Cost-driven system and method for predictive equipment failure detection

Patent metadata

Field	Value
Publication number	US-10776706-B2
Application number	US-201615053978-A
Country	US
Kind code	B2
Filing date	Feb 25, 2016
Priority date	Feb 25, 2016
Publication date	Sep 15, 2020
Grant date	Sep 15, 2020

Abstract

A method includes identifying costs associated with different outcomes of a failure prediction algorithm. The algorithm is configured to predict one or more faults with at least one piece of industrial equipment. The different outcomes include both successful and unsuccessful predictions by the algorithm. The method also includes identifying a threshold value for the algorithm using the costs, where the threshold value is used by the failure prediction algorithm to identify whether maintenance of the at least one piece of industrial equipment is needed. The method further includes providing the threshold value to the algorithm. The threshold value is selected such that a net positive economic benefit is obtained from use of the threshold value with the failure prediction algorithm. In addition, the method can include generating a signal indicating whether maintenance is needed based on a comparison of an indicator value calculated using the algorithm and the threshold value.

First claim

What is claimed is:

1. A method for the predictive detection of equipment failure comprising:
receiving from a prediction tool, equipment information indicating one or more characteristics of a piece of equipment including information from one or more sensors, actuators, controllers, or process historians,
identifying by the prediction tool an indicator value, based on the equipment information using a classifier of a failure prediction algorithm by identifying costs associated with different outcomes of the failure prediction algorithm, the failure prediction algorithm configured to predict one or more faults with at least one piece of industrial equipment, the different outcomes including both successful and unsuccessful predictions by the failure prediction algorithm, wherein the different outcomes of the failure prediction algorithm comprise a true positive, a true negative, a false positive, and a false negative;
incorporating a cost calculation into a receiver operating characteristic (ROC) curve in the form of a cost breakeven baseline that evaluates the cost benefit of repairing or replacing the equipment;
determining a threshold value, for the failure prediction algorithm using a point on the classifier's receiver operating characteristic (ROC) curve on the cost breakeven baseline showing the true positive rates relative to the false positive rates, the threshold value used by the failure prediction algorithm to identify whether maintenance of the at least one piece of industrial equipment is needed; and
providing the threshold value to the failure prediction algorithm wherein the threshold value is selected such that a net positive economic benefit is obtained from use of the threshold value with the failure prediction algorithm;
generating a signal for maintenance of the at least one piece of industrial equipment by comparing the indicator value with the threshold value.

2. The method of claim 1 , wherein the different outcomes of the failure prediction algorithm comprise:
the true positive in which the failure prediction algorithm correctly predicts that maintenance of the at least one piece of industrial equipment is needed;
the true negative in which the failure prediction algorithm correctly predicts that maintenance of the at least one piece of industrial equipment is not needed;
the false positive in which the failure prediction algorithm incorrectly predicts that maintenance of the at least one piece of industrial equipment is needed; and
the false negative in which the failure prediction algorithm incorrectly predicts that maintenance of the at least one piece of industrial equipment is not needed.

3. The method of claim 2 , wherein identifying the threshold value comprises:
identifying a cost breakeven baseline associated with the failure prediction algorithm; and
identifying true positive rates relative to false positive rates for a classifier of the failure prediction algorithm.

4. The method of claim 3 , wherein identifying the threshold value further comprises:
the identified point having a largest positive difference from the cost breakeven baseline; and
identifying the threshold value using the identified point.

5. The method of claim 3 , wherein the cost breakeven baseline is determined as a cost of repairing or replacing the at least one piece of equipment after the at least one piece of equipment has already experienced the one or more faults without prediction of the one or more faults.

6. An apparatus for the predictive detection of equipment failure comprising:
at least one processing device configured to:
receive from a prediction tool, equipment information indicating one or more characteristics of a piece of equipment including information from one or more sensors, actuators, controllers, or process historians,
identify by the prediction tool an indicator value, based on the equipment information using a classifier of a failure prediction algorithm by identifying costs associated with different outcomes of the failure prediction algorithm, the failure prediction algorithm configured to predict one or more faults with at least one piece of industrial equipment, the different outcomes including both successful and unsuccessful predictions by the failure prediction algorithm, wherein the different outcomes of the failure prediction algorithm comprise a true positive, a true negative, a false positive, and a false negative;
incorporate a cost calculation into a receiver operating characteristic (ROC) curve in the form of a cost breakeven baseline that evaluates the cost benefit of repairing or replacing the equipment;
determine a threshold value for the failure prediction algorithm using a point on the classifier's receiver operating characteristic (ROC) curve on the cost breakeven baseline showing the true positive rates relative to the false positive rates, the threshold value used by the failure prediction algorithm to identify whether maintenance of the at least one piece of industrial equipment is needed;
provide the threshold value to the failure prediction algorithm; wherein the at least one processing device is configured to select the threshold value such that a net positive economic benefit is obtained from use of the threshold value with the failure prediction algorithm, and
generate a signal for maintenance of the at least one piece of industrial equipment by comparing the indicator value with the threshold value.

7. The apparatus of claim 6 , wherein the different outcomes of the failure prediction algorithm comprise:
the true positive in which the failure prediction algorithm correctly predicts that maintenance of the at least one piece of industrial equipment is needed;
the true negative in which the failure prediction algorithm correctly predicts that maintenance of the at least one piece of industrial equipment is not needed;
the false positive in which the failure prediction algorithm incorrectly predicts that maintenance of the at least one piece of industrial equipment is needed; and
the false negative in which the failure prediction algorithm incorrectly predicts that maintenance of the at least one piece of industrial equipment is not needed.

8. The apparatus of claim 7 , wherein the at least one processing device is configured to:
identify a cost breakeven baseline associated with the failure prediction algorithm; and
identify true positive rates relative to false positive rates for a classifier of the failure prediction algorithm.

9. The apparatus of claim 8 , wherein the at least one processing device is configured to:
the identified point having a largest positive difference from the cost breakeven baseline; and
identify the threshold value using the identified point.

10. The apparatus of claim 8 , wherein the at least one processing device is configured to determine the cost breakeven baseline as a cost of repairing or replacing the at least one piece of industrial equipment after the at least one piece of equipment has already experienced the one or more faults without prediction of the one or more faults.

11. A non-transitory computer readable medium for the predictive detection of equipment failure containing instructions that, when executed by at least one processing device, cause the at least one processing device to:
receive from a prediction tool, equipment information indicating one or more characteristics of a piece of equipment including information from one or more sensors, actuators, controllers, or process historians,
identify by the prediction tool an indicator value, based on the equipment information using a classifier of a failure prediction algorithm by identifying costs associated with

Assignees

• Honeywell Int Inc

Inventors

• Zirnstein Jan
• Germann David J
• Light Marc
• Stewart Gregory E
• Grunow Jonathan T

Classifications

• G06N20/00 Primary

  Machine learning · CPC title

• G05B23/0283

  Predictive maintenance, e.g. involving the monitoring of a system and, based on the monitoring results, taking decisions on the maintenance schedule of the monitored system; Estimating remaining useful life [RUL] (preventive maintenance, i.e. planning maintenance according to the available resources without monitoring the system G06Q10/06) · CPC title

• G06N5/04 Primary

  Inference or reasoning models · CPC title