Research On Characteristic Extraction Of LFM Radar Signals

In the present complex electromagnetic environment electronic surveillance equipment intercepts and analyzes the radio signals from target emitter, and then extracts the effective characteristic parameters of emitter, which is used to identify, locate, track, interfere and attack the target. It has important meaning to occupy the information battlefield. This method could effectively overcomes the limitations at the process of active target identification, which is easily exposed and interfered, and also needs strong transmit power. This thesis mainly researches on signal analysis and characteristic extraction of LFM signals, which is basis to identify target and obtain information.Firstly, according to the linear instantaneous frequency characteristic of LFM signals, FrFT is used to extract characteristic of single period LFM signals. After analyzing the basic concept and fast algorithm of FrFT, the process to analyze LFM signals with FrFT is designed, and then the chirp constant and initial frequency characteristic is estimated. The repeated seeking method is produced to overcome the effect caused by strong noise and interference. The simulation experiment shows that FrFT is effective to analyze the LFM signals.Then, the method based on FrFT to extract characteristic of periodic LFM signals is researched. Because LFMCW signals and LFMPW signals are repetitive according to period, the idea of short time window is introduced to analyze signals hop by hop. On the base of the periodicity of the maximum amplitude in the transform and initial frequency inside the pulse, pulse period and frequency period is effectively estimated. K means method is introduced to adjust the results by the repeated seeking method at the condition of strong interference. But the pulse width estimation accuracy depends on the adjustment value with different SNR.At last, Duffing oscillator is built to estimate the pulse width. Build the Duffing oscillator model, and analyze the signal hop by hop. Use the phase criterion to estimate pulse information, and then determine the pulse width. Because Duffing oscillator is not sensitive to noise, it could be used at the condition of wide range of SNR. When the SNR is beyond -5dB and SIR is beyond 0dB, recognition rate reaches to 90% with 1% estimation error. From theoretical analysis and simulation experiment, FrFT and Duffing oscillator method could extract the characteristic parameters of LFM signals effectively, and have good anti-noise and anti-interference performance, which provides parameter basis for target identifying and locating.