Systems and methods for predicting information handling resource failures using deep recurrent neural network with a modified gated recurrent unit having missing data imputation

Patent metadata

Field	Value
Publication number	US-11227209-B2
Application number	US-201916528081-A
Country	US
Kind code	B2
Filing date	Jul 31, 2019
Priority date	Jul 31, 2019
Publication date	Jan 18, 2022
Grant date	Jan 18, 2022

Abstract

A method may include receiving telemetry data associated with one or more information handling resources, receiving failure statistics associated with the one or more information handling resources, merging the telemetry data and the failure statistics to create training data, and implementing a gated recurrent unit to: (i) impute missing values from the training data and (ii) train a pattern recognition engine configured to predict a failure status of an information handling resource from operational data associated with the information handling resource.

First claim

What is claimed is:

1. An information handling system comprising:
a processor; and
a non-transitory computer-readable medium having stored thereon a program of instructions executable by the processor, the program of instructions configured to, when read and executed by the processor:
receive telemetry data associated with one or more information handling resources;
receive failure statistics associated with the one or more information handling resources wherein the failure statistics include, for each information handling resource from which telemetry data is received, a failure status of the information handling resource;
merge the telemetry data and the failure statistics to create training data;
provide the training data to a gated recurrent unit;
impute, by the gated recurrent unit, missing values from the training data; and
train the gated recurrent unit, in accordance with the training data, to predict a future failure status of an information handling resource from operational data associated with the information handling resource, wherein the failure status is selected from a group of failure states comprising: failed, about to fail, and healthy;
wherein the gated recurrent unit is configured to impute the missing values using a last observation, a time since the last observation, and a distribution of a predictor.

2. The information handling system of claim 1 , wherein the training data comprises time series data generated from the telemetry data and the failure statistics.

3. The information handling system of claim 1 , wherein the program of instructions is further configured to, when read and executed by the processor, implement a pattern recognition engine as a recurrent neural network with the gated recurrent unit.

4. The information handling system of claim 1 , wherein the program of instructions is further configured to, when read and executed by the processor, apply a rules-based decision engine to the failure status to determine a remedial action for the information handling resource.

5. The information handling system of claim 1 , wherein the program of instructions is configured to impute the missing value and train the gated recurrent unit in a single step without storing datasets of imputed values.

6. A method comprising:
receiving telemetry data associated with one or more information handling resources;
receiving failure statistics associated with the one or more information handling resources, wherein the failure statistics include, for each information handling resource from which telemetry data is received, a failure status of the information handling resource;
merging the telemetry data and the failure statistics to create training data; and
providing the training data to a gated recurrent unit;
imputing, by the gated recurrent unit, missing values from the training data; and
training the gated recurrent unit, in accordance with the training data, to predict a future failure status of an information handling resource from operational data associated with the information handling resource, wherein the failure status is selected from a group of failure states comprising: failed, about to fail, and healthy;
wherein the gated recurrent unit is configured to impute the missing values using a last observation, a time since the last observation, and a distribution of a predictor.

7. The method of claim 6 , wherein the training data comprises time series data generated from the telemetry data and the failure statistics.

8. The method of claim 6 , further comprising implementing a pattern recognition engine as a recurrent neural network with the gated recurrent unit.

9. The method of claim 6 , further comprising applying a rules-based decision engine to the failure status to determine a remedial action for the information handling resource.

10. The method of claim 6 , wherein the program of instructions is configured to impute the missing value and train the gated recurrent unit in a single step without storing datasets of imputed values.

11. An article of manufacture comprising:
a non-transitory computer-readable medium; and
computer-executable instructions carried on the computer readable medium, the instructions readable by a processor, the instructions, when read and executed, for causing the processor to:
receive telemetry data associated with one or more information handling resources;
receive failure statistics associated with the one or more information handling resources wherein the failure statistics include, for each information handling resource from which telemetry data is received, a failure status of the information handling resource;
merge the telemetry data and the failure statistics to create training data;
provide the training data to a gated recurrent unit;
impute, by the gated recurrent unit, missing values from the training data; and
train the gated recurrent unit, in accordance with the training data, to predict a future failure status of an information handling resource from operational data associated with the information handling resource, wherein the failure status is selected from a group of failure states comprising: failed, about to fail, and healthy;
wherein the gated recurrent unit is configured to impute the missing values using a last observation, a time since the last observation, and a distribution of a predictor.

12. The article of claim 11 , wherein the training data comprises time series data generated from the telemetry data and the failure statistics.

13. The article of claim 11 , the instructions for further causing the processor to, when read and executed by the processor, a pattern recognition engine as a recurrent neural network with the gated recurrent unit.

14. The article of claim 11 , the instructions for further causing the processor to, when read and executed by the processor, apply a rules-based decision engine to the failure status to determine a remedial action for the information handling resource.

Assignees

• Dell Products Lp

Inventors

• Singh Ashutosh
• Chambers Landon Martin

Classifications

• G06N3/044

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

• G06N3/042

  Knowledge-based neural networks; Logical representations of neural networks · CPC title

• G06N3/084 Primary

  Backpropagation, e.g. using gradient descent · CPC title

• G06N3/0442 Primary

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

• G06N3/09

  Supervised learning · CPC title