摘要 |
PROBLEM TO BE SOLVED: To provide an emergency monitoring device for judging normality/abnormality about equipment whose operation changes in a specific procedure. SOLUTION: A plurality of types of operations are performed in a specific procedure by equipment X. An object signal generated according to the operation of the equipment X is captured by a signal inputted part 2, and the featured values of the object signal extracted by a feature extraction part 3 are input to a competition learning type neural network 1. The operating state of the competition learning type neural network 1 is reflected on an ignition position in a clustering map 5. The change pattern of the ignition position of the clustering map 5 corresponding to a series of operations of the equipment X is defined as an operation pattern, and the similarity of the operation pattern of the equipment X in a normal operation time stored in an operation pattern storage part 6 and the operation of the equipment X in the actual operation time, is evaluated by a judgement part 7. Thus, not only the abnormality of each operation of the equipment X but also the abnormality of the operation sequence of the equipment X is detected on the basis of the evaluation of the judgement part 7. COPYRIGHT: (C)2008,JPO&INPIT
|