Research On Array Calibration Method For High-frequency Ground Wave Radar

High-Frequency Ground Wave Radar(HFGWR) is a popular instrument for ocean state surveillance developed in last forty years. The ocean dynamic parameter inversion is based on the scattering of high frequency electromagnetic radiation from the ocean surface. Because the vertical polarized electromagnetic wave can propagate over the horizon on sea surface, HF radar can extract the surface current,wave and wind profile in wide area in real time.Ocean current and wave extraction performance is related to the single/multi-beam forming in HF radar operation. Narrow beam detection is obtained from the beamforming algorithm and direction-of-arrival(DOA) of the first order sea echoes is estimated by super-resolution method which is applied widely for current field estimation. Array signal processing is one of the important techniques in HF radar sea state monitoring. However, the existence of array errors deteriorates the performance of radar beam forming and DOA estimation, and affects the current and wave parameter inversion. With improper corrections, standard beam forming can result in failure steering of array beam, broaden main array beams, or high side lobes and direction finding algorithms would yield spurious directions and poor angular resolution. Real time array calibration is essential for HF radar long duration operation. This thesis is supported by "Eleventh Five-Year" National 863 Program project-"New Multi-Function HF Ground Wace Radar" and the Public Science and Technology Research Funds Projects of Ocean-"Smart HF Surface Wave Radar Industrializing Technology Research and Demonstrable Operation ". Our research focus on the antenna pattern error and gain/phase error in the the array calibration of HF radar. Antenna pattern distortion is estimated by using the ship echoes from various bearings and their real-time direction information from auxiliary instrument. An auto calibration method and its improved version which uses measured antenna pattern are developed to calibrate the gain and phase errors in the HF radar. The gain and phase mismatch is jointly estimated with the unknown DOA of the calibration sources in the auto calibration method. The errors are assumed independent to the bearings. The application of both methods in the Multi-frequency HF radar data and Smart array multi-frequency HF radar data is provided to verify their performance in error calibration and direction estimation. The thesis is divided into following aspects:1, The mathematical model of array receiving signal and array errors are introduced. The influence of present errors on direction estimation performance is briefly proposed.2, An active calibration technique using ships of opportunity to realize antenna pattern measurement is developed for HF radar. The ship echoes are detected from the radar data and their directions are obtained from Auto Identification System(AIS). Based on these preliminary information, Antenna pattern response is estimated through three steps, that is, Constant False Alarm Ratio (CFAR) ship detection, ship echoes preprocessing, AIS records and ship echoes matching and antenna pattern fitting estimation. Experimental application of this method in the original receiving data of multi-freuqency HF radar and Smart array multi-frequency HF radar is conducted to get the antenna pattern measurements in different operating conditions. Furthermore, the performance of direction finding considering array pattern distortion is compared with that of traditional MUSIC method after phase error calibration.3, For gain and phase error caliration in HF radar, three calibration methods are introduced in this paticular issue. Assuming with disjoint sources and nonlinear arrays, the cost functions are constructed according to three direction finding theory, that is, maximum likelihood, noise subspace fitting and signal subspace fitting.According to the principle of auto calibration, the jointly estimation of errors and directions is achieved through an solution of optimization problem. The details of optimal approach for each cost function are outlined and discussed. For the optimal solution is derived from the complex high dimensional nonlinear problems, lots of trials and efforts focus on decreasing computation burden and ensuring global convergence. Conventional methods are steepest descent method, Newton-Raphson method and quasi-Newton method which are based on the first-order approximation of proposed cost function. Besides, an iterative estimation constructer which estimates the errors and DO As' separately can save computation load and modern optimization algorithms are involved to improve convergence. This thesis puts forward a precise DOA estimating method and an initial method in order to improve the iterative procedure in computation and convergence. The DOA estimation is based on the derivation of reciprocal MUSIC spectrum using steepest descent method can decrease the computaiton loading effectively. The initial method accomplishes global convergence in large phase error condition by solving the optimal problem with additional active-calibration restriction. Convergence and estimation performance of proposed methods are verified and compared through numerical simulation.4, A phase error estimation method is proposed for HF radar using the first-order single-DOA sea echoes as the calibrating sources. Alternative methods for single-DOA sea echoes judgement, that is shift-invariant antennas method and eigenvalue gradient method, are compared and validated in the practical measured data. We provide four available schemes to calibrate the phase errors in the HF radar. Statistical performance is compared in a large number of experimental data, and we recommend to use eigenvalue gradient method to check single-DOA sea echoes and noise subspace fitting to estimate phase error which has compromise advantage in calibrating accuracy and stable real time operation. Practical application in experimental data of Multi-frequency HF radar and Smart Array multi-frequency HF radar confirms its effectiveness. Statistical performance are carried out in comparison with the active calibration method, which is based on the measured antenna pattern. Array direction finding performance before and after calibration is validated in details for different radar systems and operating conditions. Finally, an improved gain/phase error calibration method, which assuming the antenna pattern distortion is included in radar data, is provided. Compared with the gain and phase error active calibration method which using the auxiliary sources, This method can estimate the errors more frequent, and error estimation performance and direction estimation performance can be equal to the active method. It is concluded that direction finding performance after gain and phase error and antenna measurement can be improved with root mean square error decreasing to 1.1 degree for Multi-frequency HF radar and 2-5 degree for Smart array multi-frequency HF radar.