Magnetosphere Imaging Instrument (MIMI) on the Cassini Mission to Saturn/Titan

The magnetospheric imaging instrument (MIMI) is a neutral and charged particle detection system on the Cassini orbiter spacecraft designed to perform both global imaging and in-situ measurements to study the overall configuration and dynamics of Saturn’s magnetosphere and its interactions with the solar wind, Saturn’s atmosphere, Titan, and the icy satellites. The processes responsible for Saturn’s aurora will be investigated; a search will be performed for substorms at Saturn; and the origins of magnetospheric hot plasmas will be determined. Further, the Jovian magnetosphere and Io torus will be imaged during Jupiter flyby. The investigative approach is twofold. (1) Perform remote sensing of the magnetospheric energetic (E > 7 keV) ion plasmas by detecting and imaging charge-exchange neutrals, created when magnetospheric ions capture electrons from ambient neutral gas. Such escaping neutrals were detected by the Voyager l spacecraft outside Saturn’s magnetosphere and can be used like photons to form images of the emitting regions, as has been demonstrated at Earth. (2) Determine through in-situ measurements the 3-D particle distribution functions including ion composition and charge states (E > 3 keV/e). The combination of in-situ measurements with global images, together with analysis and interpretation techniques that include direct “forward modeling’’ and deconvolution by tomography, is expected to yield a global assessment of magnetospheric structure and dynamics, including (a) magnetospheric ring currents and hot plasma populations, (b) magnetic field distortions, (c) electric field configuration, (d) particle injection boundaries associated with magnetic storms and substorms, and (e) the connection of the magnetosphere to ionospheric altitudes. Titan and its torus will stand out in energetic neutral images throughout the Cassini orbit, and thus serve as a continuous remote probe of ion flux variations near 20R _S (e.g., magnetopause crossings and substorm plasma injections). The Titan exosphere and its cometary interaction with magnetospheric plasmas will be imaged in detail on each flyby. The three principal sensors of MIMI consists of an ion and neutral camera (INCA), a charge–energy–mass-spectrometer (CHEMS) essentially identical to our instrument flown on the ISTP/Geotail spacecraft, and the low energy magnetospheric measurements system (LEMMS), an advanced design of one of our sensors flown on the Galileo spacecraft. The INCA head is a large geometry factor (G ∼ 2.4 cm² sr) foil time-of-flight (TOF) camera that separately registers the incident direction of either energetic neutral atoms (ENA) or ion species (≥5^∘ full width half maximum) over the range 7 keV/nuc < E < 3 MeV/nuc. CHEMS uses electrostatic deflection, TOF, and energy measurement to determine ion energy, charge state, mass, and 3-D anisotropy in the range 3 ≤ E ≤ 220 keV/e with good (∼0.05 cm² sr) sensitivity. LEMMS is a two-ended telescope that measures ions in the range 0.03 ≤ E ≤ 18 MeV and electrons 0.015 ≤ E≤ 0.884 MeV in the forward direction (G ∼ 0.02 cm² sr), while high energy electrons (0.1–5 MeV) and ions (1.6–160 MeV) are measured from the back direction (G ∼ 0.4 cm² sr). The latter are relevant to inner magnetosphere studies of diffusion processes and satellite microsignatures as well as cosmic ray albedo neutron decay (CRAND). Our analyses of Voyager energetic neutral particle and Lyman-α measurements show that INCA will provide statistically significant global magnetospheric images from a distance of ∼60 R _S every 2–3 h (every ∼10 min from ∼20 R _S). Moreover, during Titan flybys, INCA will provide images of the interaction of the Titan exosphere with the Saturn magnetosphere every 1.5 min. Time resolution for charged particle measurements can be < 0.1 s, which is more than adequate for microsignature studies. Data obtained during Venus-2 flyby and Earth swingby in June and August 1999, respectively, and Jupiter flyby in December 2000 to January 2001 show that the instrument is performing well, has made important and heretofore unobtainable measurements in interplanetary space at Jupiter, and will likely obtain high-quality data throughout each orbit of the Cassini mission at Saturn. Sample data from each of the three sensors during the August 18 Earth swingby are shown, including the first ENA image of part of the ring current obtained by an instrument specifically designed for this purpose. Similarily, measurements in cis-Jovian space include the first detailed charge state determination of Iogenic ions and several ENA images of that planet’s magnetosphere.This revised version was published online in July 2005 with a corrected cover date.     

