Research On Space-time Series Model Of Airport Noise Based On Multi-point Monitor

The civil aviation promotes economic development and at the same time causes aiport noisepollution.Noise monitoring system is widely used in large aiport at home and abroad in responseto growing aiport noise pollution problem.Noise monitoring station located around airportmeasures and records the intensity of noise in real-time.Noise values of each monitoring stationconstitute a time series.Time series of different monitoring stations have relationship between eachother in space and time.In this paper,the theory and method of airport noise space-time seriesmodel is studied.The main research contents are as follows:(1) A new method for establishing spatial weight matrix is proposed, based on sample data.traditional methods is difficult to accurately reflect the interactions between spatial proximityvariables.Spatial weight matrix established by the new method not only has space property butalso has delay property.(2) STARMA-Network model has network structure and replace linear combination bynonlinear functions. STARMA model has only linear form and can not capture the nonlineartemporal autocorrelation structure.Airport noise space-time series prediction model is constructedbased on STARMA-Network model.(3) Space-time neural network is proposed using the function expansion technique and thelinear impulse response filter theory. Space-time neural network consists of Function expansionand Linear delay pulse.Net input space is mapped into a high dimensional space by Functionexpansion. Therefore, nonlinear mode in low dimensional space can be converted to linear modein high dimensional space. Linear delay pulse is equivalent to the temporal linear impulseresponse filter,which is responsible for fitting linear model in space-time series.