Research On Radiation Source Identification Technology Based On Artificial Intelligence

Radiation source identification refers to analyzing and judging the type of radiation source and individual through the features of the intercepted radiation source signal.Radar is the main radiation source in the battlefield environment.With the rapid development of radar technology and the improvement of anti-jamming capability,radar radiation source signals are also increasingly complex.From continuous wave and simple pulse to linear frequency modulation and pulse coding,traditional radars are used.Signal source frequency,pulse width,repetitive frequency and other parameters for radiation source identification can no longer meet the requirements for classification and identification of radar radiation sources in complex battlefield environments.In this paper,the artificial intelligence method is applied to the radar radiation source identification.The radar signal received by the detection is transformed from the time domain to the time-frequency domain by time-frequency transform.The artificial intelligence technology such as machine learning and deep learning is used to realize the radar radiation source signal.Accurate classification and identification.In the time-frequency domain,the difference in signals from different radar emitters is more pronounced.Therefore,the paper transforms the radar signal into a two-dimensional time-frequency image,and then applies the artificial intelligence related technology,such as machine learning,deep learning,etc.to automatically extract the features of the image,and finally use the automatic encoder(AE)in the artificial intelligence.Automatic Encoder Overlay(SAE)and its variants Noise Reduction Automatic Encoder(DAE)to identify radar emitter signals.The main work of the thesis is as follows: 1.The radar signal model was established,and the main radar signals such as regular pulse,linear frequency modulation,phase encoding,and frequency encoding were analyzed and simulated.2.Radar radiation source identification technology based on time-frequency analysis method is studied.STFT,wavelet transform and WVD time-frequency conversion simulation are performed on common six modulation modes of the radiation source signal through analysis and comparison,three different signal-to-noise ratios are used.The characteristics of the time-frequency diagram obtained by the time-frequency transform method show that the STFT has better performance in terms of noise immunity and aggregation degree at low signal-to-noise ratio.3.For the traditional recognition method of low signal to noise ratio recognition effect is not good,this paper studies the method based on automatic encoder recognition.Firstly,the time-frequency analysis method is used to simulate and preprocess the radiation source signal to obtain a grayscale image;then,an automatic encoder is used to perform the implicit feature extraction,and an automatic identification method and an identification effect are added.4.Lastly,SAE and DAE are used in the simulation experiment under the condition of greater than 0d B,and all of them can achieve the recognition rate of over 99%.Then the simulation experiment is performed under the condition of less than 0d B.The highest SAE recognition rate is only 66.7%,and the DAE is up to 95.0%.Finally,the DAE and CNN were compared and tested.It was found that under low SNR,the DAE has better recognition effect,but the performance difference is not high at high SNR,thus verifying the effectiveness of the method.