Monitored machine performance as a maintenance predictor
A method, system, and computer program product for predicting abnormal operation of at least one component of a machine is provided. Real time monitoring data from an operating machine is received and monitoring features that are informative of likely abnormal operation are extracted and/or calculated. The monitoring features are applied to a prediction matrix that outputs probabilities of abnormal operation within one or more prediction time horizons. If the output probabilities exceed a threshold probability, then an alert can be output. Maintenance can be automatically scheduled in response to the alert.
1. A computer-implemented method of predicting component abnormal operation in a machine, the method comprising:
receiving monitoring data from the machine during operation of the machine;
computing at least one monitoring feature from the received monitoring data;
extracting a probability of a component operating abnormally from a prediction matrix based on the computed at least one monitoring feature, wherein the prediction matrix includes probabilities of abnormal operation for a first component of the machine, the probabilities of abnormal operation corresponding to values of a first monitoring feature; and
scheduling maintenance for the component of the machine upon the extracted probability exceeding a first threshold value.
2. The computer-implemented method of claim 1 , wherein the prediction matrix includes a first prediction sub-matrix that includes the probabilities of abnormal operation for combinations of values of the first monitoring feature and time horizons.
3. The computer-implemented method of claim 2 , wherein the at least one monitoring feature includes a second monitoring feature, wherein the prediction matrix further includes a second prediction sub-matrix related to abnormal operation of a second component of the machine, wherein the second prediction sub-matrix includes probabilities of abnormal operation corresponding to combinations of values of the second monitoring feature and time horizons.
4. The computer-implemented method of claim 2 , wherein the at least one monitoring feature includes a second monitoring feature, wherein the prediction matrix further includes a second prediction sub-matrix related to abnormal operation of the first component of the machine, wherein the second prediction sub-matrix includes probabilities of abnormal operation corresponding to combinations of values of the second monitoring feature and time horizons.
5. The computer-implemented method of claim 4 , wherein extracting the probability of a component operating abnormally comprises:
extracting, from the first prediction sub-matrix and the second prediction sub-matrix, a maximum probability for a particular time horizon and values of the first monitoring feature and the second monitoring feature.
6. The computer-implemented method of claim 2 , further comprising:
generating the first prediction sub-matrix, wherein generating the first prediction sub-matrix includes:
receiving historical monitoring data related to instances of abnormal operations of at least one of the machine, one or more machines of a same type as the machine, and one or more machines of a same class as the machine, wherein the instances of abnormal operations includes data for a first one of the at least one monitoring feature for a first period before an abnormal operation of a component and data for a second period after repair of the machine following the abnormal operation;
dividing a range of values of the at least one monitoring feature into a plurality of value sub-regions; and
for each value sub-region:
calculating a probability of abnormal operation for a first time horizon by dividing a total number of abnormal operations by a total number of instances in which the value of the first one of the at least one monitoring feature is in the value sub-region; and
calculating a probability of abnormal operation for a second time horizon by dividing a total number of abnormal operations by the total number of instances in which the value of the first one of the at least one monitoring feature is in the value sub-region.
7. The computer-implemented method of claim 1 , further comprising:
determining the at least one monitoring feature, wherein determining the at least one monitoring feature comprises:
receiving historical monitoring data related to instances of abnormal operations of at least one of the machine, one or more machines of a same type as the machine, and one or more machines of a same class as the machine, wherein the instances of abnormal operations includes data for a first period before an abnormal operation of a component and data for a second period after repair of the machine following the abnormal operation; and
computing the at least one monitoring feature from the data based on a difference in value for the at least one monitoring feature in the data before the abnormal operation and after the abnormal operation.
8. A system, comprising:
a computer processor; and
a computer memory storing:
a prediction matrix for abnormal operation of a component of a machine, wherein the prediction matrix includes probabilities of abnormal operation for a first component of the machine, the probabilities of abnormal operation corresponding to values of a first monitoring feature; and
an abnormal operation prediction application, which is executable by the computer processor to:
receive monitoring data from the machine during operation of the machine;
compute at least one monitoring feature from the received monitoring data;
extract a probability of a component operating abnormally from the prediction matrix based on the computed at least one monitoring feature; and
schedule maintenance for the component of the machine upon the extracted probability exceeding a first threshold value.
9. The system of claim 8 , wherein the prediction matrix includes a first prediction sub-matrix that includes the probabilities of abnormal operation for combinations of values of the first monitoring feature and time horizons.
10. The system of claim 9 , wherein the at least one monitoring feature includes a second monitoring feature, wherein the prediction matrix further includes a second prediction sub-matrix related to abnormal operation of a second component of the machine, wherein the second prediction sub-matrix includes probabilities of abnormal operation corresponding to combinations of values of the second monitoring feature and time horizons.
