用遗传算法精确计算圆度误差

提出了一种应用遗传算法计算满足最小区域法的圆度误差的新思路,并对传统的遗传算法提出了一些改进。采用实数值编码,其计算结果精确度非常高,理论上可以获得全局最优解;保留上一代种群中适应度最好的个体到下一代,可以确保解的收敛性;对基于实数值编码的繁殖算子、交叉算子、变异算子给出了具体的操作方法。仿真结果表明,用改进的遗传算法求解圆度误差,简单明了,收敛速度快,在计算机上容易实现。     

遗传算法在含连续/离散变量结构优化中的应用

传统的优化方法难于有效地处理含有连续／离散混合变量优化问题。本文探讨了如何将遗传算法应用于含连续／离散设计变量的结构优化问题。着重讨论了连续／离散混合变量的编码方法和减少适应度函数计算次数的ｍ ｉｃｒｏ ＧＡ 技术。将遗传算法应用于数学考题和十杆结构尺寸／材料混合变量优化问题。两个算例表明,遗传算法能比较有效地解决含连续／离散混合设计变量的优化问题。     

基于基因算法与博弈论的气动高升力优化

A multi-objective evolutionary optimization method (combining genetic algorithms(GAs)and game theory(GT))is presented for high lift multi-airfoil systems in aerospace engineering.Due to large dimension global op-timization problems and the increasing importance of low cost distributed parallel environments,it is a natural idea to replace a globar optimization by decentralized local sub-optimizations using GT which introduces the notion of games associated to an optimization problem.The GT/GAs combined optimization method is used for recon-struction and optimization problems by high lift multi-air-foil desing.Numerical results are favorably compared with single global GAs.The method shows teh promising robustness and efficient parallel properties of coupled GAs with different game scenarios for future advanced multi-disciplinary aerospace techmologies.     

Blockage effect on the flow around a cylinder probe in calibration

Flow around a 2-D cylinder pressure probe placed in uniform flow,free jet flow,and wind tunnel flow was analyzed with potential flow theory and simulated with numerical method.Blockage effect was investigated under several typical flow Mach numbers.The result from numerical simulation shows a similar trend to the one from potential flow method while varies in quantity.Wind tunnel walls accelerate the flow near the probe and thus produce a blockage effect;Boundary of free jet flow,however,decelerates the flow and thus produces a "negative" blockage effect.A maximum incoming Mach number exists when the probe is calibrated in wind tunnel in high subsonic condition due to choking caused by shocks and shock induced separation.The critical Mach number varies with blockage ratio,which makes high Mach number impossible to achieve in large blockage ratio condition.The blockage effect itself is unavoidable for calibration or measurement although a sufficiently small blockage ratio brings minor effect.Correction can be implemented based on the numerical simulation result presented in this paper and further works.      

叶轮流道5轴高速铣加工刀轨计算

为适应叶轮流道的高速加工,提出了一种新的端面铣削5轴加工刀轨生成算法。该算法首先提取原加工刀轨面的上下边界边,接着采用5次NURBS曲线拟合此上下边界边,然后顺序地连接上下NURBS曲线的型值点,最终生成了全程C2连续的刀位曲线和空间平滑变化的刀轴矢量,应用于Superm an CAMⅡ原型系统中。经测试,该算法的运行速度已接近于UG,所生成的加工刀轨加工材料为铝合金的叶轮流道,刀具能以转速5 000r/m in以及进给速度600 mm/m in平稳地进行精加工,满足了高速走刀的要求。     

Geometric Optimization Design System Incorporating Hybrid GRECO-WM Scheme and Genetic Algorithm

This article seeks to outline an integrated and practical geometric optimization design system (GODS) incorporating hybrid graphical electromagnetic computing-wedge modeling (GRECO-WM) scheme and the genetic algorithm (GA) for calculating the radar cross section (RCS) and optimizing the geometric parameters of a large and complex target respectively. A new wedge modeling (WM) scheme is presented for calculating the high-frequency RCS of wedge with only one visible facet based on the method of equivalent currents (MEC). The applications of GODS to 2D cross-section and 3D surface are respectively implemented by choosing an average of monostatic RCS values corresponding to a series of incident angles over a frequency band as the optimum objective function. And the results demonstrate that the RCS can be effectively and conveniently reduced by the GODS presented in this article.     

基于快速退火MRF的改进SAR图像分割方法

基于Markov随机场(MRF,Markov Random Field)的SAR图像分割方法利用了SAR图像的灰度和结构信息,能在分割过程中有效抑制斑点噪声,获得较高的分割精度.但这类方法的缺点是模拟退火的计算量很大.针对该问题,提出了一种基于快速退火MRF的SAR图像分割处理方法.该方法根据SAR图像Gibbs分布的特性,在求取全局最优解时,首先寻找邻域系统中占有支配地位的某种标记,若存在占支配地位的标记,用此标记更新状态;反之,则沿用传统模拟退火的方法随机更新状态.由于该方法引入基于Gibbs分布的先验判决进行系统状态更新,因此能够快速求得全局最优解.最后对真实SAR图像进行处理,处理结果验证了算法的有效性.     

保证速率的AFDX交换机实时调度算法

针对AFDX(Avionics Full Duplex Switched Ethernet)网络关键技术——交换机实时调度算法,提出一种保证速率的优先级实时调度算法(PRTRG,Priority Real Time sched-uling algorithm based on Rate-Guaranteed),并运用网络演算理论分析了此调度算法的实时性.在AFDX网络典型配置下,与(FIFO,First In First Out)结果进行分析对比,证明PRTRG算法有效地减少了高优先级数据的端到端延迟上界,同时保证了低优先级数据端到端延迟的确定性,并且阻止了数据流拥塞的扩散,在交换机内部实现了流量隔离.     

一种无人机视觉导航方法及其滤波算法改进

设计了一种无人机视觉/惯性组合导航系统,将无人机和地标点的运动模型作为状态方程,视觉信息作为观测量构建了与之对应的滤波模型.在滤波处理上,采用了复杂加性噪声模型对系统噪声进行建模处理;将小波分析引入到UKF(Unscented Kalman Filter)滤波中得到小波-UKF滤波算法,以此克服视觉观测噪声对滤波的影响;采用最大后验概率准则(MAP,Maximum A Posterior)自适应估计观测噪声协方差阵,并将其反馈到滤波过程中克服了小波处理后观测噪声方差阵不易确定的不足.仿真结果证明:对滤波算法的改进可以有效地提高滤波估计的精度.     

新的适于计算机辅助工程的算法集成模型

为解决计算机辅助工程中"以算法为核心"类软件系统在算法集成方面存在的重要问题,以火箭发动机设计这个较具代表性的特例为切入点,深入分析工程实践中算法的运用特点,并指出现有软件系统的不足.提出2种算法模式,基于这2种算法模式建立一种新的算法集成模型--轮循与层次化算法集成模型,其具有松耦合、层次化、灵活等特点,基于该模型可建立一个跨领域、通用、高效的算法集成软件环境.