Prediction of Atmospheric Bioaerosol Number Concentration Based on PKO–AGA–SVM Fusion Algorithm and Fluorescence Lidar Telemetry
In order to realize early warning prediction of the distribution characteristics of atmospheric bioaerosol content, this paper proposes using fluorescence lidar as a technical means to establish a prediction model of atmospheric bioaerosol concentration by obtaining the observation data set of bioae...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2025-05-01
|
| Series: | Atmosphere |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2073-4433/16/6/638 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | In order to realize early warning prediction of the distribution characteristics of atmospheric bioaerosol content, this paper proposes using fluorescence lidar as a technical means to establish a prediction model of atmospheric bioaerosol concentration by obtaining the observation data set of bioaerosol concentration, combining it with the data set of atmospheric environmental parameters related to bioaerosol content, and utilizing the fusion algorithm PKO–AGA–SVM. The trained model was then used to predict the atmospheric bioaerosol concentration and compared with the bioaerosol concentration detected by fluorescence lidar to analyze the relative error of the model in predicting the bioaerosol number concentration with different algorithms as well as the bioaerosol number concentration at different pollution levels of atmospheric environmental quality. The experimental results show that the model prediction using the PKO–AGA–SVM fusion algorithm is better than the SVM, AGA–SVM, and PKO–SVM algorithms, with mean relative errors of 25.79, 20.75, 16.93, and 11.57%, respectively. Then, environmental data with different pollution levels were introduced for model prediction experiments, and the results show that the mean relative error of prediction was 12.75% when the air quality was excellent, the mean relative error of prediction was 13.01% when the air quality was good, the mean error of prediction was 10.53% when the air quality was mildly polluted, and the mean error of prediction was 13.72% when the air quality was moderately polluted. When the air quality was heavily polluted, the mean prediction error was 11.83%. The experimental results show that the prediction model has high accuracy and stability under different atmospheric conditions, which can provide a new research approach and technical support for the early warning system of atmospheric bioaerosol concentration. |
|---|---|
| ISSN: | 2073-4433 |