Research On Air Quality Evaluation And Prediction Based On Machine Algorithm

The relevant data of the WHO on urban air quality show that the concentration of pollutants in the air of more than 80%of the world’s cities is higher than the standard set by the WHO.In recent years,the state has attached great importance to environmental protection and issued a series of rectification measures.The air quality has been improved to a certain extent,but some new environmental problems have emerged.With the development of economy and technology,people’s living standards are constantly improving,and the types of travel tools are becoming more and more diverse.The emissions of NO_xand CO are on the rise,making automobile exhaust become the main factor of air pollution in some cities.These acid gases are oxidized in the air to form acid rain.Causing corrosive damage to soil and water resources,and serious smog pollution in some cities,which disturbs people’s daily activities and physical health.In view of the current problems that are not optimistic,adopting objective and effective air quality forecasting and monitoring has important reference value and guiding significance for vigorously carrying out comprehensive urban environmental management and air pollution prevention and control work.However,the content and process involved in prediction are relatively complex and have strong coupling with many parameters,which is still a challenge.As a category of artificial intelligence,machine learning is valued and favored by researchers in many technical fields due to its strong expressive power.It focuses on finding patterns in the data,and learns through training and analyzing a large amount of data.It can process data with very large samples,is robust to noise data,has strong fault tolerance,and has high classification accuracy,which can approximate any non-linear relationship.Strong learning ability.In this paper,a new prediction model is formed through the combination of intelligent optimization algorithms and neural network algorithms in machine learning.The air quality index AQI monitoring data of Taiyuan from 2018 to 2020 is evaluated and predicted.The main research contents of the article are as follows:(1)In order to evaluate,analyze and predict the air quality of Taiyuan,Shanxi Province initiated the first-level emergency response on January 25,2020 before and after the epidemic prevention and control,and the policy adjustment was adjusted to the second-level emergency response on February 24,using gray correlation The method of fuzzy comprehensive evaluation of the law evaluates the air quality trends before and after the epidemic prevention and control in Taiyuan,and analyzes the changes in related pollutant concentrations,hoping to provide a valuable reference for the evaluation and control of air pollution in Taiyuan.(2)The single-term prediction model has certain limitations,and can only reflect the local information characteristics of the research data,so that the prediction accuracy of the model is not high.This paper proposes a combined prediction model based on the ARIMA model and the Elman model through two different combinations of residual combination and weight combination.By comparing the experimental results,it can be obtained that the prediction accuracy of the two combined prediction models is significantly higher than that of the single model,and the prediction accuracy of the residual combination model is higher than that of the weight combination model,which effectively optimizes the prediction of the single prediction model Accuracy.(3)The AQI in Taiyuan was predicted using a Long-Term Memory Neural Network(LSTM)prediction model optimized by a stochastic gradient descent algorithm(Adagrad、Ada Delta、Adam)based on adaptive regulated learning rate.Comparing experimental results:all three stochastic gradient descent algorithms improve the prediction accuracy of LSTM neural network model,and LSTM neural network model based on Adam has the highest prediction accuracy and fastest convergence,so the model has higher feasibility.The model proposed in this paper:grey fuzzy comprehensive evaluation model for air quality evaluation before and after the epidemic in Taiyuan,ARIMA-Elman combinationmodel and Adagrad-LSTM、Ada Delta-LSTM、Adam-LSTM,for air quality prediction in Taiyuan,which provide a scientific and reasonable theoretical basis for air pollution prevention and control.