改进的六步相移法确定等倾角与等差线相位

光弹性应力分析是一种重要的实验方法,确定等倾线参数和等差线参数是光弹法的关键。常用的六步相移法可以自动确定等倾线和等差线参数,但在求解等倾角时,由于不能避免等倾线相位图中耦合等差线的影响,会造成误差。针对这个问题,本文提出优化方案--多次载荷-六步相移法,实现在等倾线相位图求解中消除等差线的影响。通过处理对称受压圆盘在不同载荷、不同光场下得到的光弹图像,实现光弹性等倾线和等差线的自动判读,验证了所提方法的可行性。     

光弹条纹图像的全场相移技术

本文详细分析了光弹性条纹图像的特点，利用图像处理技术，结合相移法及Ｔａｒｄｙ补偿法，实现了等倾线和等差线的全场相移，使得模型中任一点的主应力差值及第一主应力方向值可由计算机自动判读，实现了光弹性实验数据的自动处理。     

彩色光弹性干涉影像分析系统

自行开发的“彩色光弹性干涉影像分析系统”首先利用CCD成像和图像采集设备,将光弹图像以数字图像的形式存储到计算机,然后通过对存储的光弹图像进行处理,得到物体边界、等差线、等倾线等数据。最后根据这些数据,绘制出主应力迹线,并进行二维的和三维的应力分析。本文着重介绍了系统整体设计以及系统研制的难点问题(彩色光弹图像处理、主应力迹线的绘制等)。系统可以通过对彩色图像进行分解,应用目前已经比较成熟的灰度光弹图像处理技术,来完成彩色图像的处理;也可以直接应用彩色信息来确定条纹级数,进行相关处理。彩色图像能够比灰度图像提供更精确的图像信息,以满足高精度测量的要求。     

摆辗成形后的摩托车启动齿轮三维塑性应变场的光塑性分析

本文叙述了采用聚碳酸脂材料模拟低碳钢和低碳合金钢一类金属的摩托车启动齿轮(下面陶称为端面齿轮)摆辗成形过程。运用光塑性法计算分析了氐形后的光塑性模型内部三维塑性应雯分布;对采用螺旋形轨迹和玫瑰线轨迹成形后的光塑性模型内部的三维塑性应变分布进行了分析比较。该方法已为摆辗最佳工艺参数的选择提供了有用的依据。     

递阶遗传-H∞滤波混合优化的前向神经网络在捷联惯导对准中的应用

将前向神经网络用于捷联惯性导航系统(SINS)的对准问题。首先,运用递阶遗传算法(HGA)优化神经网络(NNW)的拓扑结构,并对网络其余参数进行全局粗调;然后运用H滤波算法对具有最优结构的神经网络的其余参数在线自适应精调,并对这一过程与常规算法进行了计算机仿真比较。仿真结果表明:该算法能根据实际问题自适应确定网络结构,而且精度、实时性与常规方法相仿。     

基于遗传小波神经网络的MEMS陀螺误差建模

研究了一种基于遗传算法改进的小波神经网络.该方法采用小波神经网络为主要逼近手段,并通过遗传算法优化网络的关键参数.由于小波神经网络兼容了神经网络的自学习特性和小波分析的时-频局部性,而遗传算法具有较强的全局搜索能力,二者结合形成的遗传小波神经网络因此具有较高的逼近与容错能力,从而弥补了传统的方差建模方法的不足.将该算法应用到某型MEMS陀螺仪的随机误差建模中,结果表明:遗传小波网络对原始信号的逼近精度误差在以内,较之传统的方差建模方法有了显著的提高,这一精度基本上可以满足MEMS陀螺工程化应用的要求.     

随机激励下柔性智能梁结构振动的模糊自适应控制

对柔性智能梁结构提出了一种具有结构和参数学习能力的模糊神经网络控制方法,该法摒弃了常规的以BP算法来优化整个网络参数的作法,利用遗传算法对网络全局性参数进行离线优化,利用BP学习算法对网络局部性参数进行在线调节。以柔性智能悬臂梁为例,实现了对其在随机激励下的振动控制。仿真结果表明,模糊神经网络控制算法对智能结构的振动控制具有一定的鲁棒性和自适应性。     

有限空间爆炸静态压力的温度补偿方法

为改善压阻式压力传感器的温度漂移特性,构建了基于遗传算法和小波神经网络的压力传感器温度补偿模型。针对小波神经网络收敛速度慢且易陷入局部最优解的问题,采用遗传算法对小波神经网络的连接权值、伸缩参数和平移参数进行优化。基于压力传感器的标定数据,分别采用BP神经网络、小波神经网络和遗传小波神经网络对其进行温度补偿研究,结果表明:遗传小波神经网络兼容了小波分析的时频局部特性和神经网络的自学习能力,表现出良好的收敛速度和补偿精度,经补偿后传感器的输出值更接近于标定值,其最大误差由?17.44 kPa变至0.38 kPa,最大相对误差由?14.0%变至0.38%。将该模型应用于有限空间爆炸静态压力的温度补偿中,取得了较好的实际应用效果。     

