Design And Implementation Of Aviation Radar Information Fusion System

With the air flying target becoming more and more intensive,the requirement of radar information fusion is becoming higher and higher in Civil Aviation Surveillance System.Nowadays,the Civil Aviation Radar Information Processing System uses the degenerate Kalman filter algorithm,the degenerate Kalman filter algorithm's gain is a fixed value,and different gain-values are needed in different aviation fields,while the selection of fixed values depend on a lot of prophase experiments.At the same time,the system only receives the highest level of data at each time according to radar priority,not taking full advantage of every radar's observational data.In order to improve the limitation of the fixed gain Kalman filtering algorithm in the filter,and to fully use the observation data of each radar,the methods presented in this paper are as follow.Firstly,the dynamic-gain Kalman filtering algorithm is realized by designing the state equation of the flight targets.And experimental results show that the algorithm is free from the dependence of the fixed gain-value in the different aero-fields,which is different from the Kalman filtering algorithm used at present,furthermore,it can be directly applied to each aviation field without affecting the filtering processes in real time.Secondly,select the three-layer BP neural network model,and train the neural network based on the relationship between the radar observation data and the flight target.And it is proved by experiments that this method can fuse the result which is coincident with the real trajectory of the flying target,in this way,the system makes full use of the radar data.Based on the algorithm,this paper designs three modules for the Aviation Radar Information Fusion System,including data preprocessing module,single radar filter module and multi radar data fusion module.The data preprocessing module is used to realize the resolution of radar message and the time and space alignment of multi radar data.The single radar filter module realizes the dynamic-gain Kalman filtering algorithm where the effect of noise on radar observation data is reduced.Multi-radar Data Fusion module realizes multi-radar information fusion based on BP neural network.In addition,in order to verify the effect of filtering and fusion algorithm,this paper designs the evaluation module besides the system.Finally,this paper realizes the design of the system with the help of SpringMVC framework.