Multi-Component Signal Separation Technology Based On Optimized Time-Frequency Distribution

With the development of electronic technology,electronic warfare has become an important part of military warfare,and even affects strategy making.Nowadays,the battlefield electromagnetic environment is more and more complex,with complex signal components and high density.Usually,the received signal contains multiple signals,and these signals are interwoven in the time domain and frequency domain.So that simple operation can not separate them.The traditional forward jamming method is no longer applicable to achieve accurate electronic jamming.It is necessary to separate multi-component signals.In this paper,instantaneous frequency extraction and time-domain recovery of overlapping multi-component signals are studied,based on the time-frequency distribution of multi-component signals.Firstly,we analyze the high resolution time-frequency distribution algorithm suitable for analyzing multi-component signals,and then discusse how to suppress the cross terms to the maximum extent and keep the signal resolution without distortion in the quadratic time-frequency distribution.Experiments verify that the adaptive angle time-frequency distribution has the best time-frequency resolution for multi-component signals.And then the optimization scheme of the adaptive Angle time-frequency distribution is discussed.We propose a parameter optimization ADTFD based on genetic algorithm,which matches the best parameters for the ADTFD.Secondly,the instantaneous frequency is estimated based on the high resolution time-frequency distribution,and the problems of existing algorithms are summarized.The improved Viterbi algorithm is an existing frequency estimation method which can avoid switching errors in signal crossover region to the greatest extent.At the same time,in order to adapt to the instantaneous frequency of the nonlinear FM signal,the results of hough transform are obtained by dividing the time-frequency distribution and computing separately,which can not only resist the interference of noise but also correct the frequency estimation and minimize the error.Finally,on the basis of the instantaneous frequency information of the signal,the time domain information of the signal can be reconstructed by inverting the time-frequency distribution of the filtered single component.In this process,the generalized S-transform is proposed to carry out time-varying filtering,in which the gaussian window function is related to the value of instantaneous frequency,so that the obtained instantaneous frequency information can be used to guide filtering operations to the maximum extent.In this paper,the separation of multi-component signals is the ultimate goal,based on the optimized time-frequency distribution,the location of signal components in the time-frequency distribution is located by extracting instantaneous frequency,and the time-varying filtering method is used to recover the signal waveform in time domain.