probabilistic graphical models, e.g. probabilistic networks

Abstract

A hierarchical multi-stage model of asset failure risk for complex heterogeneously distributed physical assets is built. The hierarchical multi-stage model considers heterogeneity of failure patterns for the assets. At least one data stream is analyzed to determine whether the hierarchical multi-stage model needs to be updated due to a change in the failure patterns. If the analysis indicates that the hierarchical multi-stage model needs to be updated, the hierarchical multi-stage model is dynamically updated to obtain an updated hierarchical multi-stage model.

Claims

1. A method comprising:

building a hierarchical multi-stage model of asset failure risk for complex heterogeneously distributed physical assets, said hierarchical multi-stage model considering heterogeneity of failure patterns for said assets including seasonality and geographical distribution, by;

building a failure frequency analysis engine of said hierarchical multi-stage model;

building a failure association analysis engine of said hierarchical multi-stage model; and

building a failure prediction analysis engine of said hierarchical multi-stage model;

analyzing at least one data stream to determine a change in said failure patterns, including analyzing at least one of equipment failure history, maintenance records, equipment attributes, and sensor data; and

dynamically updating said hierarchical multi-stage model to obtain an updated hierarchical multi-stage model based at least in part on said change in said failure patterns;

wherein said considering heterogeneity of failure patterns for said assets comprises;

performing multi-dimensional segmentation on data comprising failure history for said assets to obtain segmented data; and

implementing hierarchical modeling to integrate heterogeneous clusters of failure risks for said assets from said segmented data.

2. The method of claim 1, wherein said analyzing of said data stream comprises carrying out statistical model-based change detection.

3. The method of claim 1, wherein said analyzing of said data stream comprises carrying out data mining model-based change detection.

4. The method of claim 1, wherein said dynamic updating of said hierarchical multi-stage model comprises applying a Bayesian inference technique.

5. The method of claim 4, further comprising using said hierarchical multi-stage model as an input to said Bayesian inference technique.

6. The method of claim 1, further comprising providing a system, wherein the system comprises distinct software modules, each of the distinct software modules being embodied on a computer-readable storage medium, and wherein the distinct software modules comprise a hierarchical modeling module, a change detection module, and a model updating module;

wherein;

said building is carried out by said hierarchical modeling module executing on at least one hardware processor;

said analyzing is carried out by said change detection module executing on said at least one hardware processor; and

said dynamic updating is carried out by said model updating module executing on said at least one hardware processor.

7. A non-transitory computer readable storage medium comprising computer executable instructions which when executed by a computer cause the computer to perform the method of:

building a hierarchical multi-stage model of asset failure risk for complex heterogeneously distributed physical assets, said hierarchical multi-stage model considering heterogeneity of failure patterns for said assets including seasonality and geographical distribution, by;

building a failure frequency analysis engine of said hierarchical multi-stage model;

building a failure association analysis engine of said hierarchical multi-stage model; and

building a failure prediction analysis engine of said hierarchical multi-stage model;

analyzing at least one data stream to determine a change in said failure patterns, including analyzing at least one of equipment failure history, maintenance records, equipment attributes, and sensor data; and

dynamically updating said hierarchical multi-stage model to obtain an updated hierarchical multi-stage model based at least in part on said change in said failure patterns;

wherein said considering heterogeneity of failure patterns for said assets comprises;

performing multi-dimensional segmentation on data comprising failure history for said assets to obtain segmented data; and

implementing hierarchical modeling to integrate heterogeneous clusters of failure risks for said assets from said segmented data.

8. The non-transitory computer readable storage medium of claim 7, wherein said analyzing of said data stream comprises carrying out statistical model-based change detection.

9. The non-transitory computer readable storage medium of claim 7, wherein said analyzing of said data stream comprises carrying out data mining model-based change detection.

10. The non-transitory computer readable storage medium of claim 7, wherein said dynamic updating of said hierarchical multi-stage model comprises applying a Bayesian inference technique.

11. The non-transitory computer readable storage medium of claim 10, further comprising computer executable instructions which when executed by said computer cause said computer to use said hierarchical multi-stage model as an input to said Bayesian inference technique.

12. The non-transitory computer readable storage medium of claim 7, wherein said computer-executable instructions comprise distinct software modules, each of the distinct software modules being embodied on said non-transitory computer readable storage medium, and wherein the distinct software modules comprise a hierarchical modeling module, a change detection module, and a model updating module;

wherein;

said building is carried out by said hierarchical modeling module;

said analyzing is carried out by said change detection module; and

said dynamic updating is carried out by said model updating module.

13. An apparatus comprising:

a memory;

at least one processor, coupled to said memory, and;

a non-transitory computer readable storage medium comprising computer executable instructions which when loaded into said memory configure said at least one processor to implement;

a hierarchical multi-stage model of asset failure risk for complex heterogeneously distributed physical assets, said hierarchical multi-stage model considering heterogeneity of failure patterns for said assets including seasonality and geographical distribution, said hierarchical multi-stage model in turn comprising;

a failure frequency analysis engine;

a failure association analysis engine; and

a failure prediction analysis engine;

a change detection component which analyzes at least one data stream to determine a change in said failure patterns, including analyzing at least one of equipment failure history, maintenance records, equipment attributes, and sensor data; and

a sequential model updating component which, based at least in part on said change in said failure patterns, dynamically updates said hierarchical multi-stage model to obtain an updated hierarchical multi-stage model;

wherein said hierarchical multi-stage model considers said heterogeneity of failure patterns for said assets by;

performing multi-dimensional segmentation on data comprising failure history for said assets to obtain segmented data; and

implementing hierarchical modeling to integrate heterogeneous clusters of failure risks for said assets from said segmented data.

14. The apparatus of claim 13, wherein said change detection component analyzes said at least one data stream by carrying out statistical model-based change detection.

15. The apparatus of claim 13, wherein said change detection component analyzes said at least one data stream by carrying out data mining model-based change detection.

16. The apparatus of claim 13, wherein said sequential model updating component dynamically updates said hierarchical multi-stage model by applying a Bayesian inference technique.

17. The apparatus of claim 16, wherein said non-transitory computer readable storage medium further comprises computer executable instructions which when loaded into said memory configure said at least one processor to use said hierarchical multi-stage model as an input to said Bayesian inference technique.