Raman spectrum fermentation process model fast correction method based on tradaboost algorithm
By using the Tradaboost algorithm to quickly correct Raman spectral data, the problem of insufficient model prediction accuracy during fermentation is solved, and efficient real-time monitoring under different conditions is achieved.
CN117609747BActive Publication Date: 2026-07-21GUILIN UNIV OF ELECTRONIC TECH +1
View PDF 2 Cites 0 Cited by
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- GUILIN UNIV OF ELECTRONIC TECH
- Filing Date
- 2023-11-27
- Publication Date
- 2026-07-21
AI Technical Summary
Technical Problem
Raman spectroscopy models have insufficient prediction accuracy during fermentation due to changes in fermentation conditions and batch-to-batch differences, and conventional model adjustments are time-consuming and resource-intensive.
Method used
The Tradaboost algorithm is used to quickly correct Raman spectral data. Through data preprocessing and weight adjustment, multiple weak classifiers are integrated, and the model parameters are optimized by combining Bayesian optimization methods to ensure the accuracy of the model under different conditions.
Benefits of technology
It improved the model's prediction accuracy, reduced data collection time and cost, and enabled real-time monitoring under different batches and conditions.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN117609747B_ABST
Abstract
The application relates to the field of model correction and optimization, and discloses a rapid correction method for a Raman spectrum fermentation process model based on a Tradaboost algorithm, which comprises spectrum data preprocessing, initialization model training parameters and iterative training of the algorithm. Firstly, the Tradaboost algorithm is used to adjust the weight distribution of training samples, especially to increase the weight of the previous round of classification error samples, and to effectively integrate multiple weak classifiers. Then, a Bayesian optimization method is used to finely adjust the model parameters, so as to ensure the best performance. This makes the model quickly corrected under the condition of limited new data. This strategy not only improves the prediction accuracy of the model, but also significantly reduces the time and cost requirements of data acquisition.
Need to check novelty before this filing date? Find Prior Art