| 摘要 |
In an apparatus anomaly monitoring system, a model creation module (2) creates an ensemble of models formed of predictive models of an objective variable (Y) for each explanatory variable (X) based on regression analysis with using a plurality of state data items (DS) measured from an apparatus (1) to be a target. With using this, a monitoring execution module (3) monitors a state of the apparatus (1) to detect the anomaly. More particularly, the explanatory variables (X) are categorized into a collinearity item (XA) and an independency item (XB) to create an individual model for each of the collinearity items (XA) with using the collinearity item (XA) and the independency item (XB). With using the ensemble of these models, a predicted value of the objective variable (Y), an error span between the predicted value and a measurement value, an ensemble error span, and others are computed.
|
