GTEM室中无人车系统的电磁辐照效应研究

为验证无人装备的电磁环境适应性问题,在GTEM室中用连续波对无人车系统及其传感器组件进行了电磁辐照试验。发现无人车系统整体敏感频率为150 MHz、200 MHz和250 MHz,无人车系统对200 MHz的连续波干扰最敏感,到靶场强24.9 V/m时会发生碰撞。同时给出了敏感频率下传感器的故障场强阈值。被试无人车系统工作日志和传感器数据的分析表明,受电磁干扰影响,无人车系统感知模块会出现激光雷达数据异常、毫米波雷达死机、摄像头传输线缆连接中断等现象,最终造成无人车构建地图异常、异常停止、碰撞等故障。该文为无人装备的电磁防护设计提供了参考,有助于我国智能装备在复杂电磁环境下的安全性研究。

基于符号回归的静电电压测试补偿算法

为解决非接触式静电电位传感器测量数据的补偿问题,文中提出一种基于符号回归的波形补偿算法。设计一种非接触式静电电位传感器校准实验平台,通过遍历测试距离、校准电压和校准频率的参数组合,获得测试结果的样本训练数据;再利用符号回归方法建立从测试结果到校准电压的补偿函数模型,以降低测量距离对测量结果的影响。实验结果表明,基于符号回归的波形补偿算法可行、有效,能够为提高非接触式电位测量的准确性提供一种新手段。另外,基于模型参数与补偿结果,提出一种符号回归的改善方法,即引入损失函数用来减小函数模型的复杂度,利用神经网络卷积学习的方法发现种群与原始数据中的隐藏特征,从而明确演化方向,并针对特定的问题设定特定的评价指标,进一步优化符号回归模型的准确性。

电磁防护仿生研究进展

随着电磁环境日趋复杂及电子系统电磁敏感度不断升高,传统电磁防护技术的不足正日渐突出,电子系统的电磁环境适应性面临严峻挑战,亟需创新电磁防护理论与技术。受生物鲁棒性启发以提高电子系统的抗扰度为目的开展电磁防护仿生研究,为解决这些科学难题提供了新的思路,并引起国内外研究者的广泛关注。重点对生物鲁棒性启发的电磁防护仿生研究现状及相关研究进行了较全面的综述。具体阐述了开展电磁防护仿生研究的动机、内涵及研究路线,综述了近年来国内外开展生物鲁棒性启发的电磁防护仿生研究的最新进展,指出了该研究面临的挑战及未来潜在的研究方向。

Principles of electromagnetic protection bionics and research of fault self-recovery mechanism

As we have borrowed the wisdom from biological neural network, the characteristic of interference- proof in information transferring process has been carefully studied. Our study includes the response of electrical-interference-proof circuit in electromagnetic environment, by means of bionics redesign. In this way, new principles, new technics and new ways are adopted to solve the unavoidable situations in traditional electro- magnet protection design. Based on the Boolean model, the complex bionic topological networks were firstly built and the charactoristics of degeneracy were analyzed, then the bidirectional actions between degeneracy in networks and the robustness of functions were proved in order to fill up the gap to deal with the need of interference-proof. Based on the creation of the neuron networks and the electrical synapses ＇ charac- teristic of plasticity, the impact of the distribution in the weight of synapse on the character of network self- recovery was studied afterwards so that the corresponding mechanism could be defined.