【作者】 刘慧君；

【导师】 罗汉；

【作者基本信息】 湖南大学 ， 应用数学， 2014， 硕士

【摘要】 “雾霾”,是2013年最严重的空气污染现象,其产生的重要因素为PM2.5.PM2.5微粒,其来源广泛,成因复杂,为了更好地反映污染变化趋势,加强污染控制和严重污染事件的预防,开展污染治理工作刻不容缓.首先,本文利用了马尔科夫链对武汉市的PM2.5污染的演变规律进行了描述刻画.从PM2.5污染的时间序列变化讨论其发展趋势,通过其自相关系数的计算和比较,确定其马尔科夫链的步长,经检验建立了叠加马尔科夫链模型对武汉市的PM2.5污染的演变规律进行了预测,结果显示,武汉市的PM2.5污染从长期来看,出现无污染的可能性最大,其概率为0.6.其次,我们在找到武汉市PM2.5污染的演变规律之后,对于PM2.5污染的成因进行分析.我们主要通过PM2.5污染指标与空气质量指数的其余指标之间的关系来对二次颗粒物,也就是PM2.5污染中最重要的成因来进行分析,再利用逐步回归法对此问题做出定量分析.经广义差分修正后的逐步回归模型为：y*=-11.144+2.280x*co+0.752x*pm10-0.182x*o3+0.397x*no2最后,论文借助MATLAB软件的simulink工具箱,经过数据的预处理后对武汉市的PM2.5空气污染问题构建神经网络net来进行学习、训练和仿真,构建的网络net的预测准确率达88%.论文通过对武汉市PM2.5污染的演变预测及成因分析和仿真,为当今信息社会的城市空气污染的防治和预报工作提供了一种新的思路和方法,同时还找到了一种有效而又简便的仿真软件MATLAB.更多还原

【Abstract】 Atmospheric haze, the most severe air pollution in2013, is mainly caused by PM2.5. PM2.5particles have a broad range of source and a complex course of formation. In order to reflect variation tendency of pollution, it is of great urgency to reinforce pollution control and severe pollution prevention, as well as to launch pollution treatment projects.First, we used Markov chain to describe the developing pattern of PM2.5pollution in Wuhan City. We discussed this developing process from time series of PM2.5, and identified step length of Markov chain through calculation and comparison of autocorrelation coefficient. With inspection, we established a superimposing Markov chain model to predict developing pattern of PM2.5pollution in Wuhan City. The result showed that the possibility of PM2.5-pollution-free in Wuhan City is0.6on long terms.Second, after identified developing patter of PM2.5pollution in Wuhan City, we analyzed the cause of formation of PM2.5. We mainly analyzed secondary particle, which is the most important cause of PM2.5pollution, through the relationship between PM2.5and air quality index. The step-wise regression model after differential correction is:y*=-11.144+2.280x*co+0.752x*pm10-0.182x*o3+0.397x*no2Last, we used MATLAB software toolbox Simulink to construct a neural network model for the study, exercise and simulation of PM2.5pollution problem in Wuhan City. The rate of accuracy is88%.Through the research on the PM2.5pollution problem in Wuhan City, this paper provide new thoughts and methods to the forecasting and prevention of urban air pollution in an information society, as well as an effective, simple and convenient simulation software MATLAB.更多还原