A modified Hopfield neural networks model for graphs-based kinematic structure design

Graph theory can be used efficiently for both kinematic and dynamics analysis of mechanical structures. One of the most important and difficult issues in graphs theory-based structures design is graphs isomorphism discernment. The problem is vital for graph theory-based kinematic structures enumeration, which is known to be nondeterministic polynomial-complete problem. To solve the problem, a Hopfield neural networks (HNN) model is presented and some operators are improved to prevent premature convergence. By comparing with genetic algorithm, the computation times of the HNN model shows less affection when the number of nodes were enhanced. It is concluded that the algorithm presented in this paper is efficient for large-scale graphs isomorphism problem.     

An Improved Genetic Algorithm Approach on Mechanism Kinematic Structure Enumeration with Intelligent Manufacturing

During the process of mechanism kinematic structure enumeration, isomorphism identification of graphs is an important and complicated problem. The problem is known to be a NP-complete problem. In this paper, according to the mechanism kinematic chain isomorphism identification criteria, a highly efficient hybrid genetic algorithm model is proposed for isomorphism identification. The model method is coupled with genetic algorithm, optimal choice, and optimal crossover operation. It shows a quick convergence rate of the late operation and can avoid convergence to local optimum. Simulation results show that the hybrid algorithm is more rapid and effective compared with simple genetic algorithm and the improved neural network algorithm.     

The establishment of edge-based loop algebra theory of kinematic chains and its applications

Based on the edge-based array representation of loops in the topological graphs of kinematic chains, this paper first proposes three arithmetic operations of loops. Then the concept of the independent loop set as well as its determination rules is introduced, and a new structure decomposition algorithm of kinematic chains is presented. Based on the algorithm, an automatic and efficient method for rigid sub-chain detection and driving pair selection of kinematic chains is proposed. Finally, an index is proposed to assess computation complexity of kinematic analysis with respect to different driving pair selections.     

基于快速学习型神经网络的机器人运动学模型辨识及运动控制

该文提出一种快速学习型神经网络，它不仅符合生物神经网络的基本特征，而且算法简单，学习收敛速度快，有线性，非线性系统辨识精度高优异特点。因此，此类神经网络非常适合于机器人运动学模型辨识及运动控制，仿真结果表明，基于快速学习型神经网络进行机械手运动学模型辨识有运动控制是合适的。     

支持OpenCL 的GPU 加速人工神经网络训练

人工神经网络训练所包含的运算量随着网络中神经元的数量增多而加大,对于神经元较多的网络训练很耗时。提高人工神经网络训练速度的一个方法是对训练算法优化以减少计算量。由于人工神经网络训练算法包含大量的矩阵和向量运算,如果把优化的算法用运行在GPU上的OpenCL C语言实现,则训练速度相比传统基于CPU计算的实现会提高很多。从硬件的并行计算能力着手,以RPROP算法为例,对其运行在GPU上的OpenCL C语言实现作一些研究。     

航天服手臂运动学建模及其关节力学特性的测试

介绍基于随动式测量机器人的航天服关节力学特性测量原理,根据舱内航天服手臂的特性建立基于 机器人运动学的数学模型,提出基于神经网络的逆运动学算法进行航天服关节角度的计算,并采用基于泛化能力的 网络构造算法进行神经网络拓扑结构的优化设计.通过实际舱内航天服手臂的关节力学特性测试实验验证测量原 理和计算方法.     

A recurrent neural network for minimum infinity-norm kinematiccontrol of redundant manipulators with an improved problem formulationand reduced architecture complexity

This paper presents an improved neural computation where scheme for kinematic control of redundant manipulators based on infinity-norm joint velocity minimization. Compared with a previous neural network approach to minimum infinity-non kinematic control, the present approach is less complex in terms of cost of architecture. The recurrent neural network explicitly minimizes the maximum component of the joint velocity vector while tracking a desired end-effector trajectory. The end-effector velocity vector for a given task is fed into the neural network from its input and the minimum infinity-norm joint velocity vector is generated at its output instantaneously. Analytical results are given to substantiate the asymptotic stability of the recurrent neural network. The simulation results of a four-degree-of-freedom planar robot arm and a seven-degree-of-freedom industrial robot are presented to show the proposed neural network can effectively compute the minimum infinity-norm solution to redundant manipulators.     

