Anomaly monitoring device using two competitive neural networks
View Patent ↗An anomaly monitoring device includes two neural networks which are switchable between a training mode by using training samples and a checking mode for classifying, based on a training result, whether an amount of characteristics obtained by an operation of an apparatus indicates that the operation of an apparatus is normal and a mode switching unit controlling one of the neural networks to operate in training mode and the other neural network to operate in the checking mode. Further, the anomaly monitoring device includes a switching determining unit computing a judgment evaluation value serving to evaluate reliability of a judgment result of the other neural network operating in the checking mode, and for instructing the mode switching unit to have the one of the neural networks operate in the checking mode and the other neural network operate in training mode when the judgment evaluation value does not meet evaluation criteria.
1. An anomaly monitoring device comprising:
a signal input unit for receiving a target signal generated by an operation of an apparatus;
a characteristics extraction unit for extracting an amount of characteristics having a plurality of parameters from the target signal;
two competitive learning neural networks each of which are switchable between a training mode for learning by using amounts of characteristics extracted from the characteristics extraction unit as training samples and a checking mode for classifying based on a training result from the training mode whether or not the amount of characteristics indicates that the operation of the apparatus is normal;
a output selecting unit for generating a classification result of the competitive learning neural networks;
a mode switching unit for operating each competitive learning neural network in one of the training mode and the checking mode wherein the neural networks operate in modes different from each other; and
a switching determining unit for computing a deviation (y) and a judgment evaluation value that is used in evaluating the reliability of a judgment result of the competitive learning neural network operating in the checking mode, and for instructing the mode switching unit to have said one of the competitive learning neural networks operate in the training mode and the other competitive learning neural network operate in the checking mode when the judgment evaluation value does not meet evaluation criteria.
2. The anomaly monitoring device of claim 1 , wherein the switching determining unit allows the amount of characteristics to be used as a training sample for the other competitive learning neural network operating in the training mode, only when the judgment result from said one of the competitive learning neural network operating in the checking mode is normal and meets the evaluation criteria.