基于物联网的分拣机器人故障检测系统设计

目前研究的分拣机器人故障检测系统检测准确性较低,导致检测结果误差较大、实时性较差;为此,基于物联网设计一种新的分拣机器人故障检测系统;选用滑轮式机器人载体设定分拣机器人,硬件部分采用Zigbee压力传感器采集机器人故障信息,利用XBEE模块负责数据传输,协调分拣中控机接收各个传感器采集的信息,通过STMP3550芯片实现控制器设计;通过信息标定、信息采集、特征提取、故障识别实现软件工作流程,应用非极大值最大类间方差法来筛选出最优的高低阈值解,得到连续但含有假边缘的故障信息图像边缘;将提取到的图像特征向量映射到类型空间之中,确定故障原因,完成故障识别;实验结果表明,所设计分拣机器人故障检测系统在6次检验中都准确地检测出故障原因,故障检测耗时平均值为3.27 min,能够有效提高检测准确性,加强检测结果的实时性.     

The study of a map realization system (cancellation of ambient light and swaying motion of a robot)

For better terrain adaptability, a walking robot should be equipped with a visual sensor. This paper defines the visual sensor for a walking robot as a map realization system, abbreviated MARS, and investigates the way to realize a practical 3-D range-finder which forms the major part of a MARS. First, the improvement in the range-finder's performance to separate and extract a projected laser slit ray in ambient light is investigated. A video signal processing procedure called the DDD (dual signal extraction with delay and difference) method, which extracts a pulse signal by making use of the dual pattern of positive and negative pulses in a differenced video signal, is proposed. The specific method of DDD, which extracts only the maximum pulse, is shown to be more effective and is called MaxDDD. A range-finder with a signal processing system having an optical interference filter, a MaxDDD video signal processor, and noise reduction software based on continuity of the slit ray image between scanning lines is constructed and its high S/N ratio is shown. Second, structural and algorithmic considerations are made to realize real-time compensation of the swinging motion of a walking robot and to generate a terrain map while walking. As a result, it is shown that the position measurement is executed within 240 us per point to generate a map in real time. The experimental MARS range-finder weighs 1.8 kg and is compact. This paper shows the feasibility of producing a practical visual system for walking robots.     

基于阶次跟踪和HHT瞬时相位法的轴承故障诊断

提出了一种基于阶次跟踪和HHT(Hilbert-Huang transform)瞬时相位法的轴承故障诊断新方法。对齿轮箱启动时的振动信号进行角域重采样,并对角域里的信号进行经验模态分析(EMD)分解得到多个固态模态函数(IMF)分量及其各自的瞬时相位谱,最后对包含故障信息的瞬时相位分量进行了快速傅里叶变换(FFT)。结果表明,该方法能对轴承的故障部位和类型进行准确诊断。     

基于小波旁瓣相消器的轴承故障特征提取

为解决强背景噪声下声信号提取的轴承故障特征不显著问题,提出一种基于小波旁瓣相消器的故障特征提取方法。该方法利用小波滤波器组将含噪故障轴承声信号变换到小波域,进行小波域阵列广义旁瓣相消自适应波束形成,再通过小波滤波器组重构增强后的故障轴承信号,最后对重构增强后的信号进行包络解调并提取故障特征频率进行故障诊断。实验结果表明,该方法能够在强背景噪声下有效提取滚动轴承故障特征,并且相较于传统的延时求和波束形成器具有更好的降噪和故障特征增强效果。     

轴承故障音频诊断系统设计与开发

轴承音频信号包含了其运行状态的许多重要信息,通过这些信息的恰当分析、处理就能对轴承故障进行有效诊断.与振动信号相比,音频信号的采集是非接触式的,具有使用方便、成本低廉等优势.因此,将基于音频信号的隐马尔可夫模型(Hidden Markov Model,HMM)建模方法引入到轴承故障诊断研究,以PC机为平台,在Visual C 7.0环境下设计开发了一套功能完善、操作方便、界面友好的轴承音频故障诊断系统.主要介绍了诊断系统的体系结构,HMM方法建模的步骤和故障诊断过程中的具体应用操作.多次实验结果表明,效果良好.     

