基于协同进化的多无人机航迹规划研究

关于多无人机航迹优化研究,针对复杂环境下多无人机(UAV)系统的航迹规划,达到摧毁目标最大化,解决不同无人机之间的协同和防撞问题,提出了一种利用合作型协同进化算法的多无人机三维航迹规划方法.利用数字地图建立了无人机安全飞行曲面,采用并行进化的方案,将每个无人机航迹规划当作一个子问题,通过协同函数和无人机间的防撞设计实现各无人机间的时间协同和空间防撞.各子种群采用自适应的进化方法,在保持多样性的同时,保证了算法收敛的快速性.仿真结果表明,算法有效实用,能快速得到各无人机的低空突防三维航迹,可为多无人机航迹优化提供手段.     

基于多级并行策略的复杂产品多学科设计优化

针对复杂产品设计需要进行多学科协作设计和优化的问题,结合计算机应用技术提出基于多级并行策略的多学科优化方法。该方法基于过程建模的层次化优化思想,实现复杂产品设计过程的自动化和优化。减速器标准多学科优化算例说明该方法可实现不同学科的层次化并行优化。将该方法与其他传统的多学科优化方法进行比较,验证该方法的高效性和最优设计的准确性。     

布局优化子问题的改进遗传神经网络算法

带性能约束的航天舱布局问题可分解为有限多个子问题,每个子问题克服了关于优化变量的时断时续性。本文针对子问题(关于同构布局等价类),首先构造了用于产生与已知布局方案同构的布局方案的优化算法,然后在给出组合变异策略的基础上,设计了连续空间上基于实数编码的改进遗传神经网络算法。将该算法应用于二维布局优化子问题,数值实验表明该遗传神经网络进行布局逼近是有效的。这种方法是对布局问题求解的有效探索。     

基于协同进化博弈的多学科设计优化

复杂系统的设计问题可以非层次分解为并行的多个子空间优化设计问题。多学科优化的迭代过程可看成子空间博弈的过程。各冲突子目标协商一致条件下,子空间合作博弈的均衡点能达成原系统的整体最优,并给出协同进化算法求解博弈的Nash均衡点的计算框架。以某型民用客机的总体优化设计为例,将其分解成气动和重量两个子空间优化。设计变量不重叠地分布于各子空间,两冲突子目标分配相同权值,线性加权组合而形成的单目标作为各子空间共同的优化目标。计算结果表明此方法是有效的。     

基于并行子空间优化的企业命名实体识别

针对企业命名实体的识别任务的过程复杂、学科交叉、实时性差等难点,提出了一种基于并行子空间优化的方法.首先,建立系统的目标-约束方程完成系统级优化;其次,再通过构建文字检测、文字识别两级模型,并考虑现存不同模型的优缺点进行模型选择的方法对涉及学科进行并行优化;随后,再使用图像阈值、灰度化、霍夫变换等算法构建两级模型的衔接;最后,通过仿真实验,验证了本文方法相比其他两级文字检测识别模型的识别准确率提高了9％,推理速度提升约20％.     

基于伪并行混合遗传算法的神经网络优化

在分析并行多物种遗传算法应用于神经网络拓扑结构的设计和学习之后，提出一种伪并行遗传（PPGA-MBP）混合算法，结合改进的BP算法对多层前馈神经网络的拓扑结构进行优化。算法编码采用基于实数的层次混合方式，允许两个不同结构的网络个体交叉生成有效子个体。利用该算法对N-Parity问题进行了实验仿真，并对算法中评价函数各部分系数和种群规模对算法的影响进行了分析。实验证明取得了明显的优化效果，提高了神经网络的自适应能力和泛化能力，具有全局快速收敛的性能。     

鱼雷外形多学科设计优化中的博弈求解方法

为了鱼雷系统多学科设计优化的设计过程,解决处理耦合变量的技术难点,为减小阻力,改进鱼雷性能,减少设计工作量,提出了一种新的多学科设计优化方法.该基于Nash均衡思想,把设计变量作为决策空间,设计目标作为博弈方,各子学科在某一策略下的响应设为博弈论中博弈方的收益.与以往的多学科设计优化方法相比,方法具有计算步骤简单,易于编程实现,各子学科可以独立进行学科的自主优化,改善鱼雷阻力特性.通过对具体工程算例进行仿真.结果验证了算法的可行性和适用性,可优化鱼雷性能,为设计提供了参考依据.     

FMM算法中问题规模与空间划分的关系分析

从编译优化和并行优化的角度出发,根据N-Body问题求解的FMM算法的原理,将算法分解为不同的子模块。详细分析了各子模块的计算特性,包括计算量分析、并行性分析、通信量分析和存储量分析。深入剖析问题规模与空间划分层数之间的关系,提出基于问题规模的空间划分策略。以实验验证了空间划分策略的可行性。     

基于Isight的CSSO算法框架

概述多学科优化设计(Multidisciplinary Design Optimization,MDO)在复杂产品设计中的必要性,阐述CSSO算法的基本内容,以齿轮减速器为例,利用Isight搭建CSSO计算框架并对每一步进行分析和说明.将CSSO计算结果与传统优化结果进行对比,结果表明CSSO算法比传统算法的计算效率高、准确性好.     

