Font Size: a A A

Research Of Air Pollution Tracing Based On Activation Force

Posted on:2019-06-14Degree:MasterType:Thesis
Country:ChinaCandidate:H Y ZhouFull Text:PDF
GTID:2321330545958463Subject:Information and Communication Engineering
Abstract/Summary:PDF Full Text Request
With the rapid development of economy and the continuous improvement of people's living standard,it also brings serious environmental pollution problems,especially in large and medium-sized cities,air pollution environmental problems have affected people's life in all aspects.In order to effectively control the air pollution problem,this thesis proposes to find out the source of air pollution is the reasonable and effective means of pollution control,and the core of this study is how to trace the source of pollution.Compared with the methods which usually use chemical reactions and diffusion model to study the problem of air pollution,this thesis from the data point of view,mining its internal relations through the temporal and spatial changes of data.In this thesis,a large number of pollution data in Beijing are acquired by the monitoring equipment installed on the taxi.The data is disposed by different periods and then divided into grids via longitude and latitude.After that,the data is graded for analysis.In our previous work,the Word Activation Force(WAF)model was used to model the polluted grid matrix.Then,the results of Activation Force and time series were combined to build the traceability link.Then,in view of this result is not ideal,this thesis puts forward using clustering processing to enhance the effect at first.After clustering processing,the link effect is indeed improved,but the desired result is not achieved.Therefore,this thesis proposes a new solution again.In recent years,neural networks perfonn well in a large number experiments,and the neural network has advantages in dealing with large-scale data compared with the traditional methods.Considering the data is closely related with the timing sequence,this thesis proposes air pollution modeling problems by the recurrent neural network.After demonstrating the effectiveness of air pollution grades forecast problem,this thesis innovatively combines recurrent neural network with the activation force model,solving the problem which builds the air pollution tracing links in a better way.
Keywords/Search Tags:tracing, air pollution, Word Activation Force, recurrent neural network
PDF Full Text Request
Related items