Fault features extraction for bearing prognostics

Over the past years, investigation on condition-based maintenance (CBM) technique on bearing has been conducted. Bearing diagnostics and prognostics are the important aspects in CBM. A key to the success of using vibration data for bearing fault diagnostics and bearing lifecycle prognostics is a quantified relationship between bearing damage and bearing fault features. To establish such a quantitative relationship, effective signal processing techniques to extract bearing fault features from vibration signals are needed. This paper describes a newly developed fault feature extraction method for bearing prognostics. The effectiveness of the method is demonstrated with two real bearing run-to-failure test datasets: one collected under normal operating conditions and another one under abnormal operating conditions. Experimental results show that the bearing fault features extracted using both traditional vibration analysis methods and the proposed method give clear bearing heath degradation trend for the dataset collected under normal operating conditions. However, for the data collected under abnormal operating conditions, bearing fault features obtained using traditional vibration analysis methods fail to show the bearing health degradation trend while the fault features extracted using the proposed method give consistent bearing degradation trends.     

自适应流形学习在故障诊断中的应用

针对人工干预的旋转轴承故障类型及损坏程度诊断问题,提出了一种基于自适应流形学习的故障诊断新方法。该算法借助集合经验模态分解和双谱分析提取振动信号的故障特征,用纹理分析法构建故障信息的纹理特征矩阵,通过自适应流形学习的方法对高维纹理特征矩阵进行降维。整个过程能够很好地去除噪声,同时自适应选择参数,具有很好的聚类性能和复杂信号处理能力。实验结果表明该方法能够很好地区分不同的故障类型,同时在区分内圈故障、外圈故障、滚动元素故障退化程度方面也有着较好的性能。     

轴承早期故障特征提取方法研究

针对滚动轴承早期故障信号被背景噪声淹没、故障特征不明显的问题,提出一种基于小波包分解和互补集合经验模态分解(CEEMD)的轴承早期故障信号特征提取方法.利用Matlab软件对采集到的轴承振动信号进行快速谱峭度分析,根据峭度最大化原则确定带通滤波器的中心频率和带宽,设计带通滤波器;对经过带通滤波器滤波后的信号进行小波包分解和CEEMD分解,根据峭度、相关系数筛选出有效本征模态函数(IMF)分量;利用IMF分量重构小波包信号,对重构小波包信号进行包络谱分析,提取轴承早期故障信号特征频率.该方法通过谱峭度分析降低背景噪声干扰,通过小波包分解增强故障冲击信号,并将CEEMD与小波包分解相结合,解决经典EMD分解存在的模态混叠、无效分量问题.仿真结果表明,相较于传统包络解调算法,重构后信号的背景噪声得到抑制,故障特征分量突出,验证了所提方法的可行性和有效性.     

基于SVD-AR模型与VPMCD的轴承故障诊断方法

针对强噪声背景下振动信号故障特征难以提取的问题,提出了基于奇异值分解的自回归(SVD-AR)模型,用于提取振动信号的特征,并与变量预测模型模式识别(VPMCD)方法相结合应用于轴承故障诊断.对轴承振动信号进行SVD;然后,利用奇异值差分谱对分量信号进行筛选,对能够反映故障信息的分量信号建立AR模型,提取轴承振动信号的特征信息;采用VPMCD对滚动轴承运行状态进行识别.实验证明了方法的合理性和有效性.     

