基于SVR选择性集成的机场噪声预测模型研究

Airport Noise Prediction Model Research Based on SVR Selective Ensemble

机场噪声预测对机场规划设计、航班计划制定以及机场噪声控制具有十分重要的作用。针对机场周围各个监测点上的单飞行事件进行噪声预测。由于机场噪声数据的复杂性,用单一的SVR方法对其预测往往得出局部优化结果,不能达到理想的预测效果,针对这一问题,提出一种基于SVR选择性集成的机场噪声预测方法,通过Adaboost方法对机场噪声数据进行采样训练得到多个SVR预测模型,并结合一种排序方法对预测模型进行选择集成得到最终机场噪声预测值,取得了较好的预测效果。

Airport noise prediction plays an important role in airport planning,flight plan schedule and noise control. According to different monitoring points around airport,this paper aim to predict corresponding noise of individual flight event. For the complexity of airport noise data,prediction method which only applied single SVR would cause the problem of local optimum,and cannot get an accurate prediction result as expected. To solve this problem,an airport noise prediction method based on SVR selective ensemble was proposed in this paper. Adaboost method was used to airport noise data sampling,and then multiple SVR forecasting models were trained. With the help of a sorting method,forecasting models selective ensemble was achieved and used to predict the final airport noise value,proved has a good prediction effect.