An Algorithm For Improving Error Registration Of Netted-Radar System Based On ANN

In recent years, multi-sensor data fusion technology has received extensive attention, its theory and method has been applied to many research areas. Multi-sensor data fusion technology is the basic principle of multiple sensors take full advantage of resources, and through observation of the multi-sensor information and the use of the reasonable control of the multi-sensor in space or time on redundant or complementary to the information based on a combination of criteria, object to be measured consistency of interpretation or description.First,the basic principle of data fusion is introduced in this paper, then the bad influence in data fusion for insufficiency of error registration is discussed after analyzing the system error in netted-radar . And the four kinds of primary spacial alignment algorithm were introduced and studied ,they are Real Time Quality Control method , Least Squares algorithm ,Maximum Likelihood method and Generalized Least Squares algorithm. Secondly to remove the error in coordinate transform , the Modified Least Squares was developed. And the simulation results prove it is valid. Because these algorithms mentioned above are based on the simple error model building on distance and angle error for they have the worst influence on target trace. The estimate of trace or predicting of trace still can not get better ,because many kinds of error do not be considered at the same time. To avoid the difficulty of computing in the complex error model, the error registration method with Neural Network(NN) on the base of Back Propagation algorithm was discussed. A new algorithm on the Inner Regress Network(IRN) we developed to avoid slow constringency of algorithm and weak generalizing ability .At last To compare the performance of error registration between LS and the new NN algorithm .the simulate under the condition of existing many kings of error including radar sector error and noise and these we mentioned was done, The simulated data is used to evaluate the performance and results show the new algorithm with IRN is more effective and reliable than LS algorithm.The spacial alignment in netted-radar system is studied in this paper, That is how to eliminate the system errors among those radars. There exist many kinds of error in netted-radar , such as the local error of radar site , azimuth reference error of angle, distance error and angle error in measure and error in coordinate transform and so on. The trace data fusion would be influenced for the existence of these system errors, and the veracity of trace is worse after data fusion, so it is necessary to correct the errors in multi-radar system.