基于特征解耦与物理残差校准的仪器分析迁移方法和系统
By using feature decoupling and physical residual calibration of a dual-stream deep neural network model, the problems of low migration accuracy and lack of physical basis for calibration in instrument analysis migration are solved, achieving cross-device migration with low sample cost and high robustness.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SHANGHAI DEV CENT OF COMP SOFTWARE TECH
- Filing Date
- 2026-04-21
- Publication Date
- 2026-07-17
AI Technical Summary
Existing instrumental analysis transfer methods rely on a large number of labeled samples and fail to effectively consider the differences in instrument physical characteristics, resulting in low accuracy and insufficient robustness of transfer models, making it difficult to meet actual analytical needs.
A dual-stream deep neural network model is adopted, which extracts general chemical features through a shared feature branch and extracts response deviation features through a private deviation branch. By combining feature decoupling and physical residual calibration, the decoupling and calibration of instrument signals are realized, reducing computing power requirements and improving migration accuracy.
While reducing reliance on labeled samples, it improves the accuracy and robustness of cross-instrument analysis, achieves efficient transfer at low sample cost, and solves the problems of low transfer accuracy and lack of physical basis for calibration in existing technologies.
Smart Images
Figure CN122065017B_ABST