仿生硬件及其进展

文章详细介绍了仿生硬件的概念、主要特征及用途 ,阐述了仿生硬件的进化原理及实现思想 ,重点分析了仿生硬件的关键技术———可进化硬件电路、硬件进化方法 ,并总结了研究状况与最新进展 ,讨论了需要解决的核心难点问题及其解决思路 ,最后指出了仿生硬件的研究方向及发展前景。     

直升机模糊神经控制模型

介绍了多变量系统的模糊神经控制模型、直升机飞行模糊控制规则建立的思路;建立了一个直升机模糊神经控制模型,并给出利用神经网络学习模糊控制规则的方法和步骤。     

Radio plasma imager simulations and measurements

The Radio Plasma Imager (RPI) will be the first-of-its kind instrument designed to use radio wave sounding techniques to perform repetitive remote sensing measurements of electron number density (N _e) structures and the dynamics of the magnetosphere and plasmasphere. RPI will fly on the Imager for Magnetopause-to-Aurora Global Exploration (IMAGE) mission to be launched early in the year 2000. The design of the RPI is based on recent advances in radio transmitter and receiver design and modern digital processing techniques perfected for ground-based ionospheric sounding over the last two decades. Free-space electromagnetic waves transmitted by the RPI located in the low-density magnetospheric cavity will be reflected at distant plasma cutoffs. The location and characteristics of the plasma at those remote reflection points can then be derived from measurements of the echo amplitude, phase, delay time, frequency, polarization, Doppler shift, and echo direction. The 500 m tip-to-tip X and Y (spin plane) antennas and 20 m Z axis antenna on RPI will be used to measures echoes coming from distances of several R _E. RPI will operate at frequencies between 3 kHz to 3 MHz and will provide quantitative N _e values from 10^–1 to 10⁵ cm^–3. Ray tracing calculations, combined with specific radio imager instrument characteristics, enables simulations of RPI measurements. These simulations have been performed throughout an IMAGE orbit and under different model magnetospheric conditions. They dramatically show that radio sounding can be used quite successfully to measure a wealth of magnetospheric phenomena such as magnetopause boundary motions and plasmapause dynamics. The radio imaging technique will provide a truly exciting opportunity to study global magnetospheric dynamics in a way that was never before possible.     

The Radio Plasma Imager investigation on the IMAGE spacecraft

Radio plasma imaging uses total reflection of electromagnetic waves from plasmas whose plasma frequencies equal the radio sounding frequency and whose electron density gradients are parallel to the wave normals. The Radio Plasma Imager (RPI) has two orthogonal 500-m long dipole antennas in the spin plane for near omni-directional transmission. The third antenna is a 20-m dipole along the spin axis. Echoes from the magnetopause, plasmasphere and cusp will be received with the three orthogonal antennas, allowing the determination of their angle-of-arrival. Thus it will be possible to create image fragments of the reflecting density structures. The instrument can execute a large variety of programmable measuring options at frequencies between 3 kHz and 3 MHz. Tuning of the transmit antennas provides optimum power transfer from the 10 W transmitter to the antennas. The instrument can operate in three active sounding modes: (1) remote sounding to probe magnetospheric boundaries, (2) local (relaxation) sounding to probe the local plasma frequency and scalar magnetic field, and (3) whistler stimulation sounding. In addition, there is a passive mode to record natural emissions, and to determine the local electron density, the scalar magnetic field, and temperature by using a thermal noise spectroscopy technique.     

检定几何量仪器标准器组测量不确定度评定

以标准线纹尺、小量块组、多面棱体、刻线样板、标准半圆球、渐开线样板检定／校准几何量光学仪器时,标准器组的测量不确定度评定。     

一种L波段国产GaN固态雷达发射机

为了应对国产雷达所有元器件均国产化需求,设计了一款工作频点在1.2GHZ~ 1.4 GHz的国产氮化镓（gallium nitride,GaN）固态雷达发射机。使用国产GaN功率模块对发射系统最大限度地进行小型化设计,降低组件的重量,加强设备的高可靠性及维护方便性设计,进一步完善GaN功率模块在线测试功能,使产品具有高可靠性和完善的雷达故障诊断功能。增加系统的监控、检测及显示功能,并将各种工作状态信号都传输到主监控上,使系统的各种工作状态都能够在终端上显示,使操作人员通过终端界面便可以检查发射系统的各种工作参数,判断发射机的工作状态。在电磁兼容方面通过谐波滤波器对二次谐波的抑制,测试结果表明,发射系统输出信号的二次谐波小于等于-66dB。     

基于反电势形状函数法的无刷直流电动机直接转矩控制

采用两相导通模式时通过逆变器开关状态不能确定电压空间矢量,现有方案是通过实时测量3个相电压来解决这一问题,但系统变得复杂;此外现有方案在转矩观测中包含的微分项会降低控制精度,增大计算量,影响系统实时性。针对这两点不足,提出反电势形状函数法,在软件中实现了相电压的实时计算,使转矩观测变得简洁,更易于工程实用。最后给出的仿真与实验结果证明了反电势形状函数法用于无刷直流电机直接转矩控制中的可行性与有效性。     

基于优化最小二乘支持向量机的襟翼趋势预测

将最小二乘支持向量机(LS-SVM)应用于飞机襟翼状态趋势研究。首先,通过分析飞机襟翼故障与襟翼动作耗时参数的关系,提出了利用动作耗时趋势来确定该系统未来状态的方法。然后,使用最小二乘支持向量机建立耗时回归预测模型,采用最终预报误差(FPE)准则确定回归模型嵌入维数,提出了自适应网格搜索法,优化最小二乘支持向量机的建模参数,从而实现比现有方法精度高,泛化性能好的预测模型。训练和测试样本取自飞行数据记录系统(QAR)中译码襟翼参数值。与神经网络模型的比较实践表明,该方法具有实用价值。     

低成本捷联惯导不对称动态误差的神经网络补偿

 针对低成本捷联惯导系统(SINS)中陀螺动态误差的不对称性在角振动条件下造成姿态漂移的问题,设计了多层前向神经网络的补偿模型。在标定模型参数时,为降低对外部参考信号测量精度的要求,提出用姿态解算的最终误差作为网络优化目标的训练方法。由于最终的姿态误差不是网络的期望输出,无法采用有导师的训练方法,为此采用了微粒群优化算法。仿真实验结果表明:补偿后的陀螺动态误差的不对称度减小了一个数量级。     

考虑先验信息可信度的后验加权Bayes估计

 先验信息失真及先验样本数量过大会扭曲小子样条件下Bayes融合评估的效果,引入可信度可以改善这个问题。现有的可信度度量方法大都直接基于数据层,即通过判断两种样本是否属于同一分布,这种度量在小子样情况下不太可信,为此提出了一种基于数据物理来源的可信度度量方法。同时归纳了可信度融合评估的准则,分析了现有可信度融合评估方法的不足,并结合正态分布的参数估计问题,给出了一种考虑先验信息可信度的后验加权Bayes估计方法。理论分析和算例都证实了该方法优于常用的Bayes估计方法。