Urban Air Quality Prediction Based On Hybrid Model

The quality of air quality not only directly affects the health of the public,but also affects the economic development of a city.Therefore,it is very important to predict the air quality accurately,effectively and timely.For the prediction of air quality,the traditional methods are usually numerical prediction or statistical prediction.In recent years,with the rapid development of machine learning technology,more scholars are more inclined to use machine learning methods to deeply mine historical air quality data and predict future air quality conditions.However,with the increase of the number of experiments,it is found that although the accuracy of using only a single machine learning model is greatly improved compared with the traditional model,there is still a big gap from the real value.In view of the above problems,this paper combines multiple models to establish a mixed prediction model,and uses the mixed model to predict the air quality index(AQI),further improving the accuracy of the prediction of the air quality index.The main work of this paper is as follows:1.Aiming at the prediction of individual AQI time series,the AQI prediction model based on EEMD-SE-ELM-GRU combination is established.Firstly,the EEMD algorithm is used to decompose the AQI data to obtain a group of intrinsic mode function components and residual components with different scales;Secondly,calculate the SE value of each component,recombine each component into a new sequence according to the SE value of each component,and predict the new sequence according to its complexity through GRU model or ELM model;Finally,all the results are superimposed to get the AQI prediction value.The results show that this model can predict AQI series more accurately than other prediction models.2.Aiming at the prediction of AQI sequence under the influence of multiple factors,a hybrid AQI prediction model based on CNN-Attention-GRU is proposed.Firstly,CNN is used to extract spatial features from the data,and then the Attention mechanism is further used to assign different weights to the extracted feature information to highlight the impact of key information on the prediction model.Finally,GRU network is used to predict and output the final results.In order to explore the impact of the correlation between data on the prediction results,Pearson correlation coefficient is used to screen the data.The final experimental results show that the CNN-Attention-GRU model is more accurate than other models and can better predict AQI sequence.