摘要 |
PROBLEM TO BE SOLVED: To provide an abnormality detection device which makes it detectable that there is no specific abnormal status during operation of equipment, and an abnormality detection method. SOLUTION: Existence of abnormalities of the equipment X is decided by a competitive learning type neural network 1 by using the feature amount of a target signal containing an oscillating component produced by an operation of the equipment X. A data set comprising a plurality of data of which categories are considered a specific abnormal condition is stored in a learning data storage part 6, and the competitive learning type neural network 1 is made to be learnt by this data set. A determination part 5 associates a belonging level to an Euclidean distance between a weight vector of a neuron of a category of a specific abnormal status and input data, and when the belonging level of the input data acquired during operation of the equipment X is not larger than a specified threshold, it is decided that the equipment X operates normally. COPYRIGHT: (C)2008,JPO&INPIT
|