Research On Adaptive Correction Method For Radar Network System

Systematic error registration is one of the most important problems in multi-source information fusion process of multi-sensor information system,such as Radar network.This dissertation systematically studies the principle and method of the registration algorithms in radar networking system.At present,the two-dimensional spatial registration algorithms based on stereographic projection method that have more applications in engineering,usually is not suitable for long-range registration,and cannot estimate the pitch error.A three-dimensional spatial registration algorithm based on ECEF coordinate system is introduced to solve this problem.By constructing a mathematical model,we propose a three-dimensional spatial registration algorithm based on genetic algorithm,and compared with the generalized least squares algorithm based on ECEF coordinate system.Simulated data are used to verify the effectiveness of the algorithm.This algorithm overcomes the disadvantage of linearization and the dependence on the spatial distribution of the registration target,improves the adaptive ability and automation ability of the registration,has potential engineering value.At last,we point out the deficiency of real-time in the simple genetic algorithm for radar network registration.We improve the simple genetic algorithm by combining with the heuristic search techniques,such as niche technology,interpolation method and immune algorithm,so as to improve the global optimization ability and the convergence speed.Simulation results confirm the effectiveness of the algorithm,and to a certain extent,the accuracy and time efficiency of the algorithm are improved.