厦门市空气污染指数预测方法研究——时间序列分析与神经网络的比较(英文)

Study on Air Pollution Index Forecasting in Xiamen City——Comparison Between Time Series Analysis and Neural Network

分别使用时间序列分析和BP神经网络两种方法对厦门市思明区空气污染指数进行预测比较。将数据经零均值和平稳化预处理后,根据时间序列的自相关和偏自相关图进行判断,得到合适的时间序列模型为ARMA(2,2),然后进行参数估计得到模型的各个参数。在运用神经网络方法时,为了克服传统BP算法收敛缓慢和容易达到局部极值的缺陷,采用了LM算法,得到更快的收敛速度和更小的错误率。BP神经网络的隐层节点数和传递函数经过多次的训练学习,取得了较好的预测模型。利用得到的ARMA(2,2)和BP神经网络两种模型对厦门市思明区空气污染指数进行预测,结果显示BP神经网络方法的预测效果好于时间序列分析方法。

This paper deals with the study of Air Pollution Index (API) forecasting models through application of Time Series Analysis (TSA) and BP neural network techniques in Siming District, Xiamen. For choosing a suitable time series model, the paper makes the data zero mean and steady, and utilizes the character of the autocorrelation function and partial autocorrelation of the time series. In order to overcome the shortcomings of traditional BP algorithm as being slow to converge and easy to reach extreme minimum value, the BP neural network models adopt Levenberg-Marquardt (LM) algorithm to achieve a higher speed and a lower error rate. And the number of neurons and the type of transfer functions in the hidden layer of the neural network are changed from time to time to achieve the best forecasting results. Through simulation, the testing results show that TSA is less efficient than BP neural network.