A peak recognition method and device based on a deep learning model and a medium
By using a peak identification method based on a deep learning model, the problems of noise misjudgment and low signal-to-noise ratio in spectral peak identification are solved, achieving high accuracy and high efficiency in peak identification, which is applicable to data processing such as gas chromatography, liquid chromatography, and mass spectrometry.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- ANHUI WAYEE SCI & TECH CO LTD
- Filing Date
- 2026-05-21
- Publication Date
- 2026-06-19
AI Technical Summary
Existing technologies suffer from noise misjudgment, low processing timeliness and accuracy in spectral peak identification, making it difficult to meet the high real-time and high accuracy requirements of modern analysis, especially when faced with low signal-to-noise ratio and complex waveforms.
A peak identification method based on a deep learning model is adopted. By acquiring and labeling a one-dimensional dataset, a one-dimensional convolutional deep learning model is constructed. Combining training samples and the verification process, the start and end points of peaks are identified, and negative peaks are flipped and raised to enhance the model's noise resistance.
It improves the accuracy and noise resistance of spectral peak identification, can identify complex situations such as peak clusters, has strong applicability, reduces computing power overhead, and improves the robustness and stability of computation.
Smart Images

Figure CN122241241A_ABST
Abstract
Citation Information
Patent Citations
Waveform analysis device
CN111373256B
A chromatographic peak recognition method based on convolutional neural network
CN111767790B
Data generation method and device, and recognizer generation method and device
CN115997219B
Negative peak detection method and device and readable storage medium
CN119959437A