Multi-Task Aqi Prediction Method Based on NLSTM

Air quality prediction is the main method for air pollutant monitoring, and air quality (AQI) prediction based on machine learning is also a hot topic in current research. Current research mainly uses deep learning models to predict single pollutants. There is a problem that the mutual influence between different AQIs leads to unsatisfactory prediction results for single pollutants. To address the above problems, this paper proposes a multi-task AQI prediction method based on NLSTM (Neural Long Short-Term Memory). First, each pollutant is input into the model at the same time, and processed by a one-dimensional convolution layer to reduce the dimension of the input data, reduce the computational complexity and memory usage. Secondly, the output of the convolution layer is processed by the NLSTM layer, which can fully capture the temporal relationship between pollutants and improve the performance of the model through data preprocessing. Finally, the Pearson correlation coefficient is used to construct the correlation matrix between different pollutants. The multi-task AQI prediction method based on NLSTM is verified on the original air quality index (AQI) dataset of 12 observation stations around Beijing. The mean absolute error (MAE), root mean square error (RMSE), mean absolute percentage error (MAPE) and <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{R}$</tex> square (<tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{R}^{\mathbf{2}}$</tex>) are all improved.