System and method for calculating remaining useful time of objects

• US 10,410,116 B2
• Filed: 03/11/2015
• Issued: 09/10/2019
• Est. Priority Date: 03/11/2014
• Status: Active Grant

• Current Assignee: SparkCognition Inc.
• Original Assignee: SparkCognition Inc.
• Inventors: Husain, Syed Mohammad Amir, Abel, Martin Andreas, Iqbal, Qasim
• Primary Examiner(s): Gonzales, Vincent
• Application Number:
  US14/644,346
• Publication Number:
  US 20150262060A1
• Time in Patent Office: 1,644 Days
• CPC Class Codes: G01M 13/04 Bearings
• CPC Class Codes: G06N 3/08 Learning methods

Abstract

An aspect of the present invention is to provide a system and method for predicting the remaining useful time of mechanical components such as bearings. Another aspect of the present invention is to provide a system and method for predicting the remaining useful time of bearings based on available condition monitoring data. Another aspect of the present invention is to provide a system and method for automatically deciding which columns of input information are the most significant for predicting the remaining useful life of bearings. Another aspect of the present invention is to provide a system and method for performing an analysis of both test bearings and training bearings and determining which training bearings are most similar to a given test bearing. Another aspect of the present invention is to provide a system and method for training an artificial neural network.

Claims

1. A computer implemented method comprising:

obtaining sensor data for a mechanical component;

determining a plurality of features based on the sensor data;

organizing the plurality of features into a defined matrix structure, the defined matrix structure including a plurality of columns, each column of the plurality of columns corresponding to a particular feature of the plurality of features, wherein the columns are sorted in the defined matrix structure in order of a strength of a monotonic relationship between the particular feature and time;

inputting data of the defined matrix structure into an artificial neural network; and

generating output identifying a remaining useful life of the mechanical component, the output based on a result generated by the artificial neural network responsive to the input data.

2. The method of claim 1, wherein the strength of the monotonic relationship between the particular feature and time is determined based on training data used to train the artificial neural network.

3. The method of claim 1, further comprising before organizing the plurality of features into the defined matrix structure, determining the defined matrix structure by:

extracting features from training data;

determining a correlation between each extracted feature and time; and

defining a matrix structure such that the columns of the defined matrix structure are arranged in order of decreasing strength of the correlation.

4. The method of claim 3, further comprising:

training a plurality of artificial neural networks using the training data by;

training a first artificial neural network using first training data, the first training data corresponding to a first extracted feature of the training data, the first extracted feature having a strongest correlation to time of the extracted features;

training a second artificial neural network using the first training data and second training data, the second training data corresponding to a second extracted feature of the training data, the second extracted feature having a second strongest correlation to time of the extracted features; and

training a third artificial neural network using the first training data, the second training data, and third training data, the third training data corresponding to a third extracted feature of the training data, the third extracted feature having a third strongest correlation to time of the extracted features;

determining, based on test data, an error associated with each artificial neural network of the plurality of artificial neural networks; and

selecting the artificial neural network from among the plurality of artificial neural networks, wherein the artificial neural network has a lowest error among the plurality of artificial neural networks.

5. The method of claim 1, wherein the sensor data includes vibration, temperature, pressure, magnetic information, or a combination thereof.

6. The method of claim 1, wherein determining the plurality of features based on the sensor data includes fitting a curve to the sensor data and determining feature values based on the curve.

7. The method of claim 6, wherein the curve is a Weibull distribution curve.

8. A non-transitory storage device storing computer instructions that when executed by one or more processors cause the one or more processors to:

obtain sensor data for a mechanical component;

determine a plurality of features based on the sensor data;

organize the plurality of features into a defined matrix structure, the defined matrix structure including a plurality of columns, each column of the plurality of columns corresponding to a particular feature of the plurality of features, wherein the columns are sorted in the defined matrix structure in order of a strength of a monotonic relationship between the particular feature and time;

input data of the defined matrix structure into an artificial neural network; and

generate output identifying a remaining useful life of the mechanical component, the output based on a result generated by the artificial neural network responsive to the input data.

9. The non-transitory storage device of claim 8, wherein the strength of the monotonic relationship between the particular feature and time is determined based on training data used to train the artificial neural network.

10. The non-transitory storage device of claim 8, wherein the sensor data includes vibration, temperature, pressure, magnetic information, or a combination thereof, and wherein determining the plurality of features based on the sensor data includes fitting a curve to the sensor data and determining feature values based on the curve.

11. The non-transitory storage device of claim 10, wherein the curve is a Weibull distribution curve.

12. A system comprising:

a mechanical component; and

one or more processors configured to perform operations comprising;

obtaining sensor data for the mechanical component;

determining a plurality of features based on the sensor data;

organizing the plurality of features into a defined matrix structure, the defined matrix structure including a plurality of columns, each column of the plurality of columns corresponding to a particular feature of the plurality of features, wherein the columns are sorted in the defined matrix structure in order of a strength of a monotonic relationship between the particular feature and time;

inputting data of the defined matrix structure into an artificial neural network; and

generating output identifying a remaining useful life of the mechanical component, the output based on a result generated by the artificial neural network responsive to the input data.

13. The system of claim 12, wherein the strength of the monotonic relationship between the particular feature and time is determined based on training data used to train the artificial neural network.

14. The system of claim 12, wherein the sensor data includes vibration, temperature, pressure, magnetic information, or a combination thereof.

15. The system of claim 12, wherein the determining the plurality of features based on the sensor data includes fitting a curve to the sensor data and determining feature values based on the curve.

16. The system of claim 15, wherein the curve is a Weibull distribution curve.