基于不等长序列相似度挖掘的数据关联算法

针对不等长序列数据的关联问题,提出基于滑动窗口的最优匹配增权法不等长序列相似度度量算法。以较短序列作为滑动窗口遍历较长序列得到一组滑动相似度,利用这组相似度形成最优权重,加权得到不等长序列的相似度,并根据相似度大小对序列数据进行关联判决,以解决截断法相似度度量仅能反映截断序列局部相似度的问题。仿真实验验证了所提出算法对不等长序列数据关联的有效性,并对序列长度和量测误差等因素对相似度度量和关联效果的影响进行了讨论。     

测量不确定度的研究与实践

不确定度是表征测量结果的可靠程度,它反映了测量的精确度。通过对不确定大小及其成因的分析,找到了影响测量精确度的原因并加以校正。介绍了如何使用不确定度评定测量结果的误差,并以平均点坐标法处理数据的不确定度作为示例,说明在对相关系数较低的线性相关量的实验数据处理时,使用平均点坐标法,不仅方便而且是十分理想的。用平均点坐标法处理数据时,其不确定度的评估往往比较复杂,对该处理过程,给出了一个粗略的范例。     

复杂网络数据流频繁项集人工智能挖掘仿真

针对传统的复杂网络数据流频繁项集人工智能挖掘方法存在数据挖掘时间较长、准确性较低等问题,提出一种基于时间戳的复杂网络数据流频繁项集人工智能挖掘方法。在训练阶段,利用贝叶斯分类算法找到所有复杂网络数据流频繁项集,并计算不同复杂网络数据流频繁项集的概率估值,在测试阶段,针对不同的测试样本构造不同的分类器,集成分类器,获取分类结果。通过分类结果,构建时间戳的滑动窗口模型,根据滑动窗口的大小对项集进行延迟处理,当项集的类型变化界限超过一定的阈值时,需要重新计算支持度,根据计算结果更新变化界限,完成复杂网络数据流频繁项集人工智能挖掘。实验结果表明,所提方法能够快速、准确地对数据流频繁项集进行人工智能挖掘。     

面向不均衡分类的隶属度加权模糊支持向量机

针对不均衡分类问题,提出了一种基于隶属度加权的模糊支持向量机模型。使用传统支持向量机对样本进行训练,并通过样本点与所得分类超平面之间的距离构造模糊隶属度,这不仅能够消除噪点和野值点的影响,而且可以在一定程度上约减样本;利用正负类的平均隶属度和样本数量求得平衡调节因子,消除数据不平衡时造成的分类超平面的偏移现象;通过实验结果验证了该算法的可行性和有效性。实验结果表明,该算法能有效提高分类精度,特别是对不平衡数据效果更加明显,在训练速度和分类性能上比传统支持向量机和模糊支持向量机有进一步的提升。     

基于可变容差关系的变精度粗糙集模型

针对已有不完备信息系统扩展粗糙集模型对噪声鲁棒性差的局限性,首先分析了调节基本知识粒大小的同时引入相对错误分类度的必要性;然后结合系统属性值的缺失定义了对象联系度权值矩阵,并以此为基础提出了基于可变容差关系的变精度粗糙集模型(VPRS-VPTR);接着讨论了模型的性质,分析了模型中相关参数(基本知识粒大小、相对错误分类度)对分类精度的影响,给出了分类精度随模型中相关参数变化的求解算法与时间复杂度分析;最后通过仿真实验与相关研究的扩展粗糙集模型进行对比。仿真结果显示,VPRS-VPTR分类精度更高,而且针对UCI数据库上的几组不完备数据集进行仿真实验的结果还表明,相同参数下各不完备数据集的测试集和训练集分类精度变化趋势相同,进而验证了模型的有效性、灵活性及所提算法的可行性。     

基于特征嵌入的去流行度偏差混合推荐算法

针对数据不均衡条件下贝叶斯个性化排序算法生成的推荐列表中存在强流行度偏差的问题,提出基于特征嵌入的去流行度偏差混合推荐算法。首先,利用卷积神经网络提取用户、物品特征确定用户偏好,并依据用户偏好对原始不均衡数据进行评分填充;其次,将卷积神经网络提取的用户偏好特征嵌入到贝叶斯个性化排序算法中进行混合推荐;最后,用评分填充数据训练混合推荐模型,得到去流行度偏差的个性化排序列表。为了验证算法的性能,在公开数据集MovieLens-100K和MovieLens-1M上进行分析与对比实验,实验结果显示流行度偏差降低了约50%~60%,精确度提高了约一倍。     

