Research On Weak Signal Detection Of Small Target In Sea Clutter And Its Related Methods

Sea clutter is the backscatter returning from a patch of sea surface of radar signal which can be affected by like wind,wave and other complex factors.Studies show that sea clutter has chaotic characteristics and fractal characteristics,which is a typical non-stationary signal.In addition,it has been found that the inner character of sea clutter can be used to analyze the ocean state and detect small target on sea surface or at low attitude.Actually,the small target signal is usually submerged in sea clutter and noise background,because the radar cross section of the small target is extremely small and the radar echo is susceptible to noise such as radar measurement noise and sea surface dynamic noise.Therefore,how to identify small targets accurately and stably under differnet sea conditions has become a research hotspot and difficulty in the field of signal processing.In this paper,we studied the determination of the phase space reconstruction parameters of chaotic systems,and proposed an improved phase space reconstruction parameters determination method based on LS-SVM and AD method.Simultaneously,the aspect of noise suppression in sea clutter is studied,and a new adaptive algorithm for sea clutter denoising is proposed.What's more,the weak target detection in sea clutter background is analysed,and a new small target detection method based on the fractal difference of in multi-scale and optimal transformation order,a new detection method of optimized kernel extreme learning machine?KELM?are respectively proposed.The concrete research content is as follows:According to the theory of embedding window,an improved method to determine chaotic system phase space reconstruction parameters based on average displacement?AD?method and least squares support vector machine?LSSVM?was proposed.It uses AD method to determine the width of the embedding window after determining the embedding dimension.We can calculate delay time according to the width of the embedding window and embedding dimension.The Lorenz chaotic system was selected as an example,and the influence of noise on determining chaotic system phase space reconstruction parameters was researched.The result shows that the proposed method is applicable for determining chaotic system phase space reconstruction parameters when SNR is bigger than 0dB,it can decrease the mean square by one orders of magnitude compared with the methods such as BP neural network and traditional SVM.To suppress the influence of noise on the detection of weak signals in chaotic sea clutter,we proposed a new adaptive algorithm for sea clutter denoising based on complete ensemble empirical mode decomposition with adaptive noise?CEEMDAN?theory and independent component analysis?ICA?.Firstly,the chaotic signals is decomposed into a series of intrinsic mode functions?IMFs?by CEEMDAN,and the boundary of noisy IMFs is identified according to the cross correlation coefficients of the original signal and each IMF,and then the noisy IMFs are adaptively filtered by ICA based on the distribution of the first local minimum of the cross correlation coefficients.Finally,the IMFs after filtering and the remaining IMFs are reconstructed into a new signal.R?ssler,Lorenz systems and the data from the IPIX radar sea clutter database are used in the simulation,the result shows that the proposed denoising algorithm can effectively filter out the noise submerged in chaotic sea clutter and increase the accuracy of weak signal detection.The root mean square error after denoising can be reduced by about one orders of magnitude,reaching 6.3558*10^-4,while the model before denoising can only reach 0.0058.In addition,the root mean square error is also increased by 47.04%-69.73%compared with other denoising algorithms such as EMD-ICA.To overcome the dependence of small target detection on the sea situation,this paper studies the property of sea clutter in fractal fourier triansform?FRFT?domain and proposes a new small target detection method based on the fractal difference of in multi-scale and optimal transformation order.We use fractional Brownian motion to model the measured sea clutter data from IPIX radar and combine the multifractal detrended fluctuation analysis based on dual tree complex wavelet transform to analyze the fractal property of the sea clutter in different situation,distances,and polarizations.Finally,according to the distribution of fractal difference in the optimal scale and transform order,we realize the detection of small targets under sea clutter background.The experimental results show that the proposed method can effectively magnify the fractal characteristics difference between the target unit and the pure sea clutter unit,and suppress the dependence of small target detection on the sea situation.Compared with the traditional single-scale variable order method in FRFT domain,the detection threshold is mostly increased by more than 90%.In order to further improve the detection accuracy of weak signals in the background of chaotic sea clutter,we proposed a new detection method of optimized kernel extreme learning machine based on complete ensemble empirical mode decomposition?CEEMD?theory.By CEEMD,chaotic signals containing the weak target can be decomposed into a series of intrinsic mode functions?IMFs?,and then the prediction models of each IMF reconstructed by phase space are established by KELM.Using the artificial bee colony algorithm?ABC?to continuously optimize the regularization coefficient and the kernel function parameter of KELM,after reconfiguration of predictive signal,we can detect the weak signals submerged in sea clutter background from the prediction error.Lorenz attractor and the data from the IPIX radar sea clutter database are used in the simulation,and the influence of noise at different intensities on weak signal detection is investigated in depth.The results show that the proposed method can effectively detect the weak target from chaotic signal background.When there is no noise in the system,by using the proposed method,the root mean square error can be reduced by four orders of magnitude,reaching 0.00000012?SCR=-118.9591dB?,while the conventional ELM can only reach 0.00134508 under the condition of SCR=-80.1547dB.In addition,the noise influence on the target detection performance can be ignored if SNR?0dB.This paper studies the characteristics of small targets in the chaotic sea clutter.The weak signal detection model is established based on the theories of phase space reconstruction,empirical mode decomposition,independent component analysis,kernel extreme learning machine and fractal,which can better suppress the interference of sea conditions to small target detection,and has a certain theoretical and practical application value for the recognition of small target and the safety monitoring on sea surface.