Monitoring device, method for monitoring target device, and program
An acquisition unit is configured to acquire measurement values of a target device. A likelihood calculation unit is configured to calculate an occurrence likelihood for each of a plurality of phenomena that are liable to occur to the target device based on the measurement values acquired by the acquisition unit. A table storage unit is configured to store a table in which the plurality of phenomena and occurrence causes of abnormalities of the target device are associated to each other. As estimation unit is configured to estimate the occurrence causes based on the occurrence likelihood and the table.
1. A monitoring device, comprising:
a memory configured to store instructions; and
a processor circuit configured to execute the instructions to:
acquire measurement values of a target device, the measurement values including at least a temperature and a flow rate of an input fluid to be input to the target device, and a temperature and a flow rate of an output fluid to be output from the target device;
obtain correction measurement values in which the measurement values are corrected by using thermal equilibrium calculations based on the measurement values;
calculate a Mahalanobis distance with a factor of the correction measurement values:
when the calculated Mahalanobis distance is equal to or larger than a predetermined value, calculate an SN ratio related to the Mahalanobis distance and obtain likelihoods by calculating a likelihood associated with an occurrence of each of M kinds of phenomena that are liable to occur to the target device based on the SN ratio for each of the phenomena, wherein a number of the likelihoods is M that is same as a number of the kinds of phenomena;
store a table in which each of the M kinds of phenomena and occurrence causes of N kinds of abnormalities of the target device are associated with one another;
estimate the occurrence causes of abnormalities based on a vector including N elements that is obtained by calculation using a vector including the M likelihoods as elements and a matrix including values of the table as elements; and
output the estimated occurrence causes of the abnormalities of the target device to a display screen or print the estimated occurrence causes of the abnormalities of the target device on a sheet.
2. The monitoring device according to claim 1 ,
wherein, in the table, the occurrence causes of N kinds of abnormalities, the M kinds of phenomena, and the number of times at which the phenomena are observed when abnormalities occur due to the occurrence causes of N kinds of abnormalities are associated with one another, and
wherein the processor calculates a weighted sum of the occurrence likelihood and the number of times for each of the occurrence causes of N kinds of abnormalities, and estimates the occurrence causes of abnormalities based on the weighted sum.
3. The monitoring device according to claim 1 ,
wherein, in the table, the occurrence causes of N kinds of abnormalities, the M kinds of phenomena, and values indicating whether the plurality of phenomena are observed when abnormalities occur due to the occurrence causes of N kinds of abnormalities are associated with one another, and
wherein the processor estimates the occurrence causes of abnormalities based on the number of the elements including value that indicates the occurrence of phenomena in the matrix, which each have the occurrence likelihood being equal to or larger than a predetermined threshold value, among the plurality of phenomena.
4. The monitoring device according to claim 1 ,
wherein, in the table, the occurrence causes of N kinds of abnormalities, the M kinds of phenomena, and probabilities that the plurality of phenomena occur when abnormalities occur due to the occurrence causes of N kinds of abnormalities are associated with one another, and
wherein the processor calculates a weighted sum of the occurrence likelihood and the probabilities for each of the occurrence causes of N kinds of abnormalities, and estimates the occurrence causes of abnormalities based on the weighted sum.
5. A method for monitoring a target device, comprising:
acquiring, by a monitoring device, measurement values of the target device, the measurement values including at least a temperature and a flow rate of an input fluid to be input to the target device, and a temperature and a flow rate of an output fluid to be output from the target device;
obtaining, by a monitoring device, correction measurement values in which the measurement values are corrected by using thermal equilibrium calculations based on the measurement values;
calculating, by a monitoring device, a Mahalanobis distance with a factor of the correction measurement values;
when the calculated Mahalanobis distance is equal to or larger than a predetermined value, calculating, by a monitoring device, an SN ratio related to the Mahalanobis distance and obtaining, by the monitoring device, likelihoods by calculating a likelihood associated with an occurrence of each of M kinds of phenomena that are liable to occur to the target device based on the SN ratio for each of the phenomena;
estimating, by the monitoring device, occurrence causes of abnormalities based on a vector including N elements that is obtained by calculation using a vector including the likelihoods as elements and a matrix including values of a table as elements in which each of the M kinds of phenomena and the occurrence causes of N kinds of abnormalities of the target device are associated with one another, wherein a number of the likelihoods is M that is same as a number of the kinds of phenomena; and
outputting the estimated occurrence causes of the abnormalities of the target device to a display screen or printing the estimated occurrence causes of the abnormalities of the target device on a sheet.
6. The method for monitoring a target device according to claim 5 ,
wherein, in the table, the occurrence causes of N kinds of abnormalities, the M kinds of phenomena, and the number of times at which the phenomena are observed when abnormalities occur due to the occurrence causes of N kinds of abnormalities are associated with one another, and
wherein the monitoring device calculates a weighted sum of the occurrence likelihood and the number of times for each of the occurrence causes of N kinds of abnormalities, and estimates the occurrence causes of abnormalities based on the weighted sum.
7. The method for monitoring a target device according to claim 5 ,
wherein, in the table, the occurrence causes of N kinds of abnormalities, the M kinds of phenomena, and values indicating whether the plurality of phenomena are observed when abnormalities occur due to the occurrence causes of N kinds of abnormalities are associated with one another, and
wherein the monitoring device estimates the occurrence causes of abnormalities based on the number of the elements including value that indicates the occurrence of phenomena in the matrix, which each have the occurrence likelihood being equal to or larger than a predetermined threshold value, among the plurality of phenomena.
8. The method for monitoring a target device according to claim 5 ,
wherein, in the table, the occurrence causes of N kinds of abnormalities, the M kinds of phenomena, and probabilities that the plurality of phenomena occur when abnormalities occur due to the occurrence causes of N kinds of abnormalities are associated with one another, and
wherein the monitoring device calculates a weighted sum of the occurrence likelihood and the probabilities for each of the occurrence causes of N kinds of abnormalities, and estimates the occurrence causes of abnormalities based on the weighted sum.
9. A non-transitory computer-readable medium having a program stored thereon, the program causing a computer to execute the following processing:
acquiring measurement values of the target device, the measurement values including at least a temperature and a flow rate of an input fluid to be input to the target device, and a temperature and a flow rate of an output fluid to be output from the target device;
obtaining correction measurement values in which the measurement values are corrected by using thermal equilibrium calculations based on the measurement values;
calculating a Mahalanobis distance with a factor of the correction measurement values;
when the calculated Mahalanobis distance is equal to or larger than a predetermined value, calculating an SN ratio related to the Mahalanobis distance and obtaining likelihoods by calculating a likelihood associated with an occurrence of each of M kinds of phenomena that are liable to occur to the target device based on the SN ratio for each of the phenomena;
estimating occurrence causes of abnormalities based on a vector including N elements that is obtained by calculation using a vector including the likelihoods as elements and a matrix including values of the table as elements in which each of the M kinds of phenomena and the occurrence causes of N kinds of abnormalities of the target device are associated with one another, wherein a number of the likelihoods is M that is same as a number of kinds of phenomena; and
outputting the estimated occurrence causes of the abnormalities of the target device to a display screen or printing the estimated occurrence causes of the abnormalities of the target device on a sheet.