Method for detecting pollutant concentration through deep ultraviolet Raman spectrum based on deep learning
By constructing an SSA-CNN-LSTM-Attention fusion model, the problems of spectral line overlap, weak trace signals, and nonlinear response in the detection of new pollutants by deep ultraviolet Raman spectroscopy are solved, realizing high-precision and highly anti-interference quantitative analysis of concentration, which is suitable for rapid detection of environmental and industrial samples.
Patent Information
- Application Number
- CN202511187288.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-08-25
- Publication Date
- 2025-11-14
AI Technical Summary
Existing deep ultraviolet Raman spectroscopy techniques face challenges in detecting the concentration of new pollutants, including spectral line overlap, weak trace signals, complex background noise, and nonlinear response, making it difficult to achieve high-precision and robust quantitative concentration analysis.
A fusion model of SSA-CNN-LSTM-Attention is constructed, which combines spectral preprocessing and synthetic data augmentation strategies. It extracts local spectral features through 1D-CNN, captures long-range sequence dependencies through LSTM, enhances the feature peak response weights through the attention mechanism, and optimizes hyperparameters through SSA, thereby solving the problems of spectral line overlap interference, weak trace signals, and nonlinear response.
It achieves high-precision and highly interference-resistant quantitative analysis of pollutant concentrations, improves the anti-interference capability and generalization of new pollutant detection, and is suitable for rapid and high-precision detection of environmental and industrial samples.