摘要
传统的大气污染物预测方法是基于气象资料对数据做线性处理的过程,难以对大气环境数据进行较深入的挖掘。以西北某市2002年NO2小时浓度序列为例,基于Matlab应用技术平台,经分段三次Hermite插值处理后,在小波消噪的基础上,利用相空间重构的结果构造神经网络模型对该时间序列进行预测。仿真实验表明,该预测方法应用于大气污染物浓度时间序列的分析是可行,能够较准确地预测大气污染物浓度。
Traditional approaches of atmospheric pollutants prediction which are based on meteorological data are linear process,but are difficult to work deeply. In this paper, the hourly concentration of NO2 of a city in northwest of China was taken as an example, using Matlab to built a model for predicting. The Neural Network Model bulit by phase space reconstruction was used to predict the time series. Experimen- tal results show that such a method is feasible in concentration time series of atmospheric pollutants and a- chieved a reasonable predicting result.
出处
《西南科技大学学报》
CAS
2013年第3期24-27,39,共5页
Journal of Southwest University of Science and Technology
