高动态低信噪比强相干干扰下引信近场目标瞬态特征SVM检测方法

It is very significant to study the real-time precise recognition technique of target and strong coherent jamming signal for improving the anti-jamming capacity of radio fuze in low SNR and high dynamic range under complex electromagnetic environment. In this proposal, with the aid of the non-stationary feature extraction technique of the body-target characteristics and the method of Support Vector Machine, we intend to identify the target and jamming signal for the radio fuze. Firstly, we will establish the universal information model of a radio fuze system based on the information theory, from which we intend to obtain the feature spaces of the target and jamming signal. Then, we will study the inherent features of the near-field multi-point scattering target echo signals for the radio fuze and their time-frequency characteristics based on time-frequency analysis methods, in order to get the available primary and further characteristic parameters. Thirdly, we combine the approximate entropy feature to get the optimal feature parameters, based on the principle of information entropy under uncertainty setting. Finally, the method of Support Vector Machine is used to distinguish availably between the target and jamming signal/background clutter, in which the above available characteristic varies are put in. The research findings can provide both basic theory and technical supports for improving the survivability of the radio fuze in the future battlefield environment.

复杂电磁环境下，低信噪比高动态交会过程中如何实现目标与强相干干扰信号快速识别是引信提高其战场生存力的核心问题。项目针对表征引信近程目标非平稳特性的瞬态特征提取和高实时检测方法开展研究，提出一种新的基于时频域特征参量结合支持向量机的目标检测方法。首先，从信息论角度构建无线电引信信道泄漏与信道保护模型，定量研究低信噪比条件下目标/强相干干扰信号作用下引信响应机理，获得引信目标与干扰信号的特征矢量空间模型；其次，提取目标信号经验模态分解后的本征模态函数调幅带宽、调频带宽特征，并引入近似熵特征参量，采用不确定度最小的信息熵原理实现特征参量的优化选择与融合；最后，基于二分类SVM方法给出目标与干扰信号的分类超平面，基于单分类SVM方法给出目标信号特征参量的特征值区间。本项目旨在解决高动态低信噪比强相干干扰下引信近场目标实时检测问题，研究成果将为提高引信的抗干扰能力提供有力的理论与技术支撑。

项目针对复杂电磁环境下无线电引信提高其目标识别与战场生存力的现实需求，研究了低信噪比高动态交会条件下调频引信目标与干扰信号特征提取与识别问题。首先构建了无线电引信信道泄漏与信道保护模型，定量研究了低信噪比条件下目标/强相干干扰信号下引信响应特性，提出了无线电引信抗干扰三层信道保护与量化表征方法；而后，构建了调频引信分布式目标模型，提出了以引信体目标信号的IMF调幅带宽、调频带宽和近似熵作为特征参量的目标/干扰检测方法，并经过对抗仿真与实验验证，结果表明该检测方法对目标/干扰识别率不小于90%，解决了引信目标与强相干干扰信号特征提取与识别难题；最后，提出了引信近场体目标固有特征的支持向量机目标检测方法并基于硬件实现，采用线下训练线上识别的方法有效解决了算法实时性问题，仿真和实验测试结果证明了该方法的有效性与可行性，对提升复杂战场电磁环境下引信的抗干扰能力提供了有力的理论支撑与技术储备。