Research On Sea Surface Target Detection Methods Based On Mode Decomposition

The research of sea clutter suppression and target detection methods in sea clutter background has been a difficult problem in the field of radar signal processing.Due to the interference of sea clutter in radar echoes,the target echo is submerged in sea clutter,resulting in the degradation of sea surface target detection accuracy.Therefore,to further analyze the internal characteristics of sea clutter,study the differences between sea clutter and target echoes,improve the detection accuracy of sea surface targets,and design a sea surface target detection method with good detection performance is a topic of great significance.The research content starts from the analysis of sea clutter characteristics and proposes a prediction model of sea clutter and several methods of target detection in the background of sea clutter under three different sea conditions based on the measured IPIX radar data set.The specific research contents as well as innovations are as follows:1.The measured IPIX radar dataset is introduced,and the radar echo characteristics are analyzed based on this dataset,the statistical characteristics of sea clutter amplitude are analyzed,and the commonly used sea clutter amplitude model is fitted.The analysis compares the differences between the sea clutter unit and the target unit fractal characteristics,including the features of Detrended Fluctuation Analysis(DFA),Hurst Analysis,Mass Analysis,and Multifractal Spectrum Analysis.2.Based on the difference in fractal characteristics between the sea clutter unit and target unit,Singularity Power Spectrum(SPS)method is introduced to establish an analysis model for radar echoes.The SPS-based sea surface target detection algorithm is designed,and three feature extraction methods are proposed.The effectiveness and performance of the proposed algorithm are verified by the measured radar echo data under low,medium,and high sea state scenarios.The experiments prove that the proposed algorithm can achieve good detection probability regardless of the sea state conditions.The experiments proved that the average detection probability reached 82.3% in the low sea state conditions,83.3%in the medium sea state conditions,and 66.7% in the high sea state conditions.3.An extreme learning machine(ELM)optimized by the Salp Swarm Algorithm(SSA)is designed to predict the sea clutter.The convergence speed and accuracy of the SSA algorithm are verified using several test functions,and the experiments prove that SSA has excellent convergence speed and accuracy.The SSA-ELM sea clutter prediction model is constructed and trained by using SSA to optimize the ELM transfer matrix and bias values,and the superiority of the proposed model is demonstrated by comparing various optimization algorithm models.The accuracy of the prediction model is verified by using the measured data under low,medium,and high sea state scenarios,and the advantages of the proposed algorithm are demonstrated by combining various prediction models.It is proved that the average prediction accuracy of SSA-ELM sea clutter prediction model can reach more than 94% under all three sea state conditions scenarios.4.Fuzzy Entropy(FE)is introduced as a tool to quantify the complexity of sea clutter and target echo sequences,and the feasibility of quantifying the complexity of sea surface echoes is verified,and the FE-based sea surface target detection tool is proposed.The detection is performed by distinguishing sea clutter and target echo FE features in low,medium,and high sea states,and the experiment proves that FE features can effectively distinguish the difference between the two.5.To solve the problem of low differentiation of FE features in high sea state scenarios,the Variational Mode Decomposition(VMD)tool is introduced and optimized the preset number of decomposition layers of the VMD tool,and the Adaptive Variational Mode Decomposition(A-VMD)is proposed.By performing adaptive decomposition of radar echoes under different sea state scenarios,suitable intrinsic mode functions are selected for reconstruction,and FE features are extracted from the reconstructed sequences.The experiments demonstrate that,compared with the surface target detection using only fuzzy entropy features,the surface target detection method based on the combination of A-VMD and FE features has a more obvious degree of distinction between sea clutter and targets,the entropy difference performance is greater,and the detection probability has different degrees of improvement regardless of low,medium and high sea conditions,which has a better detection performance for surface targets.