基于峭度加权VMD和熵特征的雷达脉内调制识别

针对复杂电磁环境下识别雷达信号脉内调制样式困难以及受噪声影响识别准确率受限的问题，提出了一种将变分模态分解(Variational Mode Decomposition, VMD)与熵特征提取相结合的识别方法。首先，通过基于峭度加权的改进VMD算法对雷达信号进行分解，得到由三个本征模态函数组成的最优分量集合；其次，对各分量分别计算其模糊熵、排列熵和符号熵值，从而实现对熵特征信息提取；最后，将特征向量输入到支持向量机完成识别。相较于其他方法，该方法有着较高的识别准确率和抗噪性能，在2 dB信噪比以上平均识别准确率为94.63%。

Radar intra-pulse modulation recognition based on kurtosis weighted VMD and entropy features

In view of the difficulties in identifying the intra-pulse modulation pattern of radar signal in complex electromagnetic environment and the limited recognition accuracy affected by noise,an identification method combining variational mode decomposition(VMD) and entropy feature extraction is proposed.Firstly,the radar signal is decomposed by the improved VMD algorithm based on kurtosis weighting,and the optimal component set composed of three intrinsic mode function is obtained.Secondly,the fuzzy entropy,permutation entropy and symbol entropy of each component are calculated respectively,so as to extract the entropy feature information.Finally,the feature vector is input to support vector machine(SVM) to complete the recognition.Compared with other methods,this method has higher recognition accuracy and anti-noise performance,and the average recognition accuracy above 2 dB signal-to-noise ratio is 94.63%.