A New Complete Hierarchical Classification Approach to Kinematic Chains

Based on structural properties and genetic isomorphism-identification approach, this paper proposes a classification scheme of kinematic structures to categorize the kinematic chains into different families, thus facilitating the optimum selection of a basic structure of a mechanism. The kinematic chain is represented by a graph at first. The genetic adaptive model for the graph isomorphism identification is developed, which includes the construction of an effective method to decrease the problem's dimensions and applying an evolutionary searching strategy. From the various invariants of the genetic adaptive model, which charaterize the specific features of a kinematic chain or a family of kinematic chains, we obtain a six-step hierarchical classification scheme. This scheme classifies together the kinematic chains having similar sub-sets of structures forming isomorphic sub-chains. An example illustrates the theory, procedure and utitlity of the hierarchical classification. The scheme reduces computing time and effort in the optimum selection of a kinematic structure from a large family of kinematic chains.     

Survey of neural network technology for automatic targetrecognition

A review is presented of ATR (automatic target recognition), and some of the highlights of neural network technology developments that have the potential for making a significant impact on ATR are presented. In particular, neural network technology developments in the areas of collective computation, learning algorithms, expert systems, and neurocomputer hardware could provide crucial tools for developing improved algorithms and computational hardware for ATR. The discussion covers previous ATR system efforts. ATR issues and needs, early vision and collective computation, learning and adaptation for ATR, feature extraction, higher vision and expert systems, and neurocomputer hardware.     

基于郭涛算法的演化神经网络

提出了一种新的演化神经网络算法GTEANN,该算法基于高效的郭涛算法,同时完成在网络结构空间和权值空间的搜索,以实现前馈神经网络的自动化设计。本方法采用的编码方案直观有效,基于该编码表示,神经网络的学习过程是一个复杂的混合整实数非线性规划问题,例如杂交操作包括网络的同构和规整处理。初步实验结果表明该方法收敛,能够达到根据训练样本自动优化设计多层前馈神经网络的目的。     

Tracking a maneuvering target using neural fuzzy network

A fast target maneuver detecting and highly accurate tracking technique using a neural fuzzy network based on Kalman filter is proposed in this paper. In the automatic target tracking system, there exists an important and difficult problem: how to detect the target maneuvers and fast response to avoid miss-tracking? The traditional maneuver detection algorithms, such as variable dimension filter (VDF) and input estimation (IE) etc., are computation intensive and difficult to implement in real time. To solve this problem, neural network algorithms have been issued recently. However, the normal neural networks such as backpropagation networks usually produce the extra problems of low convergence speed and/or large network size. Furthermore, the way to decide the network structure is heuristic. To overcome these defects and to make use of neural learning ability, a developed standard Kalman filter with a self-constructing neural fuzzy inference network (KF-SONFIN) algorithm for target tracking is presented in this paper. By generating possible target trajectories including maneuver information to train the SONFIN, the trained SONFIN can detect when the maneuver occurred, the magnitude of maneuver values and when the maneuver disappeared. Without having to change the structure of Kalman filter nor modeling the maneuvering target, this new algorithm, SONFIN, can always find itself an economic network size with a fast learning process. Simulation results show that the KF-SONFIN is superior to the traditional IE and VDF methods in estimation accuracy.     

基于神经网络的机器人操作手IKP精确求解

结合位置正解模型，利用BP网络求解了机器人逆运动学问题（IKP）．为提高求解 结果精度，采用迭代计算进行误差补偿，计算结果表明，该法迭代次数少，计算精度高且计 算速度接近机器人实时控制的要求．     

动态修正下的神经网络盲均衡算法

针对现有的神经网络算法收敛速度慢以及精确度低的问题,通过对传统的神经网络盲均衡算法以及前馈神经网络进行研究,提出一种具有自动修正效果的前馈神经网络盲均衡算法。该算法通过对算法中的代价函数以及迭代步长因子进行改进,来提高算法的收敛速度;通过对所获得的目标信号进行修正处理,来对所获取的信息进行修正。实验结果表明,该算法的实验结果与预期效果基本相符,具有可靠性强、收敛速度快的优势。     

多层前向神经网络的快速学习算法及其应用

针对目前多层前向神经网络学习算法存在的不足,提出一种多层前向神经网络的快速学习算法,它不仅符合生物神经网络的基本特征,而且算法简单,学习收敛速度快,具有线性、非线性逼近精度高等特性.以二杆机械手逆运动学建模作为应用实例,仿真结果表明该方法是有效的,其算法与收敛速度更优于BP网络.     

