Aiming at three specific modulation modes of FM,MSK and QPSK,the effect of differentiating MSK and QPSK modulation signals with wavelet transform amplitude variance is poor in white Gaussian noise(WGN)environment,and the maxi⁃mum instantaneous amplitude of zero center normalized spectral density has bad effect to recognize the FM,MSK and QPSK mo⁃dulation signals under the conditions of different signal⁃to⁃noise ratios,so the characteristic quantity method based on kurtosis and wavelet transform used for signal classification and recognition is put forward,in which the kurtosis is used for type identifi⁃cation of analog modulation signal and digital modulation signal,and then the characteristic quantity extracted on the basis of wavelet transform coefficient is used to recognize and classify the two digital modulation signals. The decision tree classification method is adopted to carry out simulation verification of the method by means of Matlab. The simulation results show this method has better recognition effect.%针对FM,MSK,QPSK三种具体调制方式,由于在高斯白噪声环境中小波变换幅度方差区分MSK,QPSK调制信号效果较差,以及在不同信噪比条件下零中心归一化瞬时幅度谱密度的最大值对FM,MSK和QPSK识别效果的下降,提出利用峰度和基于小波变换特征量的方法对信号分类识别。该方法首先利用峰度进行模拟与数字调制信号的类间识别,然后采用基于小波变换系数提取的特征量对两种数字调制信号进行识别分类。采用决策树分类方法,利用Matlab对该方法进行仿真验证,仿真结果表明该方法具有较好的识别效果。
展开▼