时间敏感数据流上的频繁项集挖掘算法

数据流中的数据分布随着时间动态变化,但传统基于事务的滑动窗口模型难以体现该特征,因此挖掘结果并不精确.首先提出时间敏感数据流处理中存在的问题,然后建立基于时间戳的滑动窗口模型,并转换为基于事务的可变滑动窗口进行处理,提出了频繁项集的挖掘算法FIMoTS.该算法引入了类型变化界限的概念,将项集进行动态分类,根据滑动窗口大小的变化对项集进行延迟处理,仅当项集的类型变化界限超出一定阈值的时候才进行支持度的重新计算,能够达到剪枝的目的.在4种不同密度的数据集上完成的实验结果显示,该算法能够在保证内存开销基本不变的情况下显著提高计算效率.     

基于可预测适合度的选择性模型修复

由于信息系统记录的行为不断变化,因此事件日志与给定模型之间往往存在偏差.事件日志可能产生2种不同类型的偏差,且每种偏差在偏差总数中的占比是不确定的.已有方法采用固定方式修复日志中非迭代偏差和自循环产生的迭代偏差,或在理想适合度被设定为1的前提下选择执行不同的修复方式,因而很难保证适合度与精度始终在合理范围.针对这一问题,提出一种修复方法可根据迭代可观测偏差总成本预测配置优化后的适合度,并在其满足给定阈值的情况下对所有偏差进行整体配置.当预测适合度不满足给定阈值时,进一步通过最优对齐发现事件日志与过程模型之间的变体,并根据每个变体的实际情况使用配置优化或者自循环插入的方式修复可观测偏差.仿真实验中对不同数据集进行了验证,结果表明：在始终保证适合度合理的前提下所提出方法能够最大程度地改善精度.     

双电层对纳米流体润滑与摩擦影响的MD模拟 (2.东南大学MEMS教育部重点实验室 南京 210096）

采用分子动力学方法对带电和不带电两硅板之间的水分子润滑薄膜进行模拟研究,通过加双电层的水分子薄膜润滑与未加双电层薄膜润滑摩擦属性的对比,发现摩擦系数在存在双电层的情况下比未加双电层时要小,两板相对滑动速度对摩擦系数的影响与未加双电层时相似,相对滑动速度越大,摩擦系数在一定的速度范围内平稳增大.当速度大于某个数值时,摩擦系数增大变快.两板之间水分子以及离子密度或数目分布在靠近壁面的地方较大,中间密度相对较小.     

受电弓摩擦副载流磨损特性研究

为了解决电气化铁路受电弓网中摩擦副的磨损严重问题,采用自行开发的滑动电接触实验机,基于浸铜碳滑板和纯铜接触线为实验材料,在不同条件下进行了载流磨损实验,分析了受电弓滑板的磨损率与接触压力,接触电流和滑动速度之间的关系。基于方差分析表,采用F检验方法比较接触压力、接触电流、滑动速度3个因素对磨损率的影响程度,通过数据拟合的方法,建立磨损率的数学模型,并通过实验验证了所建立模型的有效性。     

四轮驱动滑动吸盘爬壁机器人的动力学研究

研究了四轮驱动滑动吸盘式爬壁机器人在滑动导向运动方式下的动力学问题．首先分析了机器人在壁面上安全移动的运动状态和约束条件, 然后对驱动轮支撑力分布、驱动轮与壁面间的横向摩擦力以及密封圈摩擦力进行分析,进而基于牛顿---欧拉法建立了机器人的动力学方程, 并用驱动力矩安全系数来表征驱动力矩的安全程度．通过动力学仿真,分析了吸附压力、驱动轮分布、密封圈刚度等对驱动力矩的影响, 为四轮驱动滑动吸盘式爬壁机器 人结构优化和安全运动控制提供理论依据．     

Functional Redesign of Mantis 2.0, a Hybrid Leg-Wheel Robot for Surveillance and Inspection

