基于卡尔曼滤波的机场噪声预测异构集成策略  

Heterogeneous ensemble strategy for airport noise prediction based on Kalman filtering

作　　者：徐涛[1,2] 苏瀚[1] 谢继文[1,2] 

机构地区：[1]中国民航大学计算机科学与技术学院,天津300300 [2]中国民航信息技术科研基地,天津300300

出　　处：《计算机工程与设计》2016年第2期512-518,共7页Computer Engineering and Design

基　　金：国家自然科学基金重点项目(61139002);国家863高技术研究发展计划基金项目(2012AA063301);国家科技支撑计划基金项目(2014BAJ04B02);中国民用航空局科技基金项目(MHRD201006;MHRD201101);中央高校基本科研业务费专项基金项目(3122013P013)

摘　　要：为有效解决机场噪声预测问题,提升预测结果的适应性以及准确性,提出一种基于卡尔曼滤波的机场噪声预测异构集成策略。将已有预测方法归纳为两类:时间序列预测和关联分析预测,根据现行噪声标准构建同时适用两类方法的训练数据集;采用卡尔曼滤波对两类方法的预测结果进行优化,将两类方法的结果集成到同一时间维度上对未来噪声进行预测。实验结果表明,该集成策略能将两类预测方法合理结合起来,显著提高预测准确率。To solve airport noise prediction effectively and improve the adaptability and accuracy of results,a heterogeneous ensemble strategy for airport noise prediction based on Kalman filtering was presented.Existing models can be classed into two groups：time series method and incidence analysis method,and training data sets were created for two methods according to the current noise standard.The prediction results were optimized by using Kalman filtering.The results of two methods can be integrated into time dimension to predict future noise.Experimental results show that this ensemble strategy can combine two prediction methods reasonably to improve the prediction effect significantly.

关 键 词：机场噪声预测 时间序列 关联分析 卡尔曼滤波 异构集成策略 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]