Deep learning-based movable type printing plate number spraying robot collaborative control method

By constructing a topological map of the perforation space of the movable type printing plate and improving the dynamic anti-disturbance balance mechanism of the TDN model, the distortion problem of the control model for fluid regulation in the nozzle array of the movable type printing plate was solved, achieving efficient fluid deformation prediction and defect gradient tensor calculation, and reducing the scrap rate.

CN122452433APending Publication Date: 2026-07-24TIANJIN FENGHUA TAIYUAN TECHNOLOGY CO LTD +1
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
TIANJIN FENGHUA TAIYUAN TECHNOLOGY CO LTD
Filing Date
2026-05-06
Publication Date
2026-07-24

AI Technical Summary

Technical Problem

Existing classical fluid control algorithms, after increasing the density of the nozzle array in movable type printing plates and introducing the non-Newtonian fluid properties of alkali shavings, are unable to effectively predict multi-dimensional spatial deformation and accurately suppress edge diffusion and internal caking defects, resulting in structural distortion of the control model and an increase in scrap rate.

Method used

A deep learning-based collaborative control method for type printing plate marking robots is adopted. By improving the dynamic anti-disturbance balance mechanism in the TDN model, a topological map of the hole space is constructed, dynamic features are injected to generate a spatiotemporal evolution feature matrix, spatiotemporal reverse causal tracing and nonlinear topological hedging are performed, and the Lyapunov energy function is used to optimize the control of fluid pulse disturbances, thereby achieving efficient spatiotemporal deformation prediction.

Benefits of technology

It accurately filters out invalid fluid coupling interference and pulse fluctuation noise, improves the stability of fluid response and the accuracy of targeted defect gradient tensor calculation, and reduces scrap rate and material loss.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122452433A_ABST
    Figure CN122452433A_ABST
Patent Text Reader

Abstract

The application discloses a movable type printing plate number spraying robot collaborative control method based on deep learning, relates to the technical field of machine vision, and comprises the following steps: S1, outputting a hole space topology graph; S2, outputting a space-time characteristic matrix; S3, through improving a TDN model, based on a graph Laplacian spectrum transverse diffusion constraint and a dynamic anti-disturbance hedging principle, combinedly extracting fluid transverse diffusion and longitudinal accumulation evolution characteristics and dimensionally outputting a space-time deformation prediction tensor; S4, generating a targeted defect gradient tensor; S5, generating an inhibition instruction data packet; S6, outputting a anti-muddle character spraying ready signal; and S7, executing a spraying operation. The application overcomes the limitations of fluid response lag, physical space structure distortion and neglecting dynamic balance constraints of a traditional method, and provides an efficient solution for movable type printing plate number spraying robot collaborative control.
Need to check novelty before this filing date? Find Prior Art