Wire disconnection prediction device

A wire disconnection prediction device for estimating a wire disconnection risk during machining of a workpiece in a wire electric discharge machine, the wire disconnection prediction device comprising:

• a data acquisition part configured to acquire data relating to machining of the workpiece during machining of the workpiece by the wire electric discharge machine;
• a preprocessing part configured to create, based on the data acquired by the data acquisition part, machining condition data of a condition relating to the machining commanded in the machining of the workpiece, machining member data relating to a member used in the machining, and machining state data during the machining of the workpiece, as state data indicating a state of the machining;
• and
• a machine learning device configured to execute, based on the state data created by the preprocessing part, processing relating to machine learning using a learning model indicating correlation between a machining state in the wire electric discharge machine and presence/absence of a possibility of disconnection occurrence of a wire electrode in the wire electric discharge machine and a disconnection cause by a plurality of class sets, wherein
• the preprocessing part is configured to create the state data and label data based on the data acquired by the data acquisition part, the label data taking, as label values, the presence/absence of the disconnection occurrence of the wire electrode at time when the state data is acquired and the disconnection cause at time when disconnection occurs,
• the machine learning device includes a learning part configured to generate the learning model, based on the state data and the label data, by creating the class sets corresponding to each of the label values taken by the label data, and
• the label data includes disconnection pattern data including a label indicating (1) presence/absence of disconnection occurrence of the wire electrode in a state where the state data is acquired and the disconnection cause estimated when the wire electrode is disconnected and (2) a machining condition adjusted before the disconnection,
• upon acquiring the data in the state where the wire electrode is not disconnected, the preprocessing part is configured to take a label value indicating that there is no disconnection, for the disconnection pattern data of the state data, and
• upon acquiring the data in the state right before the disconnection of the wire electrode, the preprocessing part is configured to take a label value indicating the disconnection occurrence and the machining condition adjusted before the disconnection, for the disconnection pattern data of the state data.