Research On Based Data Mining Of Airport Noise Prediction Method

In the aviation airport, the airport noise has been a common phenomenon. Since the air transportcompany was established, the airport noise has affected the environment surrounding the airport. WithChina’s rapid economic development, now many airports are established and all sorts of the largeaircrafts begin to appear, so the influence of the airport noise is growing more and more concern. Atpresent there have been many methods of more mature airport noise evaluation prediction, but thesemethods lack universality. A lot of prediction methods are all from the actual factors to predict, butthese tests need some practical conditions and the simulation in the airport again and again is unlikely.The noise of the monitoring data in the airport is less used for research, so the data doesn’t get to gooduse. Based on the above reasons, through the analysis of the monitoring data in the airport, the timeseries forecasting method is applied. Using the SVR, training the monitoring data and establishingthe prediction model, the prediction model is used to predict the future data, so it can get the futuretrend of development and this helps draw the contour map of the noise combination with the existingmodel.In this paper using the time series prediction method forecast the airport noise monitoring data,through the most commonly used support vector regression method, and compared to the artificialneural network method. Based on the support vector regression in time series prediction, thisalgorithm has a certain improvement in the application of the airport noise prediction, this enables itto be used better in the data like the airport noise. In the application process with the support vectorregression algorithm in time series prediction, this article does several further research, including datapretreatment, the weighted method and the parameters training method.In the process of using the algorithm to solve practical problems, it does some pretreatment forthe data, in order to make better applied in this algorithm. Before using the support vector regressionalgorithm, the data should do standardized treatment, but in the airport noise data the smoothingprocess is added. This mainly makes the data placid a bit, and doesn’t eliminate the trend of the data.So for the kind of data at first it should do the smooth processing, and then do standardized treatment,in order to apply the support vector regression algorithm.In time series forecasting the weighted method is aimed at the time order with each sampledifferent successively weights, and then dose the training process of prediction. However for the enterdata, in addition to the time order with all the sample each other, the attributes of the sample itselfalso have the time order for some data set, so for such a sample set, the sample itself also apply the weighted method.Cross validation is the most commonly used method of training parameters. In the time seriesprediction, through adding the time characteristics to cross validation method, it not only makes thetraining data greatly reduced, but also the prediction accuracy is improved a bit. Such it makes thecross validation method get the optimal parameters in the abundant sample data, and the predictionaccuracy also is improved.