复杂环境下主被动复合探测目标信号仿真与识别技术研究

【作者】 徐达；

【导师】 肖泽龙； 胡泰洋；

【作者基本信息】 南京理工大学 ， 信号与信息处理， 2021， 硕士

【摘要】 针对日益复杂的战场环境,主被动复合探测可以在复杂环境下有效的识别装甲目标,是目前研究的热点。本文针对复杂环境下主被动复合探测的目标仿真与识别技术开展了以下研究:（1）针对大气、降雨、云雾和沙尘等复杂环境对回波信号的影响,本文建立了复杂环境下的毫米波衰减模型,分别仿真了大气、降雨、云雾和沙尘随着路径、天线仰角、温度、能见度等条件变化时的特性衰减曲线。结合复杂环境下背景亮温的变化,仿真了复杂环境下输出波形的变化情况,并给出了不同情况下的衰减值。（2）针对主被动复合探测中可能存在的低信噪比问题,本文主动通道采用小波变换去噪算法和基于ap FFT时移相位差法的频谱细化方法提高了主动通道测频精度,当信噪比为-15d B时,高斯白噪声情况下测频精度可达到10^-3倍的谱线间隔。被动通道采用基于滑动平均和卡尔曼滤波的滤波降噪算法,当信噪比低至-30d B时,也能很好的提取出钟形峰特征,降低了被动通道特征值提取难度。（3）针对传统恒虚警（CFAR）算法易于掩盖主目标附近小目标的缺点,设计了改进的一维OS-CFAR和其基于十字窗的二维形式,使阈值门限能在保证虚警率的情况下更大概率检测出相邻目标。针对传统单一决策层信息融合存在的融合不够彻底问题,设计了一种两层信息融合方法,在决策层融合的同时增加了特征层融合,将主动通道的距离信息结合复杂环境信息共同输入到被动通道,增强了复合探测识别效果。（4）为了验证本文主被动复合探测信号处理算法的有效性,完成了主被动复合探测信号处理模块的硬件设计,并用verilog编写了信号处理代码,利用高塔试验测试了主被动通道的目标信号和识别信号,通过与单一体制对比,识别概率提高13%～15%,验证了本文中主被动复合信号处理算法的可行性和有效性。更多还原

【Abstract】 In view of the increasingly complex battlefield environment,active and passive composite detection can effectively identify armored targets in complex environments,which is a hot research topic at present.In this thesis,the following researches are carried out on target simulation and recognition technology for active and passive composite detection in complex environments:（1）Aiming at the influence of complex environments such as the atmosphere,rainfall,clouds and dust on the echo signal,this thesis establishes a millimeter wave attenuation model in a complex environment,and simulates the characteristic attenuation curve of the atmosphere,rainfall,clouds,and dust with the the path,antenna elevation angle,temperature,visibility and other conditions change.Combined with the change of the background brightness temperature in the complex environment,the change of the output waveform in the complex environment is simulated,and the attenuation value under different conditions is given.（2）Aiming at the low signal-to-noise ratio problem that may exist in the active and passive composite detection,the active channel in this thesis adopts the wavelet transform denoising algorithm and the spectrum refinement method based on the ap FFT time-shift phase difference method to improve the frequency measurement accuracy of the active channel.When the noise ratio is-15d B,the frequency measurement accuracy can reach10^-3 times the spectral line interval in the case of Gaussian white noise.The passive channel adopts a filtering and noise reduction algorithm based on moving average and Kalman filter.When the signal-to-noise ratio is as low as-30d B,the bell-shaped peak feature can be extracted well,which reduces the difficulty of extracting the feature value of the passive channel.（3）Aiming at the shortcomings of traditional constant false alarm rate（CFAR）algorithms that are easy to conceal small targets near the main target,an improved one-dimensional OS-CFAR and its two-dimensional form based on a cross window are designed,so that the threshold threshold can guarantee the false alarm rate.In this case,there is a greater probability of detecting adjacent targets.Aiming at the problem of insufficient fusion of traditional single decision-making layer information fusion,a two-layer information fusion method is designed,which adds feature layer fusion at the same time as the decision-making layer fusion,and combines the distance information of the active channel with the complex environment information to be input to the passive Channel,enhance the composite detection and recognition effect.（4）In order to verify the effectiveness of the active and passive composite detection signal processing algorithm in this article,the hardware design of the active and passive composite detection signal processing module was completed,and the signal processing code was written in verilog,and the target signal of the active and passive channel was tested by the tower test.And the recognition signal,by comparing with a single system,the recognition probability is increased by 13%～15%,which verifies the feasibility and effectiveness of the active and passive composite signal processing algorithm in this article.更多还原