Research On Key Problem Of Target Parameter Estimation For The Large Scale Array Radar

In recent years,with the rapid development of stealth technology,such as absorption coating,structural absorbing material and low scattering profile,the meter-wave array radar with stealth detection ability has attracted attention.However,the traditional meter-wave array radar has low DOA/positioning accuracy,poor anti-jamming ability and the worse detection performance of multi-path sources,which is hard to realize high-precision detection.Increasing the antennas to construct Large-scale array radar can effectively improve array aperture,array freedom and anti-jamming ability,which become a vital way to improve the meter-wave array radar.However,in the application,there are still some problems such as the real-time processing problem of high-dimensional signals,low-altitude multi-path source localization problem,the mixed far-field and near-field sources localization problem.Aiming at above problems,this thesis studies from the aspects of reducing computational complexity,improving the parameter estimation performance and enhancing the adaptability of algorithm,and focuses on three scenarios: far-field point sources localization,low-altitude far-field sources localization and mixed far-field and near-field sources localization.The major research contexts are summarized as follows:1.In the parameter estimation of far-field point sources,with the increase of antennas in the large-scale array,the dimension of received data and the computational complexity of algorithm are enlarged.The compression measurement matrix is used to reduce the dimension of received data of large-scale array,which can effectively reduce the computational complexity.However,reducing the dimension of the received data will result in the performance loss of the parameter estimation algorithm.In order to find a reasonable measurement matrix,a measurement matrix optimization algorithm based on compressed sensing is proposed.Firstly,the optimization criterion is established by minimizing the overall coherence of the equivalent dictionary.Then,a closed-form based alternating minimization algorithm is used to optimize the measurement matrix.Finally,the optimized compression measurement matrix is used to design the physical structure and realize the dimensionality reduction of the received data.Since the lower the coherence between the equivalent basis,the better the recovery performance of the system,and the performance of compressed system is much closer to the performance of original system.Therefore,the algorithm can fundamentally reduce the computational complexity of large-scale array processing algorithms while ensuring the performance of algorithm parameter estimation.Additionally,a robust measurement matrix optimization algorithm is proposed to suppress noise by adding noise term constraints to the optimization function.2.In the parameter estimation of low-altitude far-field sources,the echo signals of the low-altitude sources are affected by the multi-path scattering.When the point source based estimation algorithm is used to estimate the low-altitude sources,the accuracy of algorithm will degrade or fail because the rank of the signal subspace exceeds the degree of freedom.In order to fix the low-altitude sources localization problem,a parameter estimation algorithm based on generalized non-circular characteristics is proposed for incoherently distributed sources.Firstly,the incoherently distributed source model is used to describe the multi-path low-altitude sources.Then,according to the relationship between strictly non-circular signals and circular signals,all received signals can be converted into strictly non-circular signals.Finally,the rotation invariance is constructed by subarray division,and the parameters estimation is accomplished by the generalized conjugate ESPRIT algorithm.Utilizing the non-circular characteristics of signals,the aperture and the performance of parameter estimator can be promoted.At the same time,the algorithm is search-free,which means it has low computational complexity and suitable for large-scale array.Additionally,to improve the parameter estimation performance of incoherently distributed source algorithm in low SNR,a parameter estimation algorithm based on cross-covariance decomposition is proposed,which can eliminate the uncorrelated noise by the cross-covariance of different subarrays.Then the estimation accuracy of the algorithm can be improved in low SNR.3.In parameter estimation of mixed far-field and near-field sources,due to the difference of range parameters between far-field targets and near-field targets,neither far-field sources estimation algorithms nor near-field sources estimation algorithms can handle the mixed far-field and near-field sources.To solve this problem,the received signal model is established by the concepts of plane wave and spherical wave.Moreover,a mixed sources localization algorithm is proposed based on discrete Fourier transform and orthogonal matching pursuit.In this algorithm,the anti-diagonal elements of the covariance matrix is extracted as a new data vector.The angular parameters of mixed sources can be separated and calculated by performing discrete Fourier transform on the new data vector and the received signal.Furthermore,the angular parameters of sources and the covariance matrix are used to construct a sparse reconstruction problem,which can estimate the range parameters of near-field sources with orthogonal matching pursuit algorithm.Since the discrete Fourier transform and the orthogonal matching pursuit are less affected by the SNR and the snapshots,the proposed algorithm maintain a high accuracy in low SNR and small snapshots scenarios.Meanwhile,the proposed algorithm does not require any high computational complexity matrix decomposition.Thus,it has low computational complexity and suitable for the application of large-scale array.In addition,to solve the performance degradation of estimation algorithm in the multipath scattering,a mixed far-field and near-field sources estimation algorithm is proposed based on discrete fractional Fourier transform.The spatial frequency information of the multi-path mixed sources can be extracted by performing discrete fractional Fourier transform on the received signals with one-snapshot.Then the angular parameters and range parameters of the mixed sources are calculated by the spatial frequency information.To avoid the grid error,an alternative searching algorithm is designed to realize fine searching.The algorithm can calculate the angular parameters and range parameters of multiple multi-path mixed sources,simultaneously.Therefore,in the proposed algorithm,the interaction between far-field parameters and near-field parameters can be avoided,which can further improve the parameter estimation accuracy.