Cutter wear state intelligent identification method based on heterogeneous domain adaptive transfer learning

Abstract

The invention discloses a tool wear state intelligent identification method based on heterogeneous domain adaptive transfer learning. The method comprises the following steps: constructing a tool wear state identification system based on heterogeneous domain adaptive transfer learning; acquiring source domain data S and target domain data T according to different wear curves of several cutters, and performing feature extraction and feature dimension reduction on the data; constructing an MMD matrix M, and initializing parameters and the maximum number of iterations; initializing an FWELM random input weight and calculating a hidden layer output matrix H of the FWELM random input weight; utilizing the DST-FWELM to calculate a reconstruction output weight; calculating reconstructed tool wear source domain data S'and target domain data T '; using the reconstructed source domain data to train an adaptive FWELM classification model; updating the target pseudo tag and the condition matrix Mk by using an adaptive FWELM classification model; and after the maximum number of iterations is reached, utilizing the final self-adaptive FWELM classification model to predict the tool wear state.

Description

technical field [0001] The invention belongs to the field of tool wear monitoring of numerical control manufacturing equipment, in particular to an intelligent recognition method of tool wear state based on heterogeneous domain self-adaptive migration learning. Background technique [0002] The tool is the direct executor of processing and manufacturing. The increased tool wear will lead to an increase in cutting force, an increase in the surface roughness of the workpiece, and the size of the workpiece exceeds the tolerance requirements, and even cause the processing to stop, reducing the processing efficiency. Tool condition monitoring technology can grasp the tool wear status in time, which has important and far-reaching significance for improving the processing quality and surface accuracy of workpieces, improving product economic benefits, and saving processing time. In order to find a better monitoring method, pick up the original signal that is closely related to tool...

AI Technical Summary

Problems solved by technology

However, this method needs to extract a large amount of training data, which increases the training time, resulting in insufficient real-time performance of the model, and this method cannot monitor different types of tools, which has great limitations

[0012] In view of the above problems, the present invention intends to design a tool wear state intelligent recognition method based on heterogeneous domain self-adaptive migration learning, which uses the maximum mean discrepancy (MMD) algorithm and fuzzy wavelet extreme learning machine (Fuzzy wavelet extreme learning machine , FWELM) model, combined with the idea of ​​domain migration, effectively solves the problem of single identification target and insufficient accuracy in the current tool wear state identification system in the machining process

Images