基于遗传神经网络的锚杆极限承载力预测的研究

针对BP人工神经网络具有易陷入局部极小等缺陷,提出了将遗传算法与神经网络结合,同时优化网络结构的权值与阈值的思想,建立了基于遗传算法的锚杆极限承载力预测的遗传神经网络模型。该模型以低应变动测的5个变量作为输入变量来对锚杆极限承载力进行预测,并与BP神经网络预测结果进行比较。数值算例表明,遗传神经网络在锚杆极限承载力预测中具有较高的计算效率和识别精度。     

单像素成像的数字光弹研究

数字光弹是光测力学中进行应力分析的重要方法,其核心是通过数字图像处理的方式确定等倾线和等差线参数并获得对应的相位图.传统的数字光弹使用面阵相机拍摄得到光弹性条纹图像,但是通常在特殊波段尤其是在非可见光波段中适用的面阵相机普遍比较昂贵、制造成本也相对较高.本文提出在数字光弹中引入一种具有低成本、高成像质量优势的成像方法—...     

The application of Shuffled Frog Leaping Algorithm to Wavelet Neural Networks for acoustic emission source location

When using acoustic emission to locate the friction fault source of rotating machinery, the effects of strong noise and waveform distortion make accurate locating difficult. Applying neural network for acoustic emission source location could be helpful. In the BP Wavelet Neural Network, BP is a local search algorithm, which falls into local minimum easily. The probability of successful search is low. We used Shuffled Frog Leaping Algorithm (SFLA) to optimize the parameters of the Wavelet Neural Network, and the optimized Wavelet Neural Network to locate the source. After having performed the experiments of friction acoustic emission's source location on the rotor friction test machine, the results show that the calculation of SFLA is simple and effective, and that locating is accurate with proper structure of the network and input parameters.     

Optimal design of multi-parametric nonlinear systems using a parametric continuation based Genetic Algorithm approach

In this paper, a procedure for the optimal design of multi-parametric nonlinear systems is presented which makes use of a parametric continuation strategy based on simple shooting method. Shooting method is used to determine the periodic solutions of the nonlinear system and multi-parametric continuation is then employed to trace the change in the system dynamics as the design parameters are varied. The information on the variation of system dynamics with the value of the parameter vector is then used to find out the exact parameter values for which the system attains the required response. This involves a multi-parametric optimisation procedure which is accomplished by the coupling of parameter continuation with different search algorithms. Genetic Algorithm as well as Gradient Search methods are coupled with parametric continuation to develop an optimisation scheme. Furthermore, in the coupling of continuation and Genetic Algorithm, a “norm-minimising” strategy is developed and made use of minimising the use of continuation. The optimisation procedure developed is applied to the Duffing oscillator for the minimisation of the system acceleration with nonlinear stiffness and damping coefficient as the parameters and the results are reported. It is also briefly indicated how the proposed method can be successfully used to tune nonlinear vibration absorbers.     

A genetic algorithm particle pairing technique for 3D velocity field extraction in holographic particle image velocimetry

 This research explores a novel technique, using Genetic Algorithm Particle Pairing (GAPP) to extract three-dimensional (3D) velocity fields of complex flows. It is motivated by Holographic Particle Image Velocimetry (HPIV), in which intrinsic speckle noise hinders the achievement of high particle density required for conventional correlation methods in extracting 3D velocity fields, especially in regions with large velocity gradients. The GA particle pairing method maps particles recorded at the first exposure to those at the second exposure in a 3D space, providing one velocity vector for each particle pair instead of seeking statistical averaging. Hence, particle pairing can work with sparse seeding and complex 3D velocity fields. When dealing with a large number of particles from two instants, however, the accuracy of pairing results and processing speed become major concerns. Using GA’s capability to search a large solution space parallelly, our algorithm can efficiently find the best mapping scenarios among a large number of possible particle pairing schemes. During GA iterations, different pairing schemes or solutions are evaluated based on fluid dynamics. Two types of evaluation functions are proposed, tested, and embedded into the GA procedures. Hence, our Genetic Algorithm Particle Pairing (GAPP) technique is characterized by robustness in velocity calculation, high spatial resolution, good parallelism in handling large data sets, and high processing speed on parallel architectures. It has been successfully tested on a simple HPIV measurement of a real trapped vortex flow as well as a series of numerical experiments. In this paper, we introduce the principle of GAPP, analyze its performance under different parameters, and evaluate its processing speed on different computer architectures. Received: 7 September 1997/Accepted: 3 February 1998     