基于Winograd稀疏算法的卷积神经网络加速器设计与研究

随着卷积神经网络得到愈加广泛的应用,针对其复杂运算的定制硬件加速器得到越来越多的重视与研究。但是,目前定制硬件加速器多采用传统的卷积算法,并且缺乏对神经网络稀疏性的支持,从而丧失了进一步改进硬件,提升硬件性能的空间。重新设计一款卷积神经网络加速器,该加速器基于Winograd稀疏算法,该算法被证明有效降低了卷积神经网络的计算复杂性,并可以很好地适应稀疏神经网络。通过硬件实现该算法,本文的设计可以在减少硬件资源的同时,获得相当大的计算效率。实验表明,相比于传统算法,该加速器设计方案将运算速度提升了近4.15倍;从乘法器利用率的角度出发,相比现有的其他方案,该方案将利用率最多提高了近9倍。     

基于邻居信息聚合的子图同构匹配算法

图匹配在现实中被广泛运用,而子图同构匹配是其中的研究热点,具有重要的科学意义与实践价值。现有子图同构匹配算法大多基于邻居关系来构建约束条件,而忽略了节点的局部邻域信息。对此,提出了一种基于邻居信息聚合的子图同构匹配算法。首先,将图的属性和结构导入到改进的图卷积神经网络中进行特征向量的表示学习,从而得到聚合后的节点局部邻域信息;然后,根据图的标签、度等特征对匹配顺序进行优化,以提高算法的效率;最后,将得到的特征向量和优化的匹配顺序与搜索算法相结合,建立子图同构的约束满足问题（CSP）模型,并结合CSP回溯算法对模型进行求解。实验结果表明,与经典的树搜索算法和约束求解算法相比,该算法可以有效地提高子图同构的求解效率。     

人工神经网络遗传算法研究

人工神经网络具有的自主学习的适应能力、并行信息处理的能力、非线性映射能力等,使其具有十分广泛的应用。而遗传算法是一种学习生物界之中自然遗传、自然选择机制的的一种优秀的搜索类算法,具有随机性、并行性和自适应能力等,具有群体之中自动寻优学习能力。将人工神经网络与遗传算法成功结合在一起,可以快速、准确、方便地解决网络中的相关问题,是计算机网络应用中的创举。     

基于进化策略的动态递归神经网络建模与辨识

提出一种采用进化策略实现动态递归神经网络结构、权重和自反馈增益同时进化的学习算法,以及自适应进化机制,与改进ＢＰ６算法相结合,各取所长,形成集成化动态递归神经网络建模辨识算法,实际应用结果表明,所提出算法不仅明显提高了动态递是 网络模型辨识自救的收敛速度格精度,而且实现了动态递归网络的全自动优化设计。     

A CMAC neural-network-based algorithm for the kinematic control of a walking machine

In this paper a hierarchical, neural network control architecture of a walking machine is proposed. The neural network is based on the theory of the Cerebellum Model Articulation Controller (CMAC) which is a neuromuscular control system. Some preliminary studies of kinematic control and gait synthesis are presented to demonstrate the effectiveness of the CMAC neural network. After having been trained to learn the multivariable, nonlinear relationships of the leg kinematics and gaits, CMAC is utilized to perform feedforward kinematic control of a quadruped in straight-line walking and step climbing. Simulation examples are provided and discussed. This algorithm can be extended to control other highly nonlinear processes which are hierarchical in nature and cannot be modeled by mathematical equations.     

一种神经网络辨识的混合学习算法

文章提出了一种神经网络辨识的混合学习算法。采用具有递阶结构的遗传算法来获得神经网络拓扑结构和连接权值的全局次优解,之后由BP算法来进一步调整神经网络的连接权值,从而实现神经网络的自动优化设计。仿真结果表明,所得的神经网络结构简单、精度高,并具有良好的泛化能力。