The paper discusses the redesign of the second version of the Mantis hybrid leg-wheel mobile robot, conceived for surveillance and inspection tasks in unstructured indoor and outdoor environments. This small-scale ground mobile robot is characterized by a main body equipped with two front actuated wheels, a passive rear axle and two rotating legs. Motion on flat and even ground is purely wheeled in order to obtain high speed, high energetic efficiency and stable camera vision; only in case of obstacles or ground irregularities the front legs realize a mixed leg-wheel locomotion to increase the robot climbing ability; in particular, the outer profile of the legs, inspired by the praying mantis, is specially designed to climb square steps. The multibody simulations and the experimental tests on the first prototype have shown the effectiveness of the mixed leg-wheel locomotion not only for step climbing, but also on uneven and yielding terrains. Nevertheless, extensive experimental tests have shown that the front wheels may slip in the last phase of step climbing in case of contact with some materials. In order to overcome this problem, the leg design has been modified with the introduction of auxiliary passive wheels, which reduce friction between legs and step upper surface; these wheels are connected to the legs by one-way bearings, in order to rotate only when they are pulled by the front wheels, and remaining locked when they have to push forward the robot. The influence of the auxiliary wheels on the front wheels slippage is investigated by means of theoretical analysis and multibody simulations.     

Concerning a Technique for Increasing Stability of Climbing Robots

Legged-climbing robot is considered. Each foot of the robot has an electromagnet system for robot"s holding on a metal surface. This surface can be both vertical and inclined, including negative slope. Analytical calculation of robot stability under turn over or sliding conditions has been made. Critical slopes have been determined. One of these slopes corresponds to minimal reserve of robot stability towards sliding and another to minimal reserve of robot stability towards turning-over. As total reserve of stability of a robot is always equal to the minimal one of these reserves. Additional support elements of elastic material with high coefficient of friction, along with electromagnet, allows to increase minimal reserve of robot stability towards sliding. The use of such support elements leads to redistributing force of normal support reaction between electromagnet (which surface has low coefficient of friction) and additional support element (which surface has high coefficient of friction). It is just what leads to increasing the total friction force and as a consequence to increasing of minimal reserve of robot stability towards sliding.     

Dynamic path tracking control of a vehicle on slippery terrain

This paper deals with accuracy and reliability for the path tracking control of a four wheel mobile robot with a double-steering system when moving at high dynamics on a slippery surface. An extended kinematic model of the robot is developed considering the effects of wheel–ground skidding. This bicycle type model is augmented to form a dynamic model that considers an actuation of the four wheels. Based on the extended kinematic model, an adaptive and predictive controller for the path tracking is developed to drive the wheels front and rear steering angles. The resulting control law is combined with a stabilization algorithm of the yaw motion which modulates the actuation torque of each four wheels, on the basis of the robot dynamic model. The global control architecture is experimentally evaluated on a wet grass slippery terrain, with speeds up to 7 m/s. Experimental results demonstrate enhancement of tracking performances in terms of stability and accuracy relative to the kinematic control.     

3-D Snake Robot Motion: Nonsmooth Modeling, Simulations, and Experiments

A nonsmooth (hybrid) 3-D mathematical model of a snake robot (without wheels) is developed and experimentally validated in this paper. The model is based on the framework of nonsmooth dynamics and convex analysis that allows us to easily and systematically incorporate unilateral contact forces (i.e., between the snake robot and the ground surface) and friction forces based on Coulomb's law of dry friction. Conventional numerical solvers cannot be employed directly due to set-valued force laws and possible instantaneous velocity changes. Therefore, we show how to implement the model for numerical treatment with a numerical integrator called the time-stepping method. This method helps to avoid explicit changes between equations during simulation even though the system is hybrid. Simulation results for the serpentine motion pattern lateral undulation and sidewinding are presented. In addition, experiments are performed with the snake robot ldquoAikordquo for locomotion by lateral undulation and sidewinding, both with isotropic friction. For these cases, back-to-back comparisons between numerical results and experimental results are given.     

