Predicting leading indicators of an event
Provided is a system and method for predicting leading indicators for predicting occurrence of an event at a target asset. Rather than rely on traditional manufacturer-defined leading indicators for an asset, the examples herein predict leading indicators for a target asset based on actual operating conditions at the target asset. Accordingly, unanticipated operating conditions can be considered. In one example, the method may include receiving operating data of a target resource, the operating data being associated with previous occurrences of an event at the target resource, predicting one or more leading indicators of the event at the target resource based on the received operating data, each leading indicator comprising a variable and a threshold value for the variable, and outputting information about the one or more predicted leading indicators of the target resource for display via a user interface.
1. A computing system comprising:
a storage configured to store operating data of a target resource which includes a plurality of variables; and
a processor configured to
generate a plurality of predicted outputs, via execution of a machine learning model, each predicted output indicating whether or not a future event will occur at the target resource, wherein the machine learning model receives inputs including the plurality of variables in the stored operating data to generate the plurality of predicted outputs, and
learn, via execution of a surrogate machine learning model, one or more leading indicators of the event at the target resource based on prediction patterns between the inputs and the plurality of predicted outputs of the machine learning model identified by the surrogate machine learning model, wherein a leading indicator from among the one or more leading indicators comprises a variable and a conditional threshold value for the variable which is most closely correlated to the machine learning model predicting the occurrence of the event,
wherein the processor is further configured to receive an input via a user interface, incorporate the variable and the conditional threshold value for the variable into a monitoring rule, and monitor the asset for the event based on the monitoring rule.
2. The computing system of claim 1 , wherein the target resource comprises at least one of a machine and equipment, and the event comprises at least one of a hardware failure, an error code, a warranty claim, and a customer complaint.
3. The computing system of claim 1 , wherein the target resource performs an industrial operation, and the stored operating data comprises sensor data captured of the target resource during performance of the industrial operation.
4. The computing system of claim 1 , wherein the operating data comprises at least one of sensor data of one or more upstream resources that effect operation of the target resource and historical system data of the one or more upstream resources.
5. The computing system of claim 1 , wherein at least one leading indicator comprises a time-series variable and one or more of a lower value threshold and an upper value threshold for the time-series variable.
6. The computing system of claim 1 , wherein the one or more predicted leading indicators comprise one or more identified sensors and respective conditional threshold values for the one or more identified sensors that are most closely correlated to a predicted failure of the target resource by the machine learning model.
7. The computing system of claim 1 , wherein the one or more predicted leading indicators further comprise at least one of a skill level of an operator of the target resource, an age of the target resource, a model type of the target resource, and a manufacturer of the target resource.
8. The computing system of claim 1 , wherein the processor is configured to monitor additional operating data of the target resource over time for occurrence of the event based on the monitoring rule.
9. A method comprising:
receiving operating data of a target resource which includes a plurality of variables;
generating a plurality of predicted outputs, via execution of a machine learning model, each predicted output indicating whether or not a future event will occur at the target resource, wherein the machine learning model receives inputs including the plurality of variables in the stored operating data to generate the plurality of predicted outputs;
learning, via execution of a surrogate machine learning model, one or more leading indicators of the event at the target resource based on prediction patterns between the inputs and the plurality of predicted outputs of the machine learning model identified by the surrogate machine learning model, wherein a leading indicator from among the one or more leading indicators comprises a variable and a conditional threshold value for the variable which is most closely correlated to the machine learning model predicting the occurrence of the event; and
receiving an input via a user interface, incorporating the variable and the conditional threshold value for the variable into a monitoring rule, and monitoring the asset for the event based on the monitoring rule.
10. The method of claim 9 , wherein the target resource comprises at least one of a machine and equipment, and the event comprises at least one of a hardware failure, an error code, a warranty claim, and a customer complaint.
11. The method of claim 9 , wherein the target resource comprises a model type of an equipment which represents a class of equipment having a same model type.
12. The method of claim 9 , wherein the operating data comprises at least one of sensor data captured of the target resource and sensor data captured of one or more upstream resources that effect operation of the target resource and historical system data of the one or more upstream resources.
13. The method of claim 9 , wherein at least one leading indicator comprises a time-series variable and one or more of a lower value threshold and an upper value threshold for the time-series variable.
14. The method of claim 9 , wherein the one or more predicted leading indicators comprise one or more identified sensors and respective conditional threshold values for the one or more identified sensors that are most closely correlated to a predicted failure of the target resource by the machine learning model.
15. The method of claim 9 , wherein the one or more predicted leading indicators further comprise at least one of a skill level of an operator of the target resource, an age of the target resource, a model type of the target resource, and a manufacturer of the target resource.
16. The method of claim 9 , wherein the method comprises monitoring additional operating data of the target resource over time for occurrence of the event based on the monitoring rule.
17. A non-transitory computer-readable medium storing program instructions which when executed by a processor cause a computer to perform a method comprising:
receiving operating data of a target resource which includes a plurality of variables;
generating a plurality of predicted outputs, via execution of a machine learning model, each predicted output indicating whether or not a future event will occur at the target resource, wherein the machine learning model receives inputs including the plurality of variables in the stored operating data to generate the plurality of predicted outputs;
learning, via execution of a surrogate machine learning model, one or more leading indicators of the event at the target resource based on prediction patterns between the inputs and the plurality of predicted outputs of the machine learning model identified by the surrogate machine learning model, wherein a leading indicator from among the one or more leading indicators comprises a variable and a conditional threshold value for the variable which is most closely correlated to the machine learning model predicting the occurrence of the event; and
receiving an input via a user interface, incorporating the variable and the conditional threshold value for the variable into a monitoring rule, and monitoring the asset for the event based on the monitoring rule.
18. The non-transitory computer-readable medium of claim 17 , wherein the target resource comprises at least one of a machine and equipment, and the event comprises at least one of a hardware failure, an error code, a warranty claim, and a customer complaint.
19. The non-transitory computer-readable medium of claim 17 , wherein the target resource performs an industrial operation, and the operating data comprises sensor data captured of the target resource during performance of the industrial operation.
20. The non-transitory computer-readable medium of claim 17 , wherein at least one leading indicator comprises a time-series variable and one or more of a lower value threshold and an upper value threshold for the time-series variable.
21. The computing system of claim 1 , wherein the machine learning model is configured to predict occurrence of a failure event at the target resource based on the stored operating data, and the processor is configured to predict, via execution of the surrogate model, one or more leading indicators of the failure event at the target resource based on patterns between the inputs and the predicted outputs of the machine learning model which are identified by the surrogate machine learning model.