Research On Parameter Estimation Method Of LFM Sginal

As a typical non-stationary signal,LFM signal has a wide range of applications in many fields such as geophysics,biomedicine,communication,radar,sonar,etc.Especially in the field of electronic countermeasures,LFM signal is a representative pulse compression signal which occupies an important position such as in Radar,Sonar and other equipment of various systems.Accurately obtaining the parameters of LFM signal is an important task of the reconnaissance receiving system.Therefore,it has important theoretical research significance and engineering application value to realize the parameter estimation of LFM signal under the condition of non-cooperative or limited prior information.As one of the important branches in the field of signal processing,FM signal parameter estimation technology has achieved many achievements in theoretical research and engineering applications over the years.However,with the increasingly complex channel and electromagnetic environment,there is still a weakness in the accurate estimation of the parameters of LFM signal under the condition of low signal-to-noise ratio.This dissertation focuses on the research on the estimation methods of the two important parameters,the initial frequency and the chirp rate of LFM signal and RBC-LFM signal.The algorithm is verified by theoretical analysis and simulation experiments.The performance and applicable scenarios of the parameter estimation algorithm are analyzed.The improved algorithm and implementation method are slso explored.The main contents include:(1)This dissertation introduces the background and research significance of the topic,summarizes the research status and development trend of chirp parameter estimation combined with specific applications.And the main work of this dissertation is introduced.(2)The basic theory of LFM signal and parameter estimation is introduced.The LFM signal is discussed in detail from the four perspectives of time domain,spectrum structure,fuzzy function and time-frequency analysis.The characteristics of LFM signal are analyzed.The principles,advantages and disadvantages of different time-frequency analysis methods are deduced and discussed.The main parameters of LFM signal and the performance evaluation method of the parameter estimation algorithm are introduced.(3)Four typical chirp parameter estimation methods based on time-frequency information are studied from four aspects: instantaneous frequency,line detection,fractional domain and fuzzy function.Firstly,according to the characteristics of the joint distribution of signal time-frequency domain,the least squares fitting parameter estimation method based on instantaneous frequency and the parameter estimation method based on RWT projection integration of straight line are studied respectively.Then,the parameter estimation method based on Fr FT is studied by using the energy accumulation characteristic of Fr FT on LFM signal.At last,aiming at the low accuracy of least squares fitting parameter estimation and the large amount of computations required for two-dimensional search for RWT and Fr FT,a parameter estimation algorithm based on RAT combined with MLE is introduced,so that the parameter estimation can be realized by two one-dimensional searches.(4)The dissertation explores a new method combining deep learning with chirp parameter estimation algorithms.Aiming at the problem that the maximum likelihood estimation algorithm has a large amount of computation and which will consum a long time.So a maximum likelihood parameter estimation method based on local search is proposed.The method can take full advantage of deep learning in feature extraction and representation.First,De CNN is used to filter and extract the input signal.Then,the least square fitting parameter estimation method is used to roughly estimate the parameters,and the search range of the input signal’s compression likelihood function is determined.Finally,MLE is used in the local area.This method combines deep learning technology with signal processing algorithm and broadens the way of digital signal processing.(5)The main characteristics and parameter estimation method of PRBC-LFM signal is studied.The algotirhm of the squared delay correlation demodulation parameter estimation method is derived.Aiming at the problem that algotirhm is sensitive to noise,a RAT-based square demodulation parameter estimation algorithm is inrtoduced.The method uses the chirp rate estimation result of the RAT to directly de-chirp the composite signal,which not only improves the accuracy of parameter estimation,but also improves the anti-noise performance of the algorithm.In order to further improve the accuracy of parameter estimation under the condition of low signal-to-noise ratio and remove the influence of square op-amp noise,a parameter estimation method based on Se CNN signal extraction is proposed.The correctness of the algorithm and the effectiveness under the condition of low signal-to-noise ratio are proved by simulation experiments.