Battery test system for predicting battery test results

Patent metadata

Field	Value
Publication number	US-10191116-B2
Application number	US-201514884585-A
Country	US
Kind code	B2
Filing date	Oct 15, 2015
Priority date	Oct 15, 2015
Publication date	Jan 29, 2019
Grant date	Jan 29, 2019

Abstract

A method of predicting battery test results includes using a battery test computer to predict a battery test result for a battery undergoing testing. The battery test computer comprises a tangible, non-transitory computer-readable medium storing a battery test management system implemented as one or more sets of instructions. The battery test management system includes a predictive module configured to predict the battery test result using less than all data required for the battery test to be complete, a validation module configured to validate the prediction, and a training module configured to re-train the predictive module based on results generated by the validation module. The battery test computer also includes processing circuitry configured to execute the one or more sets of instructions, and outputting, via a user interface, the prediction of the result and a confidence level associated with the prediction.

First claim

The invention claimed is:

1. A battery testing system, comprising:
a battery test computer comprising a tangible, non-transitory computer-readable medium storing a battery test management system implemented as one or more sets of instructions, wherein the battery test management system comprises a prediction module, a validation module, and a training module, and wherein the battery test computer comprises processing circuitry configured to execute the one or more sets of instructions;
a user interface communicatively coupled to the processing circuitry and configured to provide outputs to a user;
wherein the prediction module is configured predict a result of a standardized battery test conducted on a battery undergoing testing using less than all data required for the standardized battery test conducted on the battery undergoing testing to be completed, and to output, via the user interface, the predicted result and a confidence level associated with the predicted result;
wherein the validation module is configured to determine a validity of the predicted result using final test results from the standardized battery test conducted on the battery, and to output, via the user interface, a representation of the validity;
wherein the training module is configured to update training data using the validity generated by the validation module to update the prediction module;
wherein the prediction module comprises a battery test support vector machine (SVM) constructed using training data obtained from a plurality of battery samples, and the battery test SVM is a binary classification model configured to classify the battery undergoing testing into a pass category or a fail category based on a first feature and a second feature generated from battery test data obtained during the standardized battery test, and wherein the prediction module is configured to output, via the user interface, a plurality of predicted results of the standardized battery test for the battery at time intervals occurring before a full duration of the standardized battery test; and
wherein the standardized battery test simulates the normal operating conditions that would be experienced by the battery undergoing testing over an average lifetime of the particular type of the battery undergoing testing.

2. The battery testing system of claim 1 , wherein the battery test computer is configured to:
generate, using the prediction module, a first predicted result of the standardized battery test for the battery undergoing testing at a first time point using test data obtained from the onset of the standardized battery test to the first time point; and
output, via the user interface, the first predicted result in response to determining that the first predicted result has a confidence level above a confidence threshold.

3. The battery testing system of claim 2 , wherein the battery test computer is configured to:
generate, using the prediction module, a second predicted result of the standardized battery test for the battery undergoing testing at a second time point occurring after the first time point using test data obtained from the onset of the standardized battery test to the second time point in response to determining that the confidence level of the first predicted result is lower than the confidence threshold; and
output, via the user interface, the second predicted result in response to determining that the second predicted result has a confidence level above the confidence threshold.

4. The battery testing system of claim 3 , wherein the first time point is 4 weeks, the second time point is 6 weeks, and the standardized battery test duration is 18 weeks.

5. The battery testing system of claim 1 , wherein the first feature is a rate of change of discharge capacity and the second feature is a rate of change of efficiency obtained for the battery undergoing testing during the standardized battery test.

6. The battery testing system of claim 1 , wherein the battery test SVM is configured to classify the battery undergoing testing into the pass category or the fail category by constructing a data space in which historical battery test results have been separated into a pass region and a fail region, calculating a hyperplane separating the pass region from the fail region, and mapping battery test data obtained for the battery undergoing testing to the data space to identify a location of the battery test data relative to the hyperplane.

7. The battery testing system of claim 6 , wherein the battery test computer is configured to predict, using at least the prediction module, a characteristic of the battery undergoing testing at the end of the standardized battery test using linear regression and a distance from the mapped battery test data obtained for the battery undergoing testing to the hyperplane, and to output the characteristic via the user interface.

8. The battery testing system of claim 7 , wherein the characteristic is a final discharge capacity of the battery undergoing testing.

9. The battery testing system of claim 1 , wherein the user interface comprises a display configured to output visual representations of outputs from the battery management system to the user.

10. The battery testing system of claim 1 , comprising a battery testing apparatus having voltage sensing circuitry, current sensing circuitry, or a combination thereof, and configured to obtain feedback indicative of charge capacity and discharge capacity from the battery undergoing testing.

11. The battery testing system of claim 10 , wherein the battery test computer comprises a communication system configured to receive battery test data from the battery testing apparatus.

12. A tangible, non-transitory computer-readable medium storing a battery test management system implemented as one or more sets of instructions, wherein the battery test management system comprises a prediction module, a validation module, and a training module that, when executed by a processor, are configured to:
generate a prediction of a result of a standardized battery test conducted on a battery undergoing testing using less than all data required for the standardized battery test conducted on the battery undergoing testing to be completed, and to output, via the user interface, the prediction of the result and a confidence level associated with the prediction of the result;
determine a validity of the prediction of the result using final test results from the standardized battery test conducted on the battery undergoing testing, and to output, via the user interface, a representation of the validity; and
update training data based on the determined validity to update the prediction module, wherein the prediction module comprises a battery test support vector machine (SVM) constructed using training data obtained from a plurality of battery samples, and the battery test SVM is a binary classification model configured to classify the battery undergoing testing into a pass category or a fail category based on a first feature and a second feature generated from battery test data obtained during the standardized battery test, and wherein the prediction module is configured to generate a plurality of predictions of the result of the standardized battery test for the battery at time intervals occurring before a full duration of the standardized battery test; and

wherein the standardized battery test simulates the normal operating conditions that would be experienced by the battery undergoing testing over an average lifetime of the particular type of the battery undergoing testing.

13. The tangible, non-transitory computer-readable medium of

Assignees

• Johnson Controls Tech Co

Inventors

• Aliyev Timur L
• Jin Zhihong H
• Martinez Diego Hernan Diaz

Classifications

• G06F2218/12

  Classification; Matching · CPC title

• G06F18/2411

  based on the proximity to a decision surface, e.g. support vector machines · CPC title

• G01R31/3651 Primary

  Physics · mapped topic

• G06K9/6269

  Physics · mapped topic

• G01R31/3627

  Physics · mapped topic