随着雷达信号脉内调制方法日趋复杂,单纯地利用时域或频域的算法来进行信号调制类型的识别已很难奏效。借助于雷达信号的时频分布图像,提出了一种在时频联合域上进行信号识别的新算法。首先揭示了时频分布图像中确实蕴含着信号调制类型的本质特征,然后详细阐述了利用二维主分量分析来提取时频分布图像特征参数的算法,最后对算法进行了仿真,并从识别率、算法复杂度以及硬件需求和训练时间3个角度进行了比较。结果表明,该算法提取的特征参数具有很好的鲁棒性,可以取得较高的识别率,同时可以降低硬件需求,缩短训练时间。
With the increasing complexity of intra-pulse modulation method for radar signal, it is difficult to recognize the signal modulation type based separately on time domain or frequency domain. The authors propose a new method with combination of time domain and frequency domain by using the time-frequency distribution image of the radar signal. At first, it is indicated that the time-frequency distribution image does contain the essence of modulated signals Then it is discussed in detail how to extract the characteristic parameter of time-frequency distribution image by using two-dimensional principal component analysis. The simulation results are presented. The method is compared with other methods through recognition rates, algorithm complexity, hardware requirement and training costing. The results show that the feature parameters extracted by this method have good robustness, and the recognition rate is higher. Meanwhile, the proposed method consumes less time and lower requirements for hardware.