Short Term Forecasting Of Evaporation Duct Height Based On The Time Series And Neural Network

The marine atmospheric duct is a kind of abnormal atmospheric refraction,which is closely related to meteorological conditions.Among them,the evaporation duct has a higher probability occurrence in the marine atmospheric duct which has an atmosphere node.It can bind specific frequency electromagnetic wave,propagating in the waveguide layer with a low energy loss.It can give full play to the combat effectiveness of radar and increase radar detection range.Its height value is an important parameter of radar to achieve over the horizon detection and remote communication technology.Therefore,mastering the height of evaporation duct in advance has a great significance to give full play to the advantages of naval warfare.In order to study the statistical regularities of evaporation duct of the South China Sea in this paper.We build a marine automatic meteorological station based on the South China Sea oil platform.The experimental data are transmitted through Beidou satellite communication.It can obtain real-time hydrological and meteorological elements for monitoring the evaporation duct.And then,the evaporation duct height can be diagnosed with NPS model,which is based on the Monin-Obukhov Near-stratigraphic similarity theory.At present,the evaporation duct can use MM5 or WRF,which is a mesoscale in the numerical simulation,to achieve regional prediction.But these methods require a lot of background field data and sea data which is hard to obtain.Moreover,the software configuration is complex and needs large amount of calculation.So,in this paper,we uses two methods,which are the time series analysis and BP neural network,to model and predict the temperature,relative humidity,wind speed,atmospheric pressure and sea surface temperature respectively.And then,combined with the NPS diagnosis model,which can calculated the forecasting value of the evaporation duct height indirectly.Finally,it can achieve short term forecasting of the evaporation duct height.For overcoming the shortcomings of BP neural network in modeling,the weights and thresholds of BP neural network is optimized by artificial bee colony algorithm.It can improve the convergence speed,prevent falling into the local minimum,and reduce the network prediction error of BP neural network.Finally,it can achieve a higher accurate prediction of the evaporation duct height.