Fast prediction method of thermal conductivity based on phenolic resin scanning electron microscope image
By constructing a three-dimensional model and a machine learning surrogate model of phenolic resin-based composite materials and extracting structural parameters using scanning electron microscopy images, the problems of speed and accuracy in evaluating the thermal conductivity of phenolic resin-based composite materials in existing technologies have been solved, and the prediction of thermal conductivity of small-sized samples and complex-shaped specimens has been realized.
Patent Information
- Authority / Receiving Office
- CN Β· China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- BEIJING INST OF TECH
- Filing Date
- 2026-04-13
- Publication Date
- 2026-07-10
AI Technical Summary
Existing technologies struggle to quickly and accurately assess the thermal conductivity of phenolic resin-based composites, especially in small-sized specimens and complex-shaped pieces. Furthermore, existing methods neglect the influence of nanoscale pores and particle structures, leading to significant discrepancies between calculated results and actual thermal conductivity.
By constructing a three-dimensional model of phenolic resin-based composite materials, extracting structural parameters using conventional scanning electron microscopy images, and combining this with a machine learning surrogate model, rapid prediction of thermal conductivity is achieved. Specific steps include constructing a sample set, image segmentation, skeleton extraction, and structural parameter calculation, which are then input into a trained regression model for thermal conductivity prediction.
It enables rapid and accurate prediction of the thermal conductivity of phenolic resin-based composite materials, applicable to small-sized samples and complex-shaped specimens, reducing sample preparation and calculation costs, and improving evaluation efficiency and accuracy.
Smart Images

Figure CN122368320A_ABST