Research On The Application Of Intelligent Optimization Algorithm In Sky-Wave Over-the-Horizon Radar Location

The sky-wave over-the-horizon radar is free from the curvature of the earth via refraction the high-frequency sky-wave in the ionosphere.It can realize early warning and continuous observation of stealth fighters and low-altitude aircraft through its top-down detection.At present,the station of multi-station sky-wave radar not only realizes the high precision positioning of the target,but also improves the survivability of the system,which has important implications for the prevention and attack of foreign aggression targets.However,this way of positioning leads to the positioning model very complicated in solving the target position.The network which based on kernel function and extreme learning machine has excellent effect on solving complex non-liner fitting problem and it is suitable for dealing with the complex non-linear characteristic relationship between target position and the measured data from multi-static sky-wave over-the-horizon radar.This paper takes multi-static sky-wave over-the-horizon radar as the research subject.Aiming at the poor positioning accuracy of multi-static sky-wave radar by single-layer extreme learning machine and single-kernel extreme learning machine.Deep extreme learning machine and multi-kernel extreme learning machine are introduced to inform the non-linear characteristic relationship between Time Difference of Arrival,Azimuth and the target location to achieve the positioning of the detected target.Such is the specific contents:(1)Firstly,this paper detailed introduction of the sky-wave over-the-horizon radar and extreme learning machine.The basic principles of hybrid kernel extreme learning machine and deep extreme learning machine are expounded.Then,we introduce the mathematical location model and neural network positioning principle of multi-static sky-wave radar based on TDOA-AZ and multi-static sky-wave radar based on TDOA.(2)In the two-base sky-wave radar location method based on TDOA-AZ,aiming at the problem of poor positioning accuracy caused by single-kernel extreme learning machine which is failure to learn features well.Based on deep extreme learning machine which is formed by auto-encoder,the improved harris hawk optimization is used to optimized the number of hidden layer node and penalty coefficient of deep extreme learning machine.The optimized DELM can better learn the hidden features of input TDOA-AZ information.Simulation results shows that the improved model has smaller difference between the predicted target location and the actual target location.The fitting degree at each point is not only better than the location model of single-kernel extreme learning machine but also superior than the location model of DELM which is optimized by basic harris hawk optimization.(3)In the three-base sky-wave radar location method based on TDOA,it is not easy to determine the number of hidden nodes of extreme learning machine and the optimization training of weight and bias is time-consuming and poor accurate.A new position model using sparrow search algorithm to optimize the hybrid kernel extreme learning machine is proposed.With the help of the improved sparrow search algorithm to find the weight coefficients and kernel parameters,the optimized HKELM which has better non-linear mapping ability and robustness is used to locate the target detected by sky-wave radar.The results shows that the accuracy of improved sparrow search algorithm is not only superior to the HKELM location model which is optimized by the basic sparrow search algorithm,but also stronger than the extreme learning machine location model.(4)Finally,a location model of deep hybrid kernel extreme learning machine which is inherited the non-linear mapping ability of hybrid kernel extreme learning machine and feature learning ability of deep extreme learning machine is put forward.The method of two-base sky-wave radar positioning model based on TDOA-AZ which is optimized by improved harris hawk optimization and three-base positioning model based on TDOA which is optimized by improved sparrow search algorithm.In other words,the effectiveness of the method is proved.