并行智能优化算法研究进展

基于种群迭代搜索的智能优化算法在农业、交通、工业等很多领域都取得了广泛的应用.但是该类算法迭代寻优的特点使其求解效率通常较低,很难应用到大规模、高维或实时性要求较高的复杂优化问题中.随并行分布式技术的发展,国内外很多学者开始着手研究智能优化算法的并行化.本文首要介绍了并行智能优化算法的基本概念;其次从协同机制、并行模型以及硬件结构3个维度综述了几类常见的并行智能优化算法,详细分析阐述了它们优点及不足;最后对并行智能优化算法的未来研究进行了展望.     

Comparison of MDO methods with mathematical examples

Recently, engineering systems are quite large and complicated. The design requirements are fairly complex and it is not easy to satisfy them by considering only one discipline. Therefore, a design methodology that can consider various disciplines is needed. Multidisciplinary design optimization (MDO) is an emerging optimization method that considers a design environment with multiple disciplines. Seven methods have been proposed for MDO. They are Multiple-discipline-feasible (MDF), Individual-discipline-feasible (IDF), All-at-once (AAO), Concurrent subspace optimization (CSSO), Collaborative optimization (CO), Bi-level integrated system synthesis (BLISS), and Multidisciplinary design optimization based on independent subspaces (MDOIS). Through several mathematical examples, the performances of the methods are evaluated and compared. Specific requirements are defined for comparison and new types of mathematical problems are defined based on the requirements. All the methods are coded and the performances of the methods are compared qualitatively and quantitatively.     

A multidisciplinary design optimization approach for high temperature aircraft engine components

Rapid turn-around time for investigating new design concepts is a primary force driving design productivity initiatives across the industry. An integration framework focusing on the collaborative nature of rapid design automation at the preliminary and detailed design stage would ensure higher quality designs from the beginning of the product design cycle. As a result, producing reliable, robust optimum designs from the preliminary design phase would enable companies to reduce the overal design cycle time.The focus of the present work is to study the applicability of a Multidisciplinary Design Optimization (MDO) method called Concurrent SubSpace Optimization (CSSO) for the design and optimization of large scale real-life engineering systems. This work can be divided into three parts. The first part is the introduction and development of a benchmark MDO problem that simulates the design and optimization of high temperature engine components (e.g. turbines, compressors etc.). The design problem addressed herein is a stepped beam problem that couples multiple analysis codes using NASTRAN, PATRAN (The MacNeal Schwendler Corporation 1997a,b) and Response Surface Approximations (RSA). The second part focuses on the effectiveness of the polynomial based response surface approximations for capturing the temperature in a thin walled high temperature component. Specifically, quadratic response surface approximations are being investigated for their suitability. The third and the final part provides details of the generic implementation of CSSO within iSIGHT (Engenious Software Inc. 1997) and the results of testing this implementation in application to the benchmark problem mentioned above.     

A surrogate based multistage-multilevel optimization procedure for multidisciplinary design optimization

Optimization procedure is one of the key techniques to address the computational and organizational complexities of multidisciplinary design optimization (MDO). Motivated by the idea of synthetically exploiting the advantage of multiple existing optimization procedures and meanwhile complying with the general process of satellite system design optimization in conceptual design phase, a multistage-multilevel MDO procedure is proposed in this paper by integrating multiple-discipline-feasible (MDF) and concurrent subspace optimization (CSSO), termed as MDF-CSSO. In the first stage, the approximation surrogates of high-fidelity disciplinary models are built by disciplinary specialists independently, based on which the single level optimization procedure MDF is used to quickly identify the promising region and roughly locate the optimum of the MDO problem. In the second stage, the disciplinary specialists are employed to further investigate and improve the baseline design obtained in the first stage with high-fidelity disciplinary models. CSSO is used to organize the concurrent disciplinary optimization and system coordination so as to allow disciplinary autonomy. To enhance the reliability and robustness of the design under uncertainties, the probabilistic version of MDF-CSSO (PMDF-CSSO) is developed to solve uncertainty-based optimization problems. The effectiveness of the proposed methods is verified with one MDO benchmark test and one practical satellite conceptual design optimization problem, followed by conclusion remarks and future research prospects.     

