Real-time inversion method for temperature field of key components of equipment based on coordinate bridge

CN122797241BActive Publication Date: 2026-11-03ZHEJIANG UNIV
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202611248093.9
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2026-08-18
Publication Date
2026-11-03
Estimated Expiration
2046-08-18

AI Technical Summary

Technical Problem

Existing methods for temperature field reconstruction suffer from problems such as high computational cost, sensitivity to boundary conditions, difficulty in representing the true temperature field with sparse measurement points, data scarcity, and weak generalization ability. In particular, it is difficult to achieve high-precision temperature field reconstruction in real-time or online tasks.

Method used

The optimal sensor layout is determined by a coordinate bridging method combined with intrinsic orthogonal decomposition and non-dominated sorting genetic algorithm. Temporal and spatial geometric features are extracted by a dual-stream spatiotemporal coding module, and temperature reconstruction is performed using a physical information-enhanced neural network. A physical equation loss function is introduced for training, thereby achieving high-precision real-time inversion of the temperature field under sparse measurement points.

Benefits of technology

High-precision real-time reconstruction of the temperature field was achieved under extremely sparse sensor conditions, which improved spatial resolution and physical realism, reduced hardware deployment costs, and enhanced the robustness of the model under unknown conditions and its real-time online inversion capability.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122797241B_ABST
    Figure CN122797241B_ABST
Patent Text Reader

Abstract

The application discloses a kind of equipment key component temperature field real-time inversion method based on coordinate bridge, it is related to temperature field reconstruction field, the method comprises: the simulation dataset of equipment key component is constructed, and the way of combining intrinsic orthogonal decomposition and non-dominated sorting genetic algorithm is used to the joint optimization of sensor quantity and space position;Based on optimal sensor layout acquisition time series temperature data;Based on time series temperature data and the continuous space geometry coordinates in the whole computational domain of equipment key component, generate time feature and space geometry prior feature using pre-constructed double-flow space-time coding module;The feature fusion mechanism based on coordinate bridge is constructed, to use the physical coordinates of each sensor in optimal sensor layout as space anchor point, fusion time feature and space geometry prior feature, obtain joint space-time feature, then using pre-trained physical information enhancement neural network carries out temperature reconstruction.The application can realize high-precision real-time reconstruction of temperature field under the condition of sparse measuring point.
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • Multi-scale cavity residual error and space-time coupling boiler temperature field reconstruction method

    CN122347087A

  • High-temperature component temperature field reconstruction method and system based on neural network

    CN122413912A