基于BP人工神经网络的GPS／SINS组合导航算法

基于扩展Kalman滤波的GPS／SINS组合导航算法，需要对原始的非线性连续系统模型进行线性化和离散化处理，要求系统噪声和测量噪声为零均值的高斯白噪声，且易于出现滤波器发散。BP人工神经网络毋需对所求解的问题建模，能够很好地逼近系统非线性特性，获得较高精度的导航定位信息；还具有计算过程稳定，不涉及矩阵求逆，不需要迭代逼近，以及容易实现并行处理等优点。本文设计适用于GPS／SINS组合导航系统的BP网络模型，并在标准的BP算法基础上，采用共轭梯度法改进网络训练速度及精度。最后，通过仿真算例说明BP网络方法用于GPS／SINS组合导航计算的可行性。     

基于ANFIS的蒸汽发生器水位实时控制系统的设计

从蒸汽发生器的水位特性出发，对其水位高度控制原理进行了深入细致的研究。鉴于控制对象的模糊性、不确定性和非线性，采用自适应神经元模糊推理(Adaptive neuron fuzzy inference system，ANFIS)技术，建立了模糊控制规则库，实现了对蒸汽发生器水位的智能控制。本文详细阐述了ANFIS技术的结构、控制方式和系统的主要功能，完成了软、硬件的综合设计，并进行了仿真研究。控制系统的硬件采用了DSP芯片，以保证系统的实时性；软件采用了模糊一神经网络算法，以克服系统模型的不确定性。仿真结果表明，该控制系统工作稳定可靠，具有较高的控制精度和较强的鲁棒性。     

关于稀疏分布存贮模型中A连接权阵的讨论

Ｋａｎｅｒｖａ的稀疏分布存贮模型由于对寻址地址采用了稀疏编码，对数据采用了分布式存贮，从而解决了大维数向量的输入问题，ＳＤＭ实际上是一个由输入层，中间层和输出层组成的三层前向网络，其中神经元间的互迦权值在输入层与中间层是预置的（用矩阵Ａ表示），中间层与输出层的连接权阵Ｃ由外积法得到，文中假定在相同的学习规则下，就信噪比意义而言，Ａ的均匀预置能使ＳＤＭ获得最优性能，从而为Ａ的预置提供了理论依据。     

力控制反馈信号的神经网络法校正

对机器人的力反馈，主动柔顺控制和双向联想记忆的神经网络进行了综合研究，针对机器人腕力传感器输出信号的部分失真，提出了机器人擦洗平面玻璃力反馈信号神经校正的方法，并建立了相应的神经网络结构，在ＡｄｅｐｔＴｈｒｅｅ精密装配机器人上进行了试验验证，取得了满意的效果。     

三维有限元网格的自动划分

本文作者从研究结构的数学模型入手,讨论了建立三维结构有限元计算模型的方法。成功地获得了一些具有应用代表性的三维板壳和块体结构有限元计算网格的自动划分原理和算法。     

神经网络与D-S理论结合的HRR识别研究

基于D-S(Dempster-Shafer)证据理论,比较和研究了相关数据和不相关数据的融合方法,分析了多传感器数据融合的算法:集中式融合算法和分布式融合算法.经过实验证明,执行分布式有反馈融合算法时的效果最好.然后利用该算法,提出了和线性内插神经网络相结合的识别方法.利用4种飞机的步进频率雷达的高分辨率一维距离像,将神经网络的识别结果作为证据分别送入传感器进行融合,进行识别研究.实验证明,与单纯利用神经网络的方法比较,目标的正确识别率得到了改善.     

基于联邦滤波器的新型故障检测结构及算法

提出了一种基于联邦卡尔曼滤波器的故障检测结构,该结构利用各局部滤波器和参考滤波器共有状态之间的残差进行故障检测.并提出了2种故障检测算法:χ2检验法和Elman神经网络检验法.以组合导航系统为例进行了仿真研究,和其它算法相比该算法计算简单、可靠,不但可以快速检测出外部传感器及参考系统故障,且具有很好的容错性能,能快速检测出故障并进行隔离,使融合后系统依然保持较高精度.     

非线性随机动态系统学习建模方法

在动态神经网络及扩展卡尔曼滤波算法的基础上，提出了对非线性随机动态系统进行学习建模的迭代算法，用这种方法对非线性随机系统建模，可以获得更准确伯系统模型，并可对非线性随机系统进行状态估计。最后给出相应算法及仿真结果。     

无刷直流电动机的新型自适应模糊神经控制

 为无刷直流电动机提出了一种自适应模糊神经控制方法.这是一种建立在开关控制、模糊控制和自适应控制相结合基础上的控制方法,并用神经网络实现了模糊控制器和自适应机构.在无刷直流电动机的双闭环调速系统中,电流控制器是PI控制器;转速控制器是由1个开关控制器和1个包括自适应机构在内的模糊控制器相结合组成的,且用1个3层前向神经网络离线学习实现了模糊控制器,学习算法采用的是改进的BP算法.用1个单神经元通过在系统运行过程中的动态学习实现了自适应机构,学习算法选用了有监督的Hebb学习算法.由电机所处的运行阶段决定哪一个控制器工作.此控制算法的仿真结果说明,它使系统具有良好的动、静态特性和自适应性.