| Studying and developing the detection technique for high speed and accelerationtargets and the signature extraction algorithm for target with micro-motion are of greatimportance to improve the capability and extend the function of radar. The high speed andacceleration of targets induce range and Doppler migration, which pose new challenges toradar detection. Therefore, it is necessary to appropriately compensate the migration spriorto the process of target detection. The complicated attitude changes in the course of targettranslation such as oscillation, rotation and precession make the radar detection moredifficult. However, the uniqueness of micro-Doppler generated by micro-motion offers anew avenue to target identification. The substance of this paper is an in-depth investigationof the coherent integration algorithm for the targets with high speed and acceleration andthe extraction algorithm for the target’s micro-Doppler time-frequency signature based onnarrow band linear frequency modulation pulse radar system, which is supported by relatedresearch projects. The detailed research contents are as follows:1. Firstly, the trajectory and echo of high speed targets such as missile and projectileare modeled, and the relationship between range/Doppler migration and radar/targetparameters are analyzed. The influences of migration upon resolution and integration gainare evaluated quantitatively. It is found that when range or Doppler migration crosses morethan two resolution bins, the integration gain loss is over3dB. Thus migrationcompensation is necessary before integration detection is performed. The results mayprovide theoretical guidance to the study of radar detection of high speed and accelerationtargets and the selection of migration compensation algorithms.2. For the scenario of range migration associated with high speed targets, threemigration compensation algorithm architectures are suggested. After analyzing of theenvelope shift algorithm, a range migration compensation algorithm based on envelopeshift and scaling processing is developed. Envelope shift algorithm is used to shift inintegral multiples of sampling interval whilst scaling processing is used to improve theenvelope alignment accuracy. When used in combination, they may reduce computationalcomplexity and improve the detection performance. Also, given the rigorous requirementson the algorithm’s performance in real-time posed by actual operations in the case ofrelatively high SNR, an approach of Hough transform based on parallel processing isproposed to accomplish the high-speed target detection. In this approach, the echo matrix is decomposed into submatrices, and then the Hough transform of the entire echo matrix isactualized by superposition principle after the segmented Hough transformation isconducted. To reduce the runtime, the digital signal processing (DSP) technology can beemployed for the parallel processing of this approach. By decomposing the flexiblecombination of the submatrices, the local anti-disturbance capability of the algorithm canbe improved. Finally, the effectiveness of the proposed algorithm is verified by simulationand analysis.3. In consideration of Doppler frequency migration associated with accelerating targets,after analyzing DPT (discrete polynomial phase transform) and FRFT (Fractional Fouriertransform), a DPT-FRFT based compensation algorithm is proposed. Modified DPTalgorithm is used to obtain a rough estimation of frequency modulated (FM) rate whileFRFT is employed to provide quick search and accurate estimation of FM rate parameters.Ina high signal to noise ratio (SNR) situation, this technique has an estimation performanceapproximate the Cramer-Rao lower bound (CRLB), but its detection performance degradeswhen the input SNR drops. To this end, a detection technique based on segmented DPT(SDPT) is proposed where the input signals are divided into segments and coherentintegration is performed before DPT and FRFT operations among the segments. Theoreticalderivation and simulation indicate that SDPT achieves an SNR higher than that ofDPT-FRFT and DPT, which will benefit the radar detection performance for the weakaccelerating targets. Secondly, a joint migration compensation technique is introduced forthe coexistence of range migration and Doppler migration of high speed and accelerationtargets. This technique significantly reduces the computational complexity while fulfils thetarget detection requirements. Finally, a multi-frame signal detection scheme for the highspeed and acceleration target detectionhas been formulated on the basis of multi-livingagent theory, and the effectiveness and robustness of this scheme is verified by real data.4. Firstly, a micro-motion model is constructed for targets such as projectiles andhelicopters, and the mathematical expressions of micro-Doppler are given for variousmicro-motion signatures. Secondly, two typical micro-Doppler time-frequency analysistechniques are reviewed. It is found that the Wigner-Ville distribution (WVD) basedmethod has poor time-frequency analysis results due to the influence of autocorrelationfunction cross terms. Meanwhile, the short-time Fourier transform (STFT) based method,though not affected by cross terms, cannot achieve high resolution both in time andfrequency domain. In this context, a time-frequency analysis technique based on short-time fractional Fourier transform (STFRFT) is used and a mathematical expression is given forcomputing STFRFT time-frequency resolution, which facilitates the quantitativelycomparison of time-frequency resolutions. Secondly, taking account of themulti-components in the micro-motion target in practical, an algorithm that combinesmulti-order STFRFT with data fusion for the micro-Doppler time-frequency analysis isproposed. The time-frequency curve fitting of the multiple micro-motion components isaccomplished by means of multi-order matching, which improves the estimation accuracy.Finally, simulation experiment data demonstrate the validity and advantage of the presentedalgorithm.5. An experimental system is built to acquire real data for the algorithm verification.Firstly, the migration compensation algorithms proposed in2and3are verified. Thenumerical experiment with measured data indicate that when the migration compensation isconducted for the data from the detection scenarios of the civil aircraft and a particularrocket projectile, the SNR increases about5dB and11dB respectively than that of movingtarget detection (MTD) technique. When joint migration compensation is applied to theprojectile data of two different calibers, the SNR increase is up to about5dB whencompared with MTD. Secondly, the micro-motion feature extraction methods mentioned in4is proven to be effective. The test data show that, for the three types of projectile:howitzer, mortar and rocket shell, the extracted micro-motion signatures can assist thepreliminary classification of the projectile targets; for the test data of helicopter and fan, theextracted micro-motion signatures can assist to ensure the number of the blades and toestimate the rotation rate. |
PDF Full Text Request