Abnormality detection apparatus, abnormality detection method, and non-transitory computer-readable medium

An abnormality detection method comprising:

• determining whether a relevance representing a relationship on a plurality of sets of first time-series data measured in a first time period with regard to a monitoring target is satisfied for a plurality of second time-series data measured in a second period of time;
• calculating an irregularity degree representing a degree that the second time-series data is abnormal based on the relevance determined to be satisfied and the relevance determined not to be satisfied;
• calculating a first degree representing a degree that the irregularity degree is a particular value when the second time-series data is normal or abnormal based on the irregularity degree calculated with regard to the second time-series data;
• calculating a second degree representing a degree that the plurality of sets of second time-series data are related, based on the relevance with regard to the first period of time in accordance with the Ising model based on the relevance;
• and
• deriving whether the second time-series data is normal or abnormal based on the first degree and the second degree and outputting an indication as to whether the second time-series data is normal or abnormal to thereby identify whether or not the second time-series data is a factor of a failure of the monitoring target.

A non-transitory computer-readable medium recorded with an abnormality detection program for causing a computer to realize:

• a determination function configured to determine whether a relevance representing a relationship on a plurality of sets of first time-series data measured in a first time period with regard to a monitoring target is satisfied for a plurality of second time-series data measured in a second period of time;
• an irregularity degree calculation function configured to calculate an irregularity degree representing a degree that the second time-series data is abnormal based on the relevance determined to be satisfied by the determination function and the relevance determined not to be satisfied by the determination function;
• a first degree calculation function configured to calculate a first degree representing a degree that the irregularity degree is a particular value when the second time-series data is normal or abnormal based on the irregularity degree calculated with regard to the second time-series data;
• a second degree calculation function configured to calculate a second degree representing a degree that the plurality of sets of second time-series data are related, on basis of the relevance with regard to the first period of time in accordance with the Ising model based on the relevance;
• and
• a state calculation function configured to derive whether the second time-series data is normal or abnormal based on the first degree and the second degree and to output an indication as to whether the second time-series data is normal or abnormal to thereby identify whether or not the second time-series data is a factor of a failure of the monitoring target.