基于广义回归神经网络集成的宽带波束形成算法

首先构造宽带波束成形所需要的协方差矩阵,利用基于粒子群优化算法对核主成分分析方法和广义回归神经网络进行了优化。在对神经网络的输入变量进行降维处理后,生成多个复杂度低的泛回归神经网络模型。利用提出的基于聚类启发式集成算法求出波束成形时的权系数,既考虑了网络的差异性,又考虑了网络的正确性。仿真结果表明,提出的基于聚类启发式神经网络集成的波束形成算法在网络结构十分简单的情况下,仍然具有较好的性能。     

一种基于粒子滤波的双目视觉惯导SLAM技术

针对智能无人系统的定位与地图构建问题,提出一种基于粒子滤波的双目视觉惯导SLAM（即时定位与地图构建）方法。算法基于传统粒子滤波思想设计实现,后端处理时仅对位姿状态量进行滤波,有效解决了传统算法的维度爆炸问题,减少计算量的同时保证一定的精确度,兼顾SLAM算法精确性和即时性的要求。实验结果表明,方法整体定位精度相对误差低于5%,在光照条件适宜的小型场景效果更佳,误差低于3%,能够达到分米级精度,证明了SLAM方法能够完成SLAM系统的要求,实现即时定位与地图构建功能,具有一定的准确性、稳定性和鲁棒性。粒子滤波能够应用于视觉惯导SLAM领域并达到较高的精度要求。     

On the ability of sliding mode and LQR controllers optimized with PSO in attitude control of a flexible 4-DOF satellite with time-varying payload

Precise pointing of the satellite and its payload is essential in the accurate accomplishment of a space mission. In this study, the system of a satellite and its payload are considered as 4-DOF equations of motion. The time-varying payload can observe one direction of the Earth independently, and the satellite can point to the Earth station by its 3-DOF motions simultaneously. Sliding mode and LQR controllers are designed for damping disturbances, and consequently high pointing accuracy. Environmental disturbances and the associated time delay of Low Earth Orbit (LEO) are applied to the system. An algorithm based on Particle Swarm Optimization (PSO) is proposed to find the optimum values of variables and Normalized Integral Square Error (NISE) of the two aforementioned controllers. Numerical simulations indicate the optimized magnitudes of target detection errors and control efforts in four directions. The results revealed that PSO-SMC can finely track the time-varying payload and has better efficiency in comparison with PSO-LQR.     

An efficient analytical homogenization technique for mechanical-hygrothermal responses of unidirectional composites with applications to optimization and multiscale analyses

The elasticity-based Locally Exact Homogenization Theory (LEHT) is extended to study the mechanical-hygrothermal behaviors of unidirectionally-reinforced composites. Based on the framework developed previously, thermal and moisture effects are incorporated into the LEHT to study the homogenized and localized responses of heterogeneous materials, which are validated using available analytical and numerical techniques. The LEHT programs are then encapsulated as subroutines with Input/Output (I/O) interfaces, to be readily applied in different computational scenarios. In order to illustrate the efficiency of the LEHT, the theory is firstly coupled to the Particle Swarm Optimization (PSO) algorithm in order to minimize the axial thermal expansion mismatch in hexagonal and square fiber arrays by tailoring the fiber volume fraction. The LEHT is then implemented into the lamination theory to study fabrication-induced residual stresses arising during the cool-down process which introduces local laminate stresses owing to thermo-mechanical property mismatch between plies. Both of these applications illustrate the efficiency and accuracy of the LEHT in generating effective properties and local stress distributions, making the theory a golden standard in validating other analytical or numerical techniques as well as a reliable tool in composite design and practice for professionals and non-professionals alike.     

Design and performance analysis of position-based impedance control for an electrohydrostatic actuation system

Electrohydrostatic actuator (EHA) is a type of power-by-wire actuator that is widely implemented in the aerospace industry for flight control, landing gears, thrust reversers, thrust vector control, and space robots. This paper presents the development and evaluation of position-based impedance control (PBIC) for an EHA. Impedance control provides the actuator with compliance and facilitates the interaction with the environment. Most impedance control applications utilize electrical or valve-controlled hydraulic actuators, whereas this work realizes impedance control via a compact and efficient EHA. The structures of the EHA and PBIC are firstly introduced. A mathematical model of the actuation system is established, and values of its coefficients are identified by particle swarm optimization. This model facilitates the development of a position controller and the selection of target impedance parameters. A nonlinear proportional-integral position controller is developed for the EHA to achieve the accurate positioning requirement of PBIC. The controller compensates for the adverse effect of stiction, and a position accuracy of 0.08 mm is attained. Various experimental results are presented to verify the applicability of PBIC to the EHA. The compliance of the actuator is demonstrated in an impact test.     

