Method and device for detecting machining abnormalities

Abstract

To provide a machining abnormality detection method capable of reducing the number of model learning times when applying a machining abnormality detection model.SOLUTION: A machining abnormality detection method that obtains a machining abnormality degree indicating a degree of abnormality in cutting using machining information and a model, and that determines an occurrence of an abnormality in cutting when the machining abnormality degree exceeds a specified machining abnormality threshold, includes: determining, based on the machining abnormality degree, whether or not it is appropriate to use the machining information for the model to re-learn; accumulating only data that is suitable for re-learning among the machining information in a storage device; storing the machining abnormality degree corresponding to the accumulated machining information in the storage device; determining whether or not it is necessary for the model to re-learn based on a degree of similarity between a distribution of the machining abnormality degree being stored and a distribution of a degree of learning abnormality indicating an abnormality degree of the machining information used for the model to learn; and, when determining that the re-learning is necessary, causing the model to re-learn with the machining information accumulated in the storage device.SELECTED DRAWING: Figure 1

Images