Scholarlink

This study presents an improved air quality prediction method using hybrid deep learning approach named PM25-CDSCBL, which is an enhancement of the PM25-CBL model. The PM25-CDSCBL combines CNN (Convolutional Neural Network) with a Depthwise Separable Convolutions layer and Bi-LSTM (Bidirectional Long Short-Term Memory). The model is evaluated and compared with other methods such as LSTM, Bi-LSTM, CNN-LSTM, ARIMA and PM25-CBL on the HCMC Air Quality 2020 dataset. Experimental results show that PM25-CDSCBL achieves superior performance with the lowest MSE (Mean Squared Error), RMSE (Root Mean Squared Error), MAE (Mean Absolute Error), and MAPE (Mean Absolute Percentage Error) among all tested models. Notably, the PM25-CDSCBL model demonstrates high accuracy in predicting PM2.5 concentrations across time steps, with a close match between actual and predicted values. These results confirm the feasibility and effectiveness of PM25-CDSCBL in air quality prediction, especially in the context of increasing urbanization and air pollution in major cities.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

PM2.5 Prediction Model Using CNN with Depthwise Separable Convolution and Bi-LSTM

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Thi-Phuong-Trang Nguyen,
  • Duc-Cuong Nguyen,
  • Tuong Le

摘要

This study presents an improved air quality prediction method using hybrid deep learning approach named PM25-CDSCBL, which is an enhancement of the PM25-CBL model. The PM25-CDSCBL combines CNN (Convolutional Neural Network) with a Depthwise Separable Convolutions layer and Bi-LSTM (Bidirectional Long Short-Term Memory). The model is evaluated and compared with other methods such as LSTM, Bi-LSTM, CNN-LSTM, ARIMA and PM25-CBL on the HCMC Air Quality 2020 dataset. Experimental results show that PM25-CDSCBL achieves superior performance with the lowest MSE (Mean Squared Error), RMSE (Root Mean Squared Error), MAE (Mean Absolute Error), and MAPE (Mean Absolute Percentage Error) among all tested models. Notably, the PM25-CDSCBL model demonstrates high accuracy in predicting PM2.5 concentrations across time steps, with a close match between actual and predicted values. These results confirm the feasibility and effectiveness of PM25-CDSCBL in air quality prediction, especially in the context of increasing urbanization and air pollution in major cities.