Ascent guidance law for a horizontal take-off vehicle with a multi-combined cycle engine

This paper researches the ascent guidance law for the vehicle with a multi-combined cycle propulsion. The guidance law comprises two parts, namely, the off-line optimal trajectories generation and online guidance. With respect to the off-line part, disturbances are discretized and incorporated into the trajectory optimization problem; subsequently, a set of trajectories is calculated to constitute a database. To quickly obtain a database that comprises a large number of trajectories, a novel ascent profile is proposed with respect to height and velocity. Based on this profile, only inequity constraints exist in the optimization model, and the original optimization problem is converted to a parameter searching problem. The optimal trajectories are calculated using a hybrid optimization method that comprises a particle swarm optimization (PSO) method and the Hooke-Jeeves (HJ) method. With respect to online guidance, the profile is updated using a radial basis function neural network (RBFNN) based on the current flight states and the database. Simulation validates the efficiency of the proposed optimization method by comparing the method with the pseudospectral method; the robustness of the guidance law is also validated using Monte Carlo simulation.     

A Partially Orthogonal EnKF approach to atmospheric density estimation using orbital debris

A key requirement for accurate trajectory prediction and space situational awareness is knowledge of how non-conservative forces affect space object motion. These forces vary temporally and spatially, and are driven by the underlying behavior of space weather particularly in Low Earth Orbit (LEO). Existing trajectory prediction algorithms adjust space weather models based on calibration satellite observations. However, lack of sufficient data and mismodeling of non-conservative forces cause inaccuracies in space object motion prediction, especially for uncontrolled debris objects. The uncontrolled nature of debris objects makes them particularly sensitive to the variations in space weather. Our research takes advantage of this behavior by utilizing observations of debris objects to infer the space environment parameters influencing their motion.The hypothesis of this research is that it is possible to utilize debris objects as passive, indirect sensors of the space environment. We focus on estimating atmospheric density and its spatial variability to allow for more precise prediction of LEO object motion. The estimated density is parameterized as a grid of values, distributed by latitude and local sidereal time over a spherical shell encompassing Earth at a fixed altitude of 400 km. The position and velocity of each debris object are also estimated. A Partially Orthogonal Ensemble Kalman Filter (POEnKF) is used for assimilation of space object measurements to estimate density.For performance comparison, the scenario characteristics (number of objects, measurement cadence, etc.) are based on a sensor tasking campaign executed for the High Accuracy Satellite Drag Model project. The POEnKF analysis details spatial comparisons between the true and estimated density fields, and quantifies the improved accuracy in debris object motion predictions due to more accurate drag force models from density estimates. It is shown that there is an advantage to utilizing multiple debris objects instead of just one object. Although the work presented here explores the POEnKF performance when using information from only 16 debris objects, the research vision is to utilize information from all routinely observed debris objects. Overall, the filter demonstrates the ability to estimate density to within a threshold of accuracy dependent on measurement/sensor error. In the case of a geomagnetic storm, the filter is able to track the storm and provide more accurate density estimates than would be achieved using a simple exponential atmospheric density model or MSIS Atmospheric Model (when calm conditions are assumed).     

基于粒子群优化核极限学习机的北斗超快速钟差预报

针对卫星钟差序列中非线性特性较为复杂和超快速钟差预报精度较低的问题,将核极限学习机算法引入到北斗超快速钟差预报中。首先,将极限学习机进行优化,引入粒子群优化算法来选择核极限学习机所需的核参数和正则化参数;然后,将优化后的方法应用到超快速钟差预报中,并给出了利用该方法进行超快速钟差预报的步骤;最后,在分析iGMAS提供的实测北斗超快速钟差数据的基础上,选用单天和多天数据进行短期预报。结果表明：在短期预报6h范围内,利用本文提供的优化方法解算得到的超快速钟差预报精度明显优于二次多项式模型和周期项模型,并且采用此方法得到的超快速钟差预报产品与iGMAS提供的超快速钟差预报产品(ISU-P)相比,GEO、IGSO和MEO卫星的预报精度分别提升了50.51%、46.98%、40.67%,其与最终精密钟差的符合程度显著 增强 。     

基于贝叶斯方法的HPRF雷达跟踪解距离模糊方法(英文)

高脉冲重复频率(HPRF)雷达目标跟踪需要解决距离模糊问题。本文通过把脉冲间隔数和脉冲间隔变化量作为目标的离散状态,目标的距离和径向速度作为目标的连续状态,将HPRF雷达的跟踪和解距离模糊问题转换为贝叶斯框架下的混合估计问题并且利用粒子滤波方法实现。仿真结果表明本文方法在脉冲间隔数变化时仍可以正确地解距离模糊,并且跟踪精度比多假设方法高。     

改进的迭代粒子滤波法在GPS数据处理中的应用

研究粒子滤波方法(PF)处理GPS测量数据.采用迭代Kalman滤波方法产生粒子滤波器中的粒子,得到更精确的目标状态估计.同时利用目标运动方程向后外推目标状态,综合利用外推的目标状态以及历史现测数据确定各粒子的权值,降低现测随机误差的影响,从而提高粒子滤波器的性能.采用改进的粒子滤波方法处理GPS测量数据,仿真结果表明...