Cutter wear condition monitoring method based on machine learning

Abstract

The invention belongs to the technical fields of accessories and auxiliary devices of milling machines, and discloses a cutter wear condition monitoring method based on machine learning. The cutter wear condition monitoring method uses cutting force and vibration signals as the cutter condition monitoring information, is multi-feature fusion cutter wear condition monitoring and prediction method,achieves effective monitoring of milling cutter states, and meanwhile, provides the basic data source and monitoring and prediction modules for a cutting milling database. The recognition accuracy ishigher whether the cutting force or the vibration signals conduct feature fusion, and the extraction feature of the feeding direction has higher recognition accuracy than that of other directions; andcompared with all extraction features and selected partial feature recognition accuracy, a correlation selection method playing the role of reducing dimension can improve the accuracy of classification and the robustness of the system, and an important role in the cutter monitoring system is played, regression analyzing is conducted on the cutter wear through neural network, a cutter wear monitoring model is established, and the particle swarm optimization algorithm is used to improve the performance and manifestation of the neural network.

Description

technical field [0001] The invention belongs to the technical field of accessories and auxiliary devices of milling machines, and in particular relates to a method for monitoring tool wear status based on machine learning. Background technique [0002] At present, the existing technologies commonly used in the industry are as follows: [0003] At present, tool state monitoring is mainly based on the direct monitoring method represented by optical monitoring and the indirect monitoring method represented by cutting physical quantity monitoring. The direct method is to measure the shape change or quality reduction of the tool during the cutting process through optical, isotope and other monitoring methods, and to monitor the tool status through image processing or mathematical models. The direct method is mainly used for off-line tool monitoring, and the main methods include optical method, contact method and machine vision processing. The direct method is highly accurate, ...

AI Technical Summary

Problems solved by technology

[0005] At present, the indirect monitoring technology still has problems such as relatively single monitoring physical quantity, poor data fusion level, low intelligence level of prediction model, and insufficient prediction accuracy.

The cutting database has problems such as single function and low level of intelligence, and the cutting data cannot be effectively used

Images