| 摘要: |
| 为了解决现有检测算法在低信噪比下因脉冲分裂而无法准确提取到达时间与脉冲宽度的问题，本文提出基于能量预检测与时频降噪处理的脉冲检测方法。首先利用待检测信号的频谱估计出噪底功率，用于后续计算检测阈值；然后用能量检测粗定位可能含有目标脉冲的信号段；再用分裂脉冲识别与合并模块处理各信号段，将时间邻近的低信噪比信号段合并为一段信号；再用基于同步提取变换的时频降噪方法处理低信噪比信号段，以达到时频域降噪和去除虚警脉冲的目的；最后用自相关检测提取各信号段的到达时间与脉冲宽度。仿真实验结果表明，本文脉冲检测方法能够适用于线性调频、正弦调频、巴克码调相、单频与V型调频信号多种调制类型信号，当信噪比在-1.5dB及以上时，该方法有95%以上的概率检测到脉冲并准确提取到到达时间与脉冲宽度。 |
| 关键词: 脉冲检测 噪底功率估计 分裂脉冲识别与合并 时频降噪 自相关检测 |
| DOI： |
| 分类号:TN971 |
| 基金项目:国家自然科学基金项目（面上项目，重点项目，重大项目） |
|
| Signal Detection Dethod for Pulse Pre-detection and Time-frequency Denoising |
|
|
|
|
| Abstract: |
| In order to solve the problem that existing detection algorithms cannot accurately extract the arrival time and pulse width due to pulse splitting under low signal-to-noise ratio, this paper proposes a pulse detection method based on energy pre-detection and time-frequency domain denoising processing. First, the noise floor power of the signal to be detected is estimated from its spectrum, which is used for subsequent calculation of detection threshold. Then, the energy detection module is used to roughly locate the signal segments that may contain the target pulse. The split pulse recognition and merging module processes each signal segment, merging the low signal-to-noise ratio signal segments that are close in time into one signal segment. Then, the time-frequency denoising method based on synchronous extraction transform is used to process the low signal-to-noise ratio signal segments to achieve noise reduction and removal of false alarm pulses in the time-frequency domain. Finally, the autocorrelation detection method is used to extract the arrival time and pulse width of each signal segment. The simulation results show that the pulse detection method proposed in this paper can be applied to various modulation types, including linear frequency modulation, sinusoidal frequency modulation, Barker code phase modulation, single frequency and V-shaped frequency modulation signals. When the signal-to-noise ratio is greater than -1.5dB, the method has a probability of over 95% of detecting the pulse and accurately extracting the arrival time and pulse width. |
| Key words: pulse detection noise floor power estimation split pulse recognition and merging time-frequency denoising autocorrelation detection |