Font Size: a A A

Research On Chaotic Time Series Parameter Analysis And The Application On Small Target Detection

Posted on:2014-09-09Degree:MasterType:Thesis
Country:ChinaCandidate:P GongFull Text:PDF
GTID:2268330401970333Subject:Signal and Information Processing
Abstract/Summary:PDF Full Text Request
Chaos signal that is produced by nolinear system is seemingly random and extremely sensitive to initial conditions. The study of chaos is one of important research direction in the field of signal processing. In order to study the characteristics of chaos system, the analysis and research about parameters of chaotic time series has vital significance.This paper mainly studies the determination methods about phase space reconstruction parameters (embedded dimension m and time delay r) of chaotic time series and a new method based on AD method and the BP neural network was proposed. Based on the theory of dissipation synchronization, a new method of reconstruction chaos system for chaotic time series with noise is proposed, and using it to realise the weak signal detection from chaotic background. The measured IPIX sea clutter is used as research object and the chaos and fractal characteristic of sea clutter is researched. Based on the characteristic of sea clutter, new methods of small target detection from sea clutter is implemeted. The detail of research is as follows:Based on Tokens theorem, only a suitible embedded dimension and time delay is chosen for finite length chaotic time series can realise a well chaotic phase space reconstruction. Various methods have been introduced in this paper. The embedded dimension and time delay are determined respectively for the time series with different integral time step length. The result shows that embedded dimension can be better to illustrate the chaotic characteristic of time series. Emphatically analyzed the disadvantages of C-C method and improved algorithm which is used the average orbital period. Based on automated embedding phase space reconstruction algorithm, a new embedded window method is proposed by combine neural networks with average displacement method (AD method), and the simulation experiments show that this method is useful to determining the parameters of chaotic phase space reconstruction.Noise and chaotic time series are very similar. After the preprocessing, the residual noise is still inevitable in chaotic signal. Dissipation synchronization model is adopted to accomplish the reconstruction of chaotic systems by chaotic time series with noise in this paper. Chaos synchronization weakens the characteristic of sensitive to initial conditions on chaos system. Weak signal detection from chaotic background is realized by this model, and the noise influence on the performance of detection method is analysed. The research result shows that when Signal-to-Clutter Ratio is fixed, the noise influence can be ignored if the signal to noise ratio is high enough (SNR>0db). But, if SNR<-lOdb, the algorithm can not to detect weak signals from chaotic.The measured IPIX sea clutter is used as research object and the chaos and fractal characteristic of sea clutter is researched. The results show that the fractal characteristic of sea clutter changes with sea conditions and radar polarizations. Based on the relationship between chaos and fractal, it means that the chaos characteristic of sea clutter also changes with sea condition and radar polarizations and the unified chaotic forecasting model cannot analyze sea clutter in different sea condition and radar polarizations. According to the fractal characteristics of sea clutter, a method of small target detection from sea clutter based on fractal differential, and another based on the difference of H(q) on high scales are proposed. These two methods can realize the detection of small target from the background of sea clutter.
Keywords/Search Tags:chaotic time series, phase space reconstruction, fractal, sea clutter, small targetdetection
PDF Full Text Request
Related items