Research On Robust Adaptive Detection Algorithm For Radar Distributed Targets

With the continuous advancement of chip technology and signal processing methods,radar resolution has significantly improved.Traditional radar point targets can now be decomposed into spread targets containing numerous discrete scattering centers across multiple dimensions such as range,Doppler,and azimuth.Spread targets not only provide abundant information about target shape and motion but also effectively reduce fluctuations in target echo amplitude.Therefore,by carefully designing detection algorithms for radar spread targets,the perception ability of radar for targets in complex detection environments can be significantly enhanced.However,constructing a universally optimal spread target detector is a challenging task in the actual radar working environment due to the diversity of target scattering characteristics and the complexity of clutter.Currently,a common approach is to design suboptimal detectors based on prior knowledge of target scattering and clutter statistical characteristics,taking into account specific application scenarios.Nevertheless,these methods often suffer from insufficient robustness.Once the prior knowledge deviates from the actual situation,detection performance can be affected and drastically reduced.To address the shortcomings of existing research,this dissertation conducts an indepth exploration of robust adaptive detection algorithms for radar spread targets,aiming to improve radar’s target perception capability in complex environments.Specifically,this dissertation focuses on studying robust adaptive detection algorithms for single spread targets under conditions of unknown target scattering characteristics and missing clutter secondary data.Additionally,it explores detection algorithms for adjacent multiple extended targets.The main contributions of this dissertation include:(1)To address the detection of range distributed targets in non-Gaussian clutter with unknown scattering center distributions,this dissertation proposes a novel adaptive estimation method for target scattering centers based on sparse regularization.This approach takes advantage of the sparsity of target scattering centers in the radar detection window and incorporates sparse signal recovery theory.By leveraging this estimation,this dissertation further derive a generalized likelihood ratio detector specifically designed for range-spread targets.This method effectively mitigates the issue where traditional detectors’ performance deteriorates significantly when the target scattering center distribution deviates from prior assumptions.As a result,it significantly improves the robustness of detecting range-spread targets with diverse scattering center distributions,especially in non-Gaussian clutter environments.(2)To address the detection of range-Doppler distributed targets in non-Gaussian clutter with an unknown target steering matrix,this dissertation proposes an adaptive estimation method for the target steering matrix based on sparse Bayesian theory.This method takes advantage of the sparsity of the target’s Doppler spread frequency components and incorporates sparse signal recovery theory.Based on this estimation,this dissertation further derive a generalized likelihood ratio detector specifically tailored for range-Doppler distributed targets.This approach effectively solves the problem of traditional detectors’ performance degrading significantly when the dimensionality of the target signal subspace is unknown.Consequently,it significantly boosts the robustness of detecting range-Doppler spread targets,particularly in non-Gaussian clutter environments.(3)To address the detection of range distributed targets in non-Gaussian clutter with missing secondary data set,this dissertation introduces a generalized likelihood ratio detection method,which operates independently of secondary data set.In contrast to the conventional “estimation-plug” two-step detection approach that relies on secondary data set,our method solely depends on the data set under tested.Through an iterative estimation process,the proposed method obtain the approximate maximum value of the generalized likelihood ratio detector’s test statistic,enabling effective range distributed target detection.This approach diminishes the detector’s dependence on secondary data set in non-Gaussian clutter scenarios,thereby boosting its flexibility and overall applicability.(4)To address the detection of adjacent multiple distributed targets when the target number is unknown,this dissertation introduces a clustering method for scattering points based on fuzzy shell clustering.This method takes into account the shape information of the target scattering point distribution,thereby improving the clustering performance in complex multi-target environments.Moreover,to overcome the challenge of estimating the number of targets,this dissertation propose a target count estimation model utilizing the modified Akaike Information Criterion.This model significantly enhances the accuracy of estimating the number of distributed targets in multi-target detection scenarios.