基于奇异谱分析的机场噪声时间序列预测模型

Prediction Model for Airport-noise Time Series Based on SSA

随着我国民航事业的不断发展,机场噪声问题日益严重。针对机场噪声时间序列预测问题,提出了一种基于奇异谱分析的噪声序列预测模型,即将机场噪声时间序列按照奇异谱分析预测的方法进行奇异值分解,得到主分量和经验正交函数,分析其趋势和振动的特点,然后选择适当的特征向量进行序列重构,通过线性重复公式建立预测模型。在此基础上,提出通过状态转移矩阵确定残差偏离方向,并根据残差的偏离方向和贡献率将重构模型忽略的次要成分计算进去,进而对预测值进行修正。在某机场实测数据中的应用表明,该方法的准确度明显优于已有SSA预测方法。

Along with the development of civil aviation in our country, the airport noises are getting more and more seri- ous. Aimed at the airport-noise time series prediction problem, this paper presented the prediction model based on SSA, in which the airport-noise time series are obtained by singular value decomposition, and the principal component and the Empirical orthogonal function are obtained, the characteristics of trend and vibration are analyzed and then the appropri- ate feature vectors are selected for sequence reconstruction, and prediction model is constructed by linear repeating for- mula. Based on the state-transition matrix and contribution ratio, the forecast values are revised. The experiment on the measured data of an airport shows that the accuracy of this model is better than other original prediction models.