System and Method for Calculating Remaining Useful Time of Objects

Patent metadata

Field	Value
Publication number	US-2015262060-A1
Application number	US-201514644346-A
Country	US
Kind code	A1
Filing date	Mar 11, 2015
Priority date	Mar 11, 2014
Publication date	Sep 17, 2015
Grant date	—

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.

First claim

1 . A computer implemented method of generating a used life percentage that is capable of identifying the remaining life in a mechanical component such as a bearing, comprising:
obtaining sensor data from the mechanical component and organizing the obtained sensor data into a defined matrix; calculating correlation coefficients for each matrix column and ranking the columns according to the correlation coefficients; inputting the ranked columns into a series of corresponding artificial neural networks and training each artificial neural network with its corresponding column; identifying the remaining useful life based on the trained artificial neural networks;

2 . The method of claim 1 , wherein the correlation coefficients are calculated for each column by determining a moving average of vibration signals and time;

3 . The method of claim 1 , wherein the identifying the remaining useful life further comprises:
calculating a prediction error for each artificial neural network and ranking the artificial neural networks based on their error; combining the results of all ANN predictions into a combination result; using the result as an indication of a used life percentage, or remaining useful life;

4 . The method of claim 3 , wherein the combination result is obtained by combining the results of all ANN predictions by the reciprocals of their errors and normalizing.

5 . The method of claim 1 , wherein the sensor data includes either temperature, pressure, magnetic information, or combinations thereof;

6 . The method of claim 1 , wherein the artificial neural networks are provided with fitted measurements;

7 . The method of claim 1 , further comprising:
obtaining training data from prior runs of the mechanical component; selecting for training bearings the artificial neural network with the least error when compared to training data;

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 from the mechanical component and organizing the obtained sensor data into a defined matrix; calculate correlation coefficients for each matrix column and ranking the columns according to the correlation coefficients; input the ranked columns into a series of corresponding artificial neural networks and training each artificial neural network with its corresponding column; identify the remaining useful life based on the trained artificial neural networks;

9 . The non-transitory storage device of claim 8 , wherein the instructions to cause the one or more processors to calculate correlation coefficients comprise calculating for each column by determining a moving average of vibration signals and time;

10 . The non-transitory storage device of claim 8 , wherein the instructions to cause the one or more processors to the identify the remaining useful life further comprises:
calculating a prediction error for each artificial neural network and ranking the artificial neural networks based on their error; combining the results of all ANN predictions into a combination result; using the result as an indication of a used life percentage, or remaining useful life;

11 . The non-transitory storage device of claim 10 , wherein the instructions to cause the one or more processors to the combine the results further comprise combining the results of all ANN predictions by the reciprocals of their errors and normalizing.

12 . The non-transitory storage device of claim 8 , wherein the sensor data includes either temperature, pressure, magnetic information, or combinations thereof;

13 . The non-transitory storage device of claim 8 , wherein the artificial neural networks are provided with fitted measurements;
The non-transitory storage device of claim 8 , further comprising instructions that when executed by the one or more processors cause the data processing apparatus to:
obtain training data from prior runs of the mechanical component;
select for training bearings the artificial neural network with the least error when compared to training data;

14 . A system comprising:
a mechanical component; one or more processors configured to perform operations comprising: obtaining sensor data from the mechanical component and organizing the obtained sensor data into a defined matrix; calculating correlation coefficients for each matrix column and ranking the columns according to the correlation coefficients; inputting the ranked columns into a series of corresponding artificial neural networks and training each artificial neural network with its corresponding column; identifying the remaining useful life based on the trained artificial neural networks;

15 . The system of claim 14 , wherein the correlation coefficients are calculated for each column by determining a moving average of vibration signals and time;

16 . The system of claim 14 , wherein the identifying the remaining useful life further comprises:
calculating a prediction error for each artificial neural network and ranking the artificial neural networks based on their error; combining the results of all ANN predictions into a combination result; using the result as an indication of a used life percentage, or remaining useful life;

17 . The system of claim 16 , wherein the combination result is obtained by combining the results of all ANN predictions by the reciprocals of their errors and normalizing.

18 . The system of claim 14 , wherein the sensor data includes either temperature, pressure, magnetic information, or combinations thereof;

19 . The system of claim 14 , wherein the artificial neural networks are provided with fitted measurements;

20 . The system of claim 14 , the operations further comprising:
obtaining training data from prior runs of the mechanical component; selecting for training bearings the artificial neural network with the least error when compared to training data;

Assignees

• Sparkcognition Inc

Inventors

• Husain Syed Mohammad Amir
• Abel Martin Andreas
• Iqbal Qasim

Classifications

• G06N3/08 Primary

  Learning methods · CPC title

• G01M13/04

  Bearings · CPC title

• G06N3/0499

  Feedforward networks · CPC title

• G06N3/09

  Supervised learning · CPC title

• G06N3/0985

  Hyperparameter optimisation; Meta-learning; Learning-to-learn · CPC title