Method for monitoring tool wear state through enhanced comparative learning of Gramer angle field

Abstract

The invention discloses a method for monitoring the wear state of a cutter by means of enhanced comparative learning of a Grubrum angle field, which comprises the following steps of: acquiring cutting force signals in three directions in a cutting process, generating a gray-scale image from the cutting force signal in each direction through the Grubrum angle field, and superposing and combining the gray-scale images in the three directions into a color image; the method comprises the following steps: inputting a feature extraction weight into a comparison learning model to obtain a pre-training model, loading the feature extraction weight in the pre-training model into a classification model, carrying out learning training on the classification model by using a small number of labeled samples, and identifying the tool wear state by using the trained classification model. According to the invention, high-precision monitoring of the wear state of the cutter can be realized by using a small number of labeled samples and a large number of unlabeled samples, and the experiment cost is greatly reduced.

Description

technical field [0001] The invention relates to the technical field of machining process monitoring, in particular to a Gramm angle field enhanced contrast learning monitoring method for tool wear state. Background technique [0002] In the process of material processing, the tool is in direct contact with the workpiece. When the worn and dull tool continues to participate in cutting, the cutting force will increase, the cutting temperature will rise, and even vibration will occur, which will eventually reduce the surface quality and dimensional accuracy of the workpiece. Moreover, with the intensification of tool wear, it may lead to tool damage or damage, resulting in interruption of the cutting process, reduced production efficiency, and even serious accidents such as workpiece scrapping and machine tool damage. Therefore, real-time monitoring of tool status in uncertain machining environment is of great significance to ensure machining quality, ensure safe operation of m...

AI Technical Summary

Problems solved by technology

[0003] At present, in terms of tool wear condition monitoring (TCM), a lot of research has been carried out at home and abroad, and many efficient methods have been proposed, such as wavelet transform, support vector machine, etc., but these methods require certain conditions, and it is difficult to choose a suitable basis. Function or kernel function and corresponding parameters rely on a lot of prior knowledge

However, the popular deep learning methods in recent years require a large number of labeled samples for training. For TCM in actual industrial scenarios, the available labeled samples are limited and difficult to implement; and under limited samples, the neural network is prone to overfitting problems, and the model is generalized. Poor chemical ability

Images