Abstract:

The airport noise data collected by monitoring nodes may be abnormal caused by damaged or aging equipment. In order to revise these abnormal data,an airportnoise associated prediction model,which is based on observational learning algorithm and considers the correlation between monitoring nodes,is proposed.Firstly,the high correlational monitoring nodes for abnormal nodes are filtered out by measuring the correlation between the failure nodes and the normal nodes.Then,the ensemble BP neural network is used to build the model.To solve the under-fitting problem caused by small samples and to improve the generalization performance of the model,we also propose a weighted observational learning algorithm,in which the weights are measured by learning outcomes.The application in the measured data of an airport shows that the proposed model has better predictive ability,and the algorithm is more stable and effective than the standard observational learning algorithm.

Key words: airport-noise monitoring;airport-noise prediction;associated prediction;observational learning;BP neural network;ensemble learning