Data processing for component failure determination

Patent metadata

Field	Value
Publication number	US-10810069-B2
Application number	US-201816037770-A
Country	US
Kind code	B2
Filing date	Jul 17, 2018
Priority date	Jul 17, 2018
Publication date	Oct 20, 2020
Grant date	Oct 20, 2020

Abstract

A component analysis platform may communicate with one or more devices to obtain prediction data relating to a type of component. The component analysis platform may process the prediction data to determine a set of predictors for failure of an instance of the component, and may generate a model for failure of the instance of the component based on the set of predictors. The component analysis platform may monitor the instance of the component to obtain component data relating to the instance of the component. The component analysis platform may determine, using the model and based on the component data relating to the instance of the component, a predicted failure for the instance of the component. The component analysis platform may perform a response action related to the predicted failure.

First claim

What is claimed is:

1. A device, comprising:
one or more memories; and
one or more processors, communicatively coupled to the one or more memories, to:
communicate with one or more devices to obtain prediction data relating to a type of component,
wherein the prediction data includes:
log data,
entity data associated with a particular client, and
real-time or near-real time component data;

process the prediction data to determine a set of predictors for failure of an instance of the component;
receive information identifying a selected subset of predictors of the set of predictors,
the selected subset of predictors being associated with a threshold score;

generate a model for failure of the instance of the component based on the selected subset of predictors;
monitor the instance of the component to obtain component data relating to the instance of the component;
determine, using the model and based on the component data relating to the instance of the component, a predicted failure for the instance of the component; and
perform a response action related to the predicted failure.

2. The device of claim 1 , wherein the prediction data includes at least one of:
a system log,
an event log,
a performance monitor log,
a processor log, or
a database log.

3. The device of claim 1 , wherein the model is generated using a long term short memory (LSTM) neural network.

4. The device of claim 1 , wherein the set of predictors are determined based on at least one of:
a deep neural network (DNN) classifier technique,
a recurrent neural network technique, or
a covariance technique.

5. The device of claim 1 , wherein the model is generated based on a set of failure thresholds corresponding to the set of predictors and determined based on the prediction data.

6. The device of claim 1 , where the one or more processors are further to:
preprocess the prediction data to organize the prediction data for processing to determine the set of predictors.

7. The device of claim 1 , wherein the response action is a remediation action to prevent the failure.

8. The device of claim 1 , wherein the one or more processors are further to:
detect a failure scenario based on monitoring the instance of the component and adjust the model based on the failure scenario.

9. The device of claim 1 , wherein the component is a group of connected components.

10. The device of claim 1 , wherein the one or more processors, when monitoring the instance of the component, are to:
monitor at least one of: an internal parameter, a database parameter, a web server parameter, an application parameter, or a performance parameter.

11. A method, comprising:
communicating, by a first device, with a second device to obtain prediction data relating to a type of component,
wherein the prediction data includes:
log data,
entity data associated with a particular client, and
real-time or near-real time component data;

receiving, by the first device, information identifying a selected subset of predictors of a set of predictors,
the selected subset of predictors being associated with a threshold score;

generating, by the first device, a model for failure of an instance of the component based on:
the selected subset of predictors relating to failure of the instance of the component, and
the prediction data relating to the type of the component;

monitoring, by the first device, the instance of the component to obtain component data relating to the instance of the component;
determining, by the first device and using the model, a predicted failure for the instance of the component based on the component data relating to the instance of the component; and
communicating, by the first device, with a third device to transmit an alert relating to the predicted failure.

12. The method of claim 11 , wherein each predictor, of the set of predictors, is associated with a failure threshold, and
wherein determining the predicted failure comprises:
determining that at least one failure threshold of at least one predictor of the set of predictors is predicted to be satisfied.

13. The method of claim 11 , further comprising:
generating a user interface to identify the predicted failure; and
wherein communicating with the third device comprises:
providing the user interface for display to identify the predicted failure.

14. The method of claim 11 , wherein
the selected subset of predictors relate to a similar type of component to the type of component.

15. The method of claim 11 , wherein generating the model comprises:
performing natural language processing using a machine learning technique to process the prediction data.

16. The method of claim 11 , further comprising:
determining a predicted impact to an entity of the predicted failure; and
performing a response action corresponding to the predicted impact.

17. A non-transitory computer-readable medium storing instructions, the instructions comprising:
one or more instructions that, when executed by one or more processors, cause the one or more processors to:
communicate with one or more devices to obtain prediction data relating to a type of component,
wherein the prediction data includes:
log data,
entity data associated with a particular client, and
real-time or near-real time component data;

process the prediction data to determine a set of predictors for failure of an instance of the component;
receive information identifying a selected subset of predictors of the set of predictors,
the selected subset of predictors being associated with a threshold score;

generate a model for failure of the instance of the component based on the selected subset of predictors;
monitor the instance of the component to obtain component data relating to the instance of the component;
determine, using the model and based on the component data relating to the instance of the component, a predicted threshold ticket inflow based on monitoring the instance of the component; and
perform a response action to accommodate the predicted threshold ticket inflow.

18. The non-transitory computer-readable medium of claim 17 , wherein the response action includes:
proactively altering a task schedule to accommodate the predicted threshold ticket inflow.

19. The non-transitory computer-readable medium of claim 17 , wherein the response action includes:
automatically assigning tickets to accommodate the predicted threshold ticket inflow.

20. The non-transitory computer-readable medium of claim 17 , wherein the response action includes:
automatically applying a fix to avoid the predicted threshold ticket inflow.

Assignees

• Accenture Global Solutions Ltd

Inventors

• Ghosh Bhaskar
• SEKHAR Mohan
• Prasad Rajendra T
• Nagarajan Rajesh
• Venkateswaran Balaji
• JAGANNATHAN Purnima
• Manjunath Roopalaxmi
• Koushik Vijayaraghavan

Classifications

• G06N3/045

  Combinations of networks · CPC title

• G06N3/044

  Recurrent networks, e.g. Hopfield networks · CPC title

• G06N3/09

  Supervised learning · CPC title

• G06N3/0442

  characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU] · CPC title

• H04L41/149

  for prediction of maintenance · CPC title