An efficient strategy for multidisciplinary reliability design and optimization based on CSSO and PMA in SORA framework

The conventional reliability-based multidisciplinary design optimization (RBMDO) integrates the reliability-based design optimization and multidisciplinary design optimization (MDO) directly, which leads to a triple-level nested optimization loop. Especially, the multidisciplinary reliability analysis in the middle layer dominates the whole efficiency of RBMDO. To tackle this problem, first of all, a sequential multidisciplinary reliability analysis (SMRA) approach that integrates the concurrent subspace optimization (CSSO) strategy and the performance measure approach is proposed, in which the multidisciplinary analysis, system sensitivity analysis and reliability analysis are decoupled and arranged sequentially, making a recursive loop. The multidisciplinary analysis and system sensitivity analysis provide the value and gradient information of limit-state function for reliability analysis respectively. As a result, a great number of repeated iterations of the whole reliability analysis are eliminated. Secondly, the CSSO has been integrated with the sequential optimization and reliability assessment (SORA) to decouple the triple-level nested RBMDO procedures into a sequence of cycles of deterministic MDO and multidisciplinary reliability analysis. Therefore, the expensive computation of the whole reliability analysis model in each iteration of RBMDO is avoided. And also, the CSSO is adopted in the deterministic MDO to deal with medium-scale and coupled multidisciplinary systems. The procedures of the proposed approaches are presented in detail. The effectiveness of the proposed strategies is demonstrated and verified with two design examples.     

无人机协助的多用户MIMO毫米波网络混合预编码方案

针对无人机（UAV）协助的毫米波网络下行链路多用户通信场景,设计一种混合预编码方案。在发射端和接收端分别使用混合预编码器和模拟合并器,并将多元联合优化问题分解为子问题进行求解。构建UAV与地面用户的三维位置模型,利用带外位置信息对波束导向向量进行优化,进而通过码本生成模拟预编码器和模拟合并器。以最小化接收数据和发送数据之间的误差为目标,利用卡尔曼滤波算法设计基带预编码器,从而减少用户之间的干扰。仿真结果表明,该方案相比模拟波束成形方案、数字预编码方案和迫零混合预编码方案可有效提升系统频谱效率和能量效率。     

面向云端FPGA的卷积神经网络加速器的设计及其调度

卷积神经网络的高计算复杂性阻碍其广泛用于实时和低功耗应用,现有软件实现方案难以满足其对运算性能与功耗的要求,传统面向FPGA的卷积神经网络构造方式具有流程复杂、周期较长和优化空间较小等问题。针对该问题,根据卷积神经网络计算模式的特点,提出一种面向云端FPGA的卷积神经网络加速器的设计及其调度机制。通过借鉴基于HLS技术、引入循环切割参数和对卷积层循环重排的设计,采用模块化方式构造网络,并进行参数拓展以进一步优化加速器处理过程;通过分析系统任务和资源的特性总结调度方案,且从控制流和数据流两方面对其进行优化设计。与其他已有工作相比,提出的设计提供了一种同时具有灵活性、低能耗、高能效和高性能的解决方案,并且探讨了加速器的高效通用调度方案。实验结果表明,该加速器可在有效提高运算整速度的同时减少功耗。     

A reliability-based multidisciplinary design optimization procedure based on combined probability and evidence theory

To address the reliability-based multidisciplinary design optimization (RBMDO) problem under mixed aleatory and epistemic uncertainties, an RBMDO procedure is proposed in this paper based on combined probability and evidence theory. The existing deterministic multistage-multilevel multidisciplinary design optimization (MDO) procedure MDF-CSSO, which combines the multiple discipline feasible (MDF) procedure and the concurrent subspace optimization (CSSO) procedure to mimic the general conceptual design process, is used as the basic framework. In the first stage, the surrogate based MDF is used to quickly identify the promising reliable regions. In the second stage, the surrogate based CSSO is used to organize the disciplinary optimization and system coordination, which allows the disciplinary specialists to investigate and optimize the design with the corresponding high-fidelity models independently and concurrently. In these two stages, the reliability-based optimization both in the system level and the disciplinary level are computationally expensive as it entails nested optimization and uncertainty analysis. To alleviate the computational burden, the sequential optimization and mixed uncertainty analysis (SOMUA) method is used to decompose the traditional double-level reliability-based optimization problem into separate deterministic optimization and mixed uncertainty analysis sub-problems, which are solved sequentially and iteratively until convergence is achieved. By integrating SOMUA into MDF-CSSO, the Mixed Uncertainty based RBMDO procedure MUMDF-CSSO is developed. The effectiveness of the proposed procedure is testified with one simple numerical example and one MDO benchmark test problem, followed by some conclusion remarks.     

Neural Network Based Feedback Linearization Control of an Unmanned Aerial Vehicle

This paper presents a flight control design for an unmanned aerial vehicle (UAV) using a nonlinear autoregressive moving average (NARMA-L2) neural network based feedback linearization and output redefinition technique.The UAV investigated is non- minimum phase.The output redefinition technique is used in such a way that the resulting system to be inverted is a minimum phase system.The NARMA-L2 neural network is trained off-line for forward dynamics of the UAV model with redefined output and is then inverted to force the real output to approximately track a command input.Simulation results show that the proposed approaches have good performance.