| With the rapid development of science and technology,the demand for anti-jamming,multi-environment and real-time technology in the field of radar target localization is becoming higher and higher.In recent years,the rapid development of neural network algorithms and various fields has made it possible.This paper mainly studies the near and far field target localization based on the sparse array of traditional phased array and the target localization algorithm based on the combination of single-frequency receiving FDA and neural network.The angle information of the target is determined according to the echo information received by the array radar and distance information.The main work of this article is as follows:1.Aiming at the problem that the localization accuracy of phased array targets is not high in the scene of near and near field targets.A near-field and far-field target localization method based on sparse array is proposed.This method uses a sparse array model to expand the array aperture and effectively improve the problem of array aperture loss.First,by selecting the received data of the special points of the three sub-arrays and the received data of each sub-array to calculate the fourth-order cumulant,a matrix related only to the angles of the far and near field target signals is obtained.After decomposing the matrix eigenvalues,MUSIC algorithm is used to perform spectral peak search to obtain the angle of the far and near field targets.Then use it as the known information,and use the MUSIC algorithm to obtain the distance information for the near field targets.This method uses twice one-dimensional MUSIC algorithm not only to reduce the complexity of algorithm calculation but also to increase the array aperture and improve the accuracy of target localization.2.The phased array antenna pattern depends only on the angle and has nothing to do with the distance.This leads to the problem that it cannot measure distance effectively in the case of distance ambiguity.For this problem,the FDA radar is used to achieve the distance and angle localization by using the unique beam pattern characteristics that depend on the distance.In order to enhance the ability to adapt to the environment,combining FDA radar and BP neural network,a Frequency Diversity Array target localization method based on improved Particle Swarm Optimization(IPSO)-BP neural network was proposed.This method uses the improved particle swarm optimization algorithm(IPSO)with nonlinear weights to optimize the BP neural network(IPSO-BP)weights and biases,so that the problem that BP neural network is easy to fall into local minimum points is solved.The target location model based on FDA-IPSO-BP neural network was established and trained by us,which avoids the problem of considering the FDA angle and distance decoupling.The computer simulation experiments show that the method has better target localization effect,and the convergence speed of the algorithm is effectively improved,which proves the validity and reliability of the method.3.Although the FDA-BP algorithm has improved the complexity and accuracy of target localization,the commonly used method to improve accuracy is to increase the sample data.However,in the case of large sample data,the fully connected BP-FDA algorithm cannot solve fast and effective calculations and is prone to overfitting.To solve this problem,this paper proposes a Frequency Diversity Array target localization method based on CNN.This method uses the CNN algorithm combined with the FDA beam characteristics to decouple the distance and angle of the target,and directly obtain the target distance and angle information.And the function of weight sharing,local connection and pooling of this method reduces the possibility of overfitting caused by huge sample data,and can quickly realize parallel computing.While this method improves the target localization accuracy,it also greatly improves the real-time performance.Computer simulations are implementing to verify the effectives of the proposed algorithm. |
PDF Full Text Request