基于时空特征提取的空气污染物PM2.5预测

为了充分挖掘多因素数据间的时空特征信息，解决在多种因素相互影响下不能准确预测PM2.5值的问题，提出了一种融合了局部加权回归的周期趋势分解（Seasonal-Trend decomposition procedure based on Loess, STL）算法、卷积长短期记忆网络（Convolutional Long Short-Term Memory Network, ConvLSTM）和门控循环单元（Gated Recurrent Unit, GRU）的PM2.5预测方法。首先利用STL算法将PM2.5数据进行分解，将分解得到的序列分别与其他因素相融合；搭建ConvLSTM-GRU模型，并利用贝叶斯寻优算法进行超参数寻优；将融合数据传入ConvLSTM网络中进行时空特征提取，再将提取后的特征序列传入GRU网络中进行预测。通过与ConvLSTM-GRU模型、CNN-GRU模型以及GRU模型的预测结果进行比较实验，证明所提模型具有误差小、预测效果好等特点。

Prediction of Air Pollutant PM2.5 based on Time-space Feature Extraction

In order to fully mine the spatiotemporal feature information between multi-factor data, and solve the problem that the PM2.5 value cannot be accurately predicted under the influence of multiple factors, and proposes a PM2.5 prediction method that combines Seasonal-Trend decomposition procedure based on Loess (STL) algorithm, Convolutional Long Short-Term Memory Network (ConvLSTM) and Gated Recurrent Unit (GRU). First, use STL algorithm to decompose PM2.5 data and fuse the decomposed sequence with other factors; Build ConvLSTM-GRU model, and use Bayesian optimization algorithm to search for super parameters; The fused data is transferred to the ConvLSTM network for time-space feature extraction, and then the extracted feature sequence is transferred to the GRU network for prediction. Compared with the prediction results of ConvLSTM-GRU model, CNN-GRU model and GRU model, the proposed model has the characteristics of small error and good prediction effect.