An ERP heterogeneous data detection method based on multi-dimensional feature fusion

By constructing a multi-dimensional feature fusion-based ERP heterogeneous data detection method, and utilizing a spatiotemporal causal topological network of visual nodes, thermal field nodes, and mechanical nodes, the method identifies the causes of defects, solves the technical bottleneck of multi-source heterogeneous data fusion and analysis, and achieves pixel-level spatial locking from quality results to the source of production, thereby improving the precision of data governance and the level of intelligence in the production process.

CN122390540APending Publication Date: 2026-07-14SHENZHEN GUMATE TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
SHENZHEN GUMATE TECH CO LTD
Filing Date
2026-04-21
Publication Date
2026-07-14

AI Technical Summary

Technical Problem

In the existing data processing technology system, there are significant technical bottlenecks in the fusion and analysis of multi-source heterogeneous data. It is difficult to accurately trace back and map the quality appearance of three-dimensional finished products to the physical property field of two-dimensional raw materials, resulting in a disconnect between the data logic of design, manufacturing, and quality, and making it impossible to achieve closed-loop optimization of process parameters and intelligent decision-making.

Method used

By extracting visual feature vectors as input to visual nodes, and using the first and second physical features as inputs to thermal and mechanical nodes respectively, a spatiotemporal causal topological network containing visual, thermal, and mechanical nodes is constructed using a graph attention mechanism to identify the correlation features between nodes, thereby identifying the causes of defects.

Benefits of technology

It achieves pixel-level spatial locking from quality results to the source of production, significantly improving the precision of data governance for manufacturing enterprises, laying a solid data foundation for digital quality control throughout the entire process, accurately tracing the root cause of each non-conforming product, and providing visualized physical evidence to guide the optimization of nesting spacing and path planning, thereby improving the level of intelligence in the production process.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122390540A_ABST
    Figure CN122390540A_ABST
Patent Text Reader

Abstract

The present application relates to the technical field of data processing, in particular to an ERP heterogeneous data detection method based on multi-dimensional feature fusion, multi-dimensional heterogeneous data associated with defective products is called, heat conduction distribution in the cutting process is reconstructed, and a dynamic heat accumulation field is constructed; based on global layout coordinates and spatial coordinate transformation, mapping back to the dynamic heat accumulation field, locking the defect corresponding area according to the visual defect data, and extracting the first physical feature; the relative angle between the product bending line and the rolling texture direction is obtained; the feature extraction is carried out on the pressure displacement curve, and the mechanical fingerprint is obtained; the relative angle and the mechanical fingerprint are nonlinearly coupled to generate the second physical feature; the space-time causal topology network is constructed, the visual feature vector is extracted as the visual node input, the first physical feature and the second physical feature are respectively input as the heat field node and the mechanical node; the correlation characteristics between nodes are recognized through the graph attention mechanism, and the defect reason is recognized.
Need to check novelty before this filing date? Find Prior Art