卷积神经网络在装备磨损颗粒识别中的研究综述

铁谱法是用于装备故障诊断的1种重要方法,其中铁谱法的重点是铁谱图像的分析,即磨损磨粒分析. 卷积神经网络是当下最流行的深度学习算法之一,其广泛应用于图像识别领域,使得图像识别领域得到突破. 随着卷积神经网络的快速发展,磨损颗粒在智能识别方面的技术取得了重大的突破. 本文中首先简述了卷积神经网络与磨粒智能识别的发展历史,针对基于卷积神经网络的磨粒识别方法进行了从图像数据集处理到模型优化技术方面的介绍,并详细说明了这些技术在磨粒识别中的具体应用实例. 然后从现有网络和自设计网络两方面分类,整理了近年来卷积神经网络应用于磨粒智能识别的代表性文献,综述了这些工作所提出的模型结构和特点,分析并阐述了各个模型主要的识别原理,各个网络结构存在的优缺点,以及它们的数据采用情况等,并对未来磨粒智能识别的主要研究方向进行了展望. 最后肯定了卷积神经网络方法在磨粒智能识别方面的重要性,同时指出了基于此方法的磨粒识别模型的缺点,并提出了应紧跟图像识别领域的最新技术以促进磨粒智能识别水平提高等建议,对磨粒智能识别的发展具有一定的意义.      

自适应联邦滤波器及其在组合导航系统中的应用

针对现有的信息分配方法在滤波精度和自适应性方面存在不足的问题,设计了一种在模糊神经网络基础上改进的自适应联邦滤波器,利用模糊神经网络自适应地调整信息分配系数,从而有效抑制干扰对子滤波器工作状态的影响,使主滤波器能够充分、有效地利用子滤波器的信息。将这种自适应联邦滤波器应用到MINS/GPS组合导航系统设计中,并对系统进行了仿真。通过与基于常规联邦滤波器的组合系统的仿真结果比较可知,自适应联邦滤波器的组合系统在性能上有了显著的提高,且速度误差标准差控制在0.06m/s以内,位置误差标准差控制在3.3m以内,具有良好的导航精度。     

Recurrent and convolutional neural networks for deep terrain classification by autonomous robots

The future challenge for field robots is to increase the level of autonomy towards long distance (>1 km) and duration (>1h) applications. One of the key technologies is the ability to accurately estimate the properties of the traversed terrain to optimize onboard control strategies and energy efficient path-planning, ensuring safety and avoiding possible immobilization conditions that would lead to mission failure. Two main hypotheses are put forward in this research. The first hypothesis is that terrain can be effectively detected by relying exclusively on the measurement of quantities that pertain to the robot-ground interaction, i.e., on proprioceptive signals. Therefore, no visual or depth information is required. Then, artificial deep neural networks can provide an accurate and robust solution to the classification problem of different terrain types. Under these hypotheses, sensory signals are classified as time series directly by a Recurrent Neural Network or by a Convolutional Neural Network in the form of higher-level features or spectrograms resulting from additional processing. In both cases, results obtained from real experiments show comparable or better performance when contrasted with standard Support Vector Machine with the additional advantage of not requiring an a priori definition of the feature space.     

基于遗传算法的铅球投掷过程分析

运用有关力学知识建立了铅球投掷过程的数学模型，采用遗传算法对投掷过程进行优化，获得了投掷过程的各项最佳参数，并给出了最大的投掷距离．还对铅球投掷过程中的影响因素进行了灵敏度分析，确定出各影响因素的主次关系，为制定科学的运动训练计划提供了参考依据。     

Aerodynamic optimization of 3D wing based on iSIGHT

A method for combining the CFD software,Fluent,with the iSIGHT design platform is presented to optimize a three-dimensional wing to ameliorate its aerodynamics performance.In the optimization design,two kinds of genetic algorithms,the Neighborhood Cultivation Genetic Algorithm（NCGA）and the Non-dominated Sorting Genetic Algorithm（NSGAII）,are employed and the Navier-Stoke（N-S）equations are adopted to derive the aerodynamics functions of the 3D wing.The aerodynamic performance of the optimized wing has been significantly improved,which shows that the approach can be extended and employed in other cases.     

基于神经网络的界面裂纹尖端场分析

通过构造反向传播神经网络,对裂纹尖端的应力场进行模拟,进而实现对裂纹尖端应力场甬数的逼近。得到的网络具有较高的联想、记忆能力和相当的稳定性,并且可以快速、准确地得到带裂纹构件的裂纹尖端应力场,从而确定裂纹尖端的塑性区和分析裂纹的扩展。数值计算给出了LY12-CZ材料裂纹扩展方向的计算结果,与实验结果吻合较好,还给出了两相材料含界面裂纹在复合型载荷作用下的塑性区形状的变化情况,并对两相材料含界面裂纹在复合型载荷作用下裂纹的扩展方向进行了预测。