Design and analysis of a pneumatic high-impact force drive mechanism for in-pipe inspection robots

A pneumatic actuator is suitable and safe for in-pipe inspection robots in inflammable circumstances because of its ability to withstand explosions. However, ordinary pneumatic actuators limit driving speed because of their slow response. In this paper, we propose a novel pneumatic drive mechanism that can produce a high-impact force to move the in-pipe robot forward with sufficient speed by a catastrophic phenomenon using rapid release between a magnet and springs. We also introduce an anisotropic friction mechanism that uses a self-locking phenomenon to transmit the impact force to the pipe walls efficiently. A pin retraction mechanism that releases the self-locking condition is applied to retrieve the robot from the pipes. Optimizations of the proposed design were conducted based on a motion simulation model and verified in experiments. The experimental results obtained for maximum driving speeds in a straight pipe with different material types were approximately 90 and 50 mm/s for horizontal and vertical pipes, respectively. Stable strokes were also observed at different driving frequencies from 0.5 to 2.0 Hz.     

Dynamic Analysis Tool for Legged Robots

The paper introduces a systematic approach for dealing with legged robot mechanism analysis. First, we briefly summarize basic mathematical tools for studying the dynamics of these multi-loop and parallel mechanisms using a unified spatial formulation which is useful for computer algorithms. The dynamic behavior analysis is based on two stages. The first one deals with establishing the equations of motion of the whole mechanism including legs tip impact effects and allowing us to solve the direct and inverse dynamic problems. The second concerns the feet–ground interaction aspect which is one of the major problem in the context of dynamic simulation for walking devices. We focus on the phenomenon of contact by introducing a general model for dynamic simulation of contacts between a walking robot and ground. This model considers a force distribution and uses an analytical form for each force depending only on the known state of the robot system. Finally, some simulation results of biped robot are given. The simulation includes all phenomena that may occur during the locomotion cycle: impact, transition from impact to contact, contact during support with static friction, transition from static to sliding friction and sliding friction.     

Design principle of micromechanical probe with an electrostatic actuator for friction force microscopy

It is important to understand friction force in micro/nano mechanical devices both at high sliding speed and with high lateral resolution. Dual-axis friction force microscopes that can provide high lateral resolution and accuracy have been proposed; however, the sliding speed is limited by the probe scan speed. While a micro mechanical probe (MMP) with an electrostatic actuator can overcome this problem, details of probe design have not been established yet. This paper presents the principle of the mechanical design for an MMP with high force sensitivity and sufficient drive force. The dimensions of the double cantilever beam control the spring constants, resonant frequencies, and drive force. The use of an actuated MMP enables accurate friction force microscopy at high sliding speeds, which is required for the design of micro/nano mechanical devices.     

Backstepping sliding mode control with FWNN for strict output feedback non-smooth nonlinear dynamic system

An output feedback backstepping sliding mode control scheme was developed for precision positioning of a strict single-input and single-output (SISO) non-smooth nonlinear dynamic system that could compensate for deadzone, dynamic friction, uncertainty and estimations of immeasurable states. An adaptive fuzzy wavelet neural networks (FWNNs) technique was used to provide improved approximation ability to the system uncertainty. The adaptive laws were derived for application to estimate the deadzone and friction parameters using recursive backstepping controller design procedures. In addition, the sliding mode control method was also combined to enforce the robustness of the output feedback backstepping controller against disturbance. The Lyapunov stability theorem was used to prove stability of the proposed control system. The usefulness of the proposed control system was verified by simulations and experiments on a robot manipulator in the presence of a deadzone and friction in the actuator.     

Highly Precise Dynamic Simulation Environment for Humanoid Robots

In this paper we present a simulation environment for humanoid robots with a precise and efficient method of handling ground contact, and experiments empirically validating the simulator. Highly accurate dynamic simulation is an essential tool for research and development in humanoid robotics, and a simulator should ideally provide a transparent interface with pathways for control and sensing information identical to those of the actual robot(s) it models. We identified ground contact as the chief source of divergence from reality in work to date and have tackled this problem by developing an algorithm for resolving ground contact for humanoid robots. Our objective was to produce an algorithm that is accurate, efficient and easy to implement. The algorithm is general with respect to the complexity of the foot model; is based on empirically measurable characteristics of the foot–ground interaction, i.e., friction, which we have obtained using experiments described; provides an exact implementation of the Coulomb friction model (avoiding polyhedral approximation of the friction cone); runs in real-time; is also amenable to a straightforward accuracy–speed trade-off; and is relatively easy to implement as a constraint selection method. The simulation environment embodies generality, and we have applied it to two different humanoid robots, Hoap-2 and CB. We present experiments comparing the results of simulation with identical motions performed by real robots, and comparing the full contact resolution algorithm, the modification trading accuracy for computational speed and a penalty-based method.