| 摘要 |
PROBLEM TO BE SOLVED: To solve the problem that a false report may occur because of shortage in learning data since a state may differ greatly because of an operation pattern, such as length of stop time although diversified normal states need to be collected cyclopaedically in abnormality sign detection of an instance base in facilities, and execution becomes difficult since memory capacity and calculation time increase when a learning data period is extended until the shortage in learning data is solved.SOLUTION: In a method for monitoring a state of facilities on the basis of a time-series signal outputted by the facilities, an operation pattern label is granted at each fixed time on the basis of the time-series signal, learning data are selected on the basis of the operation pattern label for each fixed period, a normal model is created on the basis of the selected learning data, an abnormal measure is calculated on the basis of the time-series signal and the normal mode, and it is identified whether the state of facilities is abnormal or normal on the basis of the calculated abnormal measure. |
