Paper-plastic mold dynamic positioning method and system based on AI visual detection

By combining AI visual inspection and LSTM prediction models, the positioning deviation problem caused by thermal deformation and servo delay in the paper-plastic mold positioning system was solved, realizing accurate adaptation and stable control of mold dynamic positioning, and improving production quality and efficiency.

CN122199665APending Publication Date: 2026-06-12FOSHAN CITY MEIWANBANG TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
FOSHAN CITY MEIWANBANG TECH CO LTD
Filing Date
2026-02-24
Publication Date
2026-06-12

AI Technical Summary

⚠Technical Problem

The existing paper-plastic mold positioning system has insufficient positioning accuracy in handling the reciprocating motion of the mold. This is mainly due to the coupling effect of thermal deformation and servo execution delay, which leads to uncontrollable deviation of the positioning reference, affecting production stability and product qualification rate.

⚗Method used

A dynamic positioning method based on AI vision detection is adopted. The mold data is analyzed collaboratively by subpixel-level image segmentation, template matching and temperature-deformation correlation model. The positioning deviation is predicted and the servo execution delay is compensated by combining LSTM prediction model to construct dynamic positioning benchmark correction model and form closed-loop control.

🎯Benefits of technology

It achieves precise compensation for thermal deformation and servo delay, improves the accuracy and stability of mold dynamic positioning, and ensures product quality and production continuity.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122199665A_ABST
    Figure CN122199665A_ABST
Patent Text Reader

Abstract

The application discloses a paper-plastic mold dynamic positioning method and system based on AI vision detection, and the method comprises the following steps: synchronously collecting vision image data, surface temperature data, motion state data, positioning deviation data of the previous 25 motion cycles and positioning reference values of mold reciprocating motion; obtaining real-time positioning deviation values by adopting sub-pixel level image segmentation algorithm, template matching algorithm and temperature-deformation correlation model collaborative analysis; combining servo execution delay time to construct a dynamic positioning reference correction model to output corrected positioning reference values; and predicting the positioning deviation of the next motion node by an LSTM prediction model and forming a closed-loop control. The system corresponds to execute the above method. The application solves the core problem of low dynamic positioning precision in the prior art, realizes accurate dynamic positioning adaptation, and improves positioning stability.
Need to check novelty before this filing date? Find Prior Art