A method for constructing a multi-fidelity proxy model of cutting force based on a bidirectional recurrent neural network

By constructing a multi-fidelity surrogate model for cutting force based on a bidirectional recurrent neural network, and combining mechanistic cutting force data and multi-source sensor signals, the problem of low accuracy in cutting force prediction in existing technologies is solved, and efficient cutting force prediction under complex working conditions is achieved.

CN122433488APending Publication Date: 2026-07-21DALIAN UNIV OF TECH +1
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
DALIAN UNIV OF TECH
Filing Date
2026-04-17
Publication Date
2026-07-21

AI Technical Summary

Technical Problem

Existing cutting force prediction methods suffer from low computational efficiency, strong dependence on high-fidelity measured samples, and insufficient fusion of multi-source signals under complex working conditions, resulting in low prediction accuracy.

Method used

A multi-fidelity proxy model for cutting force based on a bidirectional recurrent neural network is constructed. Low-fidelity data is generated through the mechanistic cutting force model. Combined with multi-source sensor signals and measured cutting force label samples, a DTCN time-series feature extraction module, a BGRF time-series feature enhancement module, and an extreme learning machine are used to achieve rapid prediction of cutting force.

Benefits of technology

It improves the accuracy of cutting force prediction under complex working conditions, reduces the need for high-fidelity experimental samples, and enhances the interpretability of the model and its ability to represent the dynamic response of cutting forces.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122433488A_ABST
    Figure CN122433488A_ABST
Patent Text Reader

Abstract

The application belongs to the field of cutting force prediction, and discloses a cutting force multi-fidelity proxy model construction method based on a bidirectional recurrent neural network. First, low-fidelity cutting force data is calculated through a cutting force mechanism model, and multi-source sensing signals in the cutting process and measured cutting force data under corresponding working conditions are collected; time series alignment and normalization processing are performed on various data, and a sliding time window is used to construct a multi-fidelity sample set; second, a cutting force multi-fidelity proxy model composed of a DTCN time series feature extraction module, a BGRF time series feature enhancement module and an extreme learning machine is constructed, and the model is trained and updated using sample set data; finally, low-fidelity cutting force data and multi-source sensing signals under a to-be-measured working condition are input into the trained model to realize rapid prediction of the cutting force.
Need to check novelty before this filing date? Find Prior Art