Radar Signal Intra-pulse Modulation Type Recognition And Character Analysis

Radar signal intra-pulse modulation recognition and analysis have encountered challenges, as new system radars have rapidly developed and electromagnetism environment in space was more and more complex. First, two algorithms of modulation recognition is researched based on wavelet packet and BP neural network. Then, intra-pulse character of radar signal is researched from radar signal modulation parameter extraction and radar signal inter-pulse analysis. My main work is summarized as follow:1. Radar signal intra-pulse modulation recognition algorthms based on wavelet packet is studied. Wavelet packet translation is studied deeply, using optimization algorithm of wavelet packet decomposed, reconstruct coefficients which contained large information was extracted. To solve the mesh of the wavelet packet reconstruct coefficients too high and how to construct the recognition feature, two method are proposed: First one, the energy of wavelet packet reconstruct coefficient is calculated, statistical features of energy are recognition feature; Another one, feature selecting based on SVD from Reconstruct coefficients of wavelet packet is proposed, six big singular values was recognition feature. How to setup parameter of BP model is introduced, and establish the BP classifier. The performance of two kinds of features is analyzed by simulating.2. Radar signal intra-pulse modulation recognition algorthms based on instantaneous frequency is studied. To solve the lower recognition rate issue, a radar signal intra-pulse modulation recognition algorthms based on instantaneous frequency image is proposed. The instantaneous frequency was studied deeply, a method of transforming the instantaneous frequency into binary image is proposed. The moment feature of the binary image is extracted, and normalizing the moment feature eliminates the affect of sampling frequency and the length of signal. Two classifiers of and a BP classifier is established, finally the performance of this method is analyzed by simulating.3. Bayesian parameter estimation algorithm of LFM signal and code width of PSK signal based on Correlation receive algorithm is studied. The model of Bayesian estimation is studied for estimating the modulation parameter of LFM signal and the frequency of sine signal. The MCMC algorithm is introduced to calculate the Bayesian estimation for eliminating the computation burden. To improve the convergence speed of MCMC, a hybrid sampling between Rand Walks and Independent Markov Chain is proposed. Correlation receive algorithm is studied for estimating the width of code. The precision of code width is improved by smoothing, fine-tuning the width of code and averaging the real part'result and imaginary part'result. 4. The algorithm of envelop feature analysis of radar signal is studied. The performance of Hilbert transform and complex wavelet transform for extracting envelop is studied deeply. In order to eliminating the affect of noise, complex wavelet transform is used to extract envelop. To eliminate the affect of noisy, envelop is smoothed. Finally the envelop feature is extracted from the last layers of complex wavelet transform, and the PCA method is used to eliminate the redundancy of the envelop feature and enhance the stability of the envelop feature.