一种具有较强泛化能力的神经网络模型研究与应用

分析了影响神经网络模型泛化能力的因素。以电解铝过程中氧化铝浓度的神经网络软测量为例 ,提出了利用先验知识确定网络结构 ,采用特定实验保证样本数量和质量 ,离线训练加在线学习修正模型等措施 ,改善神经网络模型的泛化能力。现场应用表明这些措施是有效的。这样建立的神经网络模型准确 ,泛化能力强 ,为实现过程的先进控制提供了可靠保障     

振动对空间光通信系统误码率影响的分析

在空间光通信系统中，由于卫星平台的振动，使得光发射机发生振动，从而对通信系统的误码率产生影响，本文针对光射光束为高斯光束时，振动对误码率的影响问题进行研究。结果表明，在振动时不仅振幅影响着误码率，而且波长也对误码率有影响，但通信距离对误码率的影响不大，一般可以忽略。     

线性多变量系统闭环特征结构的最优配置

本文利用特征结构配置法所揭示的线性多变量反馈控制系统闭环特征值和特征向量的配置自由度,求解具有二次型性能指标的控制系统设计问题,使得闭环系统在具有任意给定的希望特征值集合条件下,性能指标达到极小。文中建立了最优二次型性能指标与闭环特征结构的关系,提出了逐个优化特征向量的算法,这一算法概念直观并具有较高的运算效率,可以推广到一般的特征结构优化问题。     

Time-Optimal Control of a Bounded Phase-Coordinate Process II: High-Order Systems with Multiple Control Inputs

This paper considers the problem arising from the design of an autopilot for a large booster. The motion-controlling actuators of the booster have both position and rate limits. The problem is formulated as a bounded phase-coordinate problem and analyzed by the ``backing out of the target' procedure. A method of constructing the optimal control is presented. An example of an oscillatory system with two control inputs is given, and the optimal control is expressed as an explicit time function.     

Comments on the Communication and Data Problems Associated with a Mars Trip During a Conjunction Phase

The analysis and comments presented in this paper are meant to establish the general communication parameters associated with Martian flyby probes and with lander and manned vehicles. Fundamental data transfer problems are reviewed to define comparisons and trends of tradeoffs for future studies. Selected focal points are based upon the long propagation path length, with inherent time delays, and the high noise produced by the sun. These problems are magnified because large quantities of data must be obtained to satisfy the needs of the scientific community and the curiosity of an interested public. A comparison of two communication systems is provided: the microwave spectrum and the optical spectrum, as represented by the microwaves at 2.3 GHz and the laser at 6328 ?. A method of cost effectiveness or value received from space missions (a criterion of power input for data quantity received) is also presented.     

Ground and on-orbit command and data handling architectures for the Active Rack Isolation System microgravity flight experiment

The Active Rack Isolation System [ARIS] International Space Station [ISS] Characterization Experiment, or ARIS-ICE for short, is a long duration microgravity characterization experiment aboard the ISS. The objective of the experiment is to fully characterize active microgravity performance of the first ARIS rack deployed on the ISS. Efficient ground and on-orbit command and data handling [C&DH] segments are the crux in achieving the challenging objectives of the mission. The objective of the paper is to provide an overview of the C&DH architectures developed for ARIS-ICE, with the view that these architectures may serve as a model for future ISS microgravity payloads. Both ground and on-orbit segments, and their interaction with corresponding ISS C&DH systems are presented. The heart of the on-orbit segment is the ARIS-ICE Payload On-orbit Processor, ARIS-ICE POP for short. The POP manages communication with the ISS C&DH system and other ISS subsystems and payloads, enables automation of test/data collection sequences, and provides a wide range of utilities such as efficient file downlinks/uplinks, data post-processing, data compression and data storage. The hardware and software architecture of the POP is presented and it is shown that the built-in functionality helps to dramatically streamline the efficiency of on-orbit operations. The ground segment has at its heart special ARIS-ICE Ground Support Equipment [GSE] software developed for the experiment. The software enables efficient command and file uplinks, and reconstruction and display of science telemetry packets. The GSE software architecture is discussed along with its interactions with ISS ground C&DH elements. A test sequence example is used to demonstrate the interplay between the ground and on-orbit segments.     

One year old and growing: a status report of the International Space Station and its partners.

The International Space Station (ISS), as the largest international science and engineering program in history, features unprecedented technical, cost, scheduling, managerial, and international complexity. A number of major milestones have been accomplished to date, including the construction of major elements of flight hardware, the development of operations and sustaining engineering centers, astronaut training, and eight Space Shuttle/Mir docking missions. International partner contributions and levels of participation have been baselined, and negotiations and discussions are nearing completion regarding bartering arrangements for services and new hardware. As ISS is successfully executed, it can pave the way for more inspiring cooperative achievements in the future.     

Solar particle events and the International Space Station

The high inclination orbit for the International Space Station poses a risk to astronauts on EVA during occasional periods of enhanced high energy particle flux from the sun known as Solar Particle Events. We are currently unable to predict these events within the few-hour lead time required for evasive action. Compounding the threat is the fact that station construction occurs during increasing solar activity and through the peak of the solar cycle. In this paper we present an overview of the risk, the current methods to provide forecasts of SPEs, and potential risk mitigation options.     

Control-structure interaction study of the proposed space station freedom

A relatively general formulation for studying the dynamics and control of an arbitrary spacecraft with interconnected flexible bodies has been developed accounting for transient system properties, shift in the center of mass, shear deformations, rotary inertias and geometric nonlinearities. This self-contained, comprehensive, numerical algorithm using system modes is applicable to a large class of spacecraft configurations of contemporary and future interests. Here, versatility of the approach is demonstrated through the dynamics and control studies aimed at the evolving Space Station Freedom.     

Remote sensing of planetary properties and biosignatures on extrasolar terrestrial planets

The major goals of NASA's Terrestrial Planet Finder (TPF) and the European Space Agency's Darwin missions are to detect terrestrial-sized extrasolar planets directly and to seek spectroscopic evidence of habitable conditions and life. Here we recommend wavelength ranges and spectral features for these missions. We assess known spectroscopic molecular band features of Earth, Venus, and Mars in the context of putative extrasolar analogs. The preferred wavelength ranges are 7-25 microns in the mid-IR and 0.5 to approximately 1.1 microns in the visible to near-IR. Detection of O2 or its photolytic product O3 merits highest priority. Liquid H2O is not a bioindicator, but it is considered essential to life. Substantial CO2 indicates an atmosphere and oxidation state typical of a terrestrial planet. Abundant CH4 might require a biological source, yet abundant CH4 also can arise from a crust and upper mantle more reduced than that of Earth. The range of characteristics of extrasolar rocky planets might far exceed that of the Solar System. Planetary size and mass are very important indicators of habitability and can be estimated in the mid-IR and potentially also in the visible to near-IR. Additional spectroscopic features merit study, for example, features created by other biosignature compounds in the atmosphere or on the surface and features due to Rayleigh scattering. In summary, we find that both the mid-IR and the visible to near-IR wavelength ranges offer valuable information regarding biosignatures and planetary properties; therefore both merit serious scientific consideration for TPF and Darwin.