A Study On Air Quality Prediction And Space Distribution Based On Genetic Algorithms And BP Neural Networks In GuangZhou

With the rapid development of economy,a lot of pollution emissions,had a great damage to the environment,make the ecological deteriorating.The impact of air pollution on people's health not only,also will directly lead to the sustainable development of the society.As a result,air pollution prevention and control work,is particularly important.In order to more fully understand and grasp the trend of atmospheric pollutants,provide more comprehensive and timely information to air pollution prevention and control work,to carry out research work to predict atmospheric pollutants is very important.Environmental forecasting of atmospheric pollutants after decades of development,forecasting methods and forecasting techniques have been fully improved.However,how to improve the prediction accuracy of prediction of air pollutants is critical.In this paper,take Guangzhou as the study area,according to the Guangzhou City 2014 and 2015 in control site pollutant monitoring data and meteorological data,using genetic algorithm and BP neural network,to construct the air quality forecast model,the model in Guangzhou City air quality forecasting experiments were carried out.At the same time,using the inverse distance weighted interpolation method,the temporal and spatial distribution characteristics of AQI in Guangzhou city were analyzed.The main work and results of this paper are as follows:First,on Application of artificial neural network in air quality forecasting in the domestic and foreign research status are described,determine the research content and technical route of this paper,the geographical location of the study area of the basic situation and air quality monitoring stations are analyzed,also described the acquisition of air quality monitoring data and weather data of data content and data sources.Secondly,the BP neural network and genetic algorithm are deeply studied,the basic principle and process of the algorithm are introduced in detail,and the advantages and disadvantages of the two algorithms are analyzed.Using genetic algorithm has the advantage of global searching,to optimize the weights and thresholds of BP neural network and avoid the premature convergence of neural network to the local minimum value,and to enhance the generalization ability of the BP neural network is proposed in this paper.Again,according to the air quality data,considering the influence of meteorological factors,combined with genetic algorithm and BP neural network algorithm,the design and implementation of a neural network based on genetic algorithm optimization of air quality prediction model.And then,select the 2014 air quality monitoring data and meteorological data were used for model training sample data,train the prediction model,through repeated experiments to determine the neural network prediction model of the network structure and parameters and the prediction model is applied in Guangzhou City monitoring station by 2015 from January to March air quality prediction experiment in,and the results of the experiment and single BP neural network model of the experimental results are compared and analyzed,predicted experimental results show that the prediction model has better prediction accuracy and to achieve the desired effect.Finally,using the inverse distance weighted spatial interpolation method,the spatial and temporal distribution characteristics of AQI in the first half of 2015 in Guangzhou city were analyzed.The results show that the 1-3 months,the air quality index(AQI)showed a downward trend,in north than south.Between April and June,AQI was gradually rising trend,South than in the north and south of downtown AQI is significantly higher.And the prediction model to predict the spatial distribution of AQI value corresponding to contrast,both the space distribution features of relatively consistent,thus the accuracy of the prediction model is verified.