| 摘要: |
| 为满足战场监控系统对海量历史数据中脉冲信号的快速处理需求，提出了一种面向海量数据的脉冲信号两级快速检测与参数估计方法。首先，在脉冲存在性检测阶段采用Z分数与降采样协同判别方法实现脉冲信号的初步检测。该检测阶段通过降采样降低数据规模，利用Z分数的统计特性差异快速区分脉冲信号与背景噪声，有效减少计算冗余，提升检测效率。其次，在脉冲时域定位检测阶段提出了一种基于反馈机制的滑窗步长动态调整方法。该方法基于实时反馈因子计算，动态优化能量窗的滑动步长，在保证定位准确的同时提高计算效率。最后，根据检测结果估计到达时间、脉冲宽度和中心频率参数。实验结果表明，所提方法在参数估计较准确的同时，与3种对比方法相比，检测时间分别减少约93.90%、80.29%和35.14%，且随着数据量的增大优势将更明显。 |
| 关键词: 脉冲信号检测 海量数据 两级结构 动态能量窗 参数估计 |
| DOI：10.20079/j.issn.1001-893x.250429001 |
|
| 基金项目:国家自然科学基金资助项目（42274159） |
|
| Two-phase Fast Detection and Parameter Estimation of Pulse Signals in Massive Data |
| SUN Lianghai,SUN Weifeng,WAN Yong,ZHANG Peng,ZHANG Chao,DAI Yongshou |
| (1.College of Oceanography and Space Informatics，China University of PetroleumEast China，Qingdao 266580，China;2.Ceyear Technologies Co.，Ltd.，Qingdao 266555，China) |
| Abstract: |
| To meet battlefield surveillance system requirements for fast pulse signal processing in massive historical data，the authors present a two-phase fast detection and parameter estimation method for pulse signals in massive data.Firstly，in the pulse existence detection stage，the Z-score and downsampling collaborative discrimination method is used to achieve preliminary detection of pulse signals.In this detection stage，downsampling is used to reduce the data size，and the statistical characteristics of Z-scores are utilized to quickly distinguish pulse signals from background noise，effectively reducing computational redundancy and improving detection efficiency.Secondly，a sliding window step size dynamic adjustment method based on feedback mechanism is proposed in the pulse time-domain positioning detection stage.This method is based on real-time feedback factor calculation，dynamically optimizing the sliding step size of the energy window to improve computational efficiency while ensuring accurate positioning.Finally，the arrival time，pulse width，and center frequency parameters are estimated based on the detection results.Experimental results demonstrate that the proposed method reduces detection time by approximately 9390%，80.29% and 35.14%，respectively，compared with three methods proposed in references,while achieving accurate parameter estimation.Moreover，the advantage becomes more pronounced as the data volume increases. |
| Key words: pulse detection massive data two-phase structure dynamic energy window parameter estimation |
《电讯技术》编辑部 