11. The system of claim 9 , wherein the at least one monitoring feature includes a second monitoring feature, wherein the prediction matrix further includes a second prediction sub-matrix related to abnormal operation of the first component of the machine, wherein the second prediction sub-matrix includes probabilities of abnormal operation corresponding to combinations of values of the second monitoring feature and time horizons.
12. The system of claim 11 , wherein extracting the probability of a component operating abnormally from comprises:
extracting, from the first prediction sub-matrix and the second prediction sub-matrix, a maximum probability for a particular time horizon and values of the first monitoring feature and the second monitoring feature.
13. The system of claim 9 , wherein the computer memory further stores a prediction matrix generating application, which is executable by the computer processor to generate the first prediction sub-matrix by:
receiving historical monitoring data related to instances of abnormal operations of at least one of the machine, one or more machines of a same type as the machine, and one or more machines of a same class as the machine, wherein the instances of abnormal operations includes data for a first one of the at least one monitoring feature for a first period before an abnormal operation of a component and data for a second period after repair of the machine following the abnormal operation;
dividing a range of values of the at least one monitoring feature into a plurality of value sub-regions; and
for each value sub-region:
calculating a probability of abnormal operation for a first time horizon by dividing a total number of abnormal operations by a total number of instances in which the value of the first one of the at least one monitoring feature is in the value sub-region; and
calculating a probability of abnormal operation for a second time horizon by dividing a total number of abnormal operations by the total number of instances in which the value of the first one of the at least one monitoring feature is in the value sub-region.
14. The system of claim 8 , wherein the computer memory further stores a monitoring feature generation application, which is executable by the computer processor to determine the at least one monitoring feature by:
receiving historical monitoring data related to instances of abnormal operations of at least one of the machine, one or more machines of a same type as the machine, and one or more machines of a same class as the machine, wherein the instances of abnormal operations includes data for a first period before an abnormal operation of a component and data for a second period after repair of the machine following the abnormal operation; and
computing the at least one monitoring feature from the data based on a difference in value for the at least one monitoring feature in the data before the abnormal operation and after the abnormal operation.
15. A computer program product for calculating a predicted abnormal operation of a machine, the computer program product comprising:
a non-transitory computer-readable storage medium having computer-readable program code embodied therewith, the computer-readable program code executable by one or more computer processors to:
receive monitoring data from the machine during operation of the machine;
compute at least one monitoring feature from the received monitoring data;
extract a probability of a component operating abnormally from a prediction matrix based on the computed at least one monitoring feature, wherein the prediction matrix includes probabilities of abnormal operation for a first component of the machine, the probabilities of abnormal operation corresponding to values of a first monitoring feature; and
schedule maintenance for the component of the machine upon the extracted probability exceeding a first threshold value.
16. The computer program product of claim 15 , wherein the prediction matrix includes a first prediction sub-matrix that includes the probabilities of abnormal operation for combinations of values of the first monitoring feature and time horizons.
17. The computer program product of claim 16 , wherein the at least one monitoring feature includes a second monitoring feature, wherein the prediction matrix further includes a second prediction sub-matrix related to abnormal operation of a second component of the machine, wherein the second prediction sub-matrix includes probabilities of abnormal operation corresponding to combinations of values of the second monitoring feature and time horizons.
18. The computer program product of claim 16 , wherein the at least one monitoring feature includes a second monitoring feature, wherein the prediction matrix further includes a second prediction sub-matrix related to abnormal operation of the first component of the machine, wherein the second prediction sub-matrix includes probabilities of abnormal operation corresponding to combinations of values of the second monitoring feature and time horizons.
19. The computer program product of claim 16 , wherein the computer-readable program code is further executable by the one or more computer processors to generate the first prediction sub-matrix by:
receiving historical monitoring data related to instances of abnormal operations of at least one of the machine, one or more machines of a same type as the machine, and one or more machines of a same class as the machine, wherein the instances of abnormal operations includes data for a first one of the at least one monitoring feature for a first period before an abnormal operation of a component and data for a second period after repair of the machine following the abnormal operation;
dividing a range of values of the at least one monitoring feature into a plurality of value sub-regions; and
for each value sub-region:
calculating a probability of abnormal operation for a first time horizon by dividing a total number of abnormal operations by a total number of instances in which the value of the first one of the at least one monitoring feature is in the value sub-region; and
calculating a probability of abnormal operation for a second time horizon by dividing a total number of abnormal operations by the total number of instances in which the value of the first one of the at least one monitoring feature is in the value sub-region.
20. The computer program product of claim 15 , wherein the computer-readable program code is further executable by the one or more computer processors to determine the at least one monitoring feature by:
receiving historical monitoring data related to instances of abnormal operations of at least one of the machine, one or more machines of a same type as the machine, and one or more machines of a same class as the machine, wherein the instances of abnormal operations includes data for a first period before an abnormal operation of a component and data for a second period after repair of the machine following the abnormal operation; and
computing the at least one monitoring feature from the data based on a difference in value for the at least one monitoring feature in the data before the abnormal operation and after the abnormal operation.