Adjustment of Home Posture of Biped Humanoid Robot Using Sensory Feedback Control

Humanoid robot dynamic walking is seriously affected by the initial home posture (walking ready posture). If the initial home posture is not accurate, the robot may fall down during walking despite using robust walking control algorithm. Moreover, the initial home posture of a real physical robot is slightly different at every setting because the zero position of the joint is not exactly the same. Therefore, an accurate and consistent initial home posture is essential when we compare and analyze walking control algorithms. In order to find a zero position, an incremental encoder with a limit switch or an absolute encoder such as a potentiometer can generally be used. However, the initial calibration of this method for a multi-axis humanoid robot that enables the desired initial sensor signal is difficult and time-consuming. Furthermore, it has the disadvantage that additional limit switches or absolute encoders can interfere with the design objective of compactness. Therefore, this paper describes a novel adjustment method of the home posture for a biped humanoid robot utilizing incremental encoders, an inertial sensor and force torque sensors. Four kinds of controllers are proposed for the adjustment of the home posture and adjusted offsets are measured when the outputs of the controllers have converged. Experimental results from KHR-2 show the effectiveness of the proposed adjustment algorithm.     

水下目标定向及遗传算法仿真

传统的水声定位需要一个庞大的声压水听器阵,技术复杂,耗费可观。矢量传感器是一种新型的水声传感器,仅用单个的矢量传感器就可以进行水下目标的方位估计。该文分析了矢量传感器声压梯度法和互谱声强法的定向原理,在互谱声强法仿真的基础上,将遗传算法引入水下目标定向中。通过构建适应度函数,根据信号适应度函数进行遗传选择,舍弃不稳定信号,对有用信号进行杂交,从而获得较高的定向精度。仿真结果表明,该算法可以有效地进行目标方位估计,进一步提高定向精度。     

Using multi-sensor data fusion for vibration fault diagnosis of rolling element bearings by accelerometer and load cell

This paper presents a new method for bearing fault diagnosis using the fusion of two primary sensors: an accelerometer and a load cell. A novel condition-based monitoring (CBM) system consisting of six modules: sensing, signal processing, feature extraction, classification, high-level fusion and decision making module has been proposed. To obtain acceleration and load signals, a work bench has been used. In the next stage, signal indices for each signal in both time and frequency domains have been calculated. After calculation of signal indices, principal component analysis is employed for redundancy reduction. Two principal features have been extracted from load and acceleration indices. In the fourth module, K-Nearest Neighbor (KNN) classifier has been used in order to identify the condition of the ball bearing based on vibration signal and load signal. In the fifth module, a high-level sensor fusion is used to derive information that would not be available from single sensor. Based on situation assessment carried out during the training process of classifier, a relationship between bearing condition and sensor performance has been found. Finally, a logical program has been used to decide about the condition of the ball bearing. The test results demonstrate that the load cell is powerful to detect the healthy ball bearings from the defected ones, and the accelerometer is useful to detect the location of fault. Experimental results show the effectiveness of this method.     

图像识别的机器手抓握滑移检测系统开发

为实现机器手抓握物体时不发生脱落,首先应检测其与被抓握物体接触面上的滑移信号.提出一种基于图像识别的机器手抓握滑移检测方法,采用中心区域匹配思想的归一化互相关算法(NCC)匹配由视觉传感器实时采集到的被抓握物体表面图像,得到被抓握物体在采集图像期间的滑移情况.实验结果表明:此系统可以准确检测被抓握物体是否发生滑移及滑移的方向和大小,具有高准确度、高灵敏度等优点.     

基于AR-MED和TEO的滚动轴承早期故障诊断方法

针对强背景噪声下滚动轴承早期故障信号信噪比低、特征提取难度大的问题,提出一种将自回归-最小熵解卷积(autoregressive-minimum entropy deconvolution,AR-MED)与Teager能量算子(teager energy operator,TEO)相结合的滚动轴承故障诊断方法.为了达到...     

结合马田系统-SVM的滚动轴承故障模式分类研究

为了有效地确定滚动轴承的故障类型和受损程度,提出了结合马田系统和SVM的滚动轴承故障模式分类方法。利用EEMD方法对原始振动信号进行分解,得到一系列IMF。经过故障敏感IMF选取方法筛选IMF后计算其时域和频域特征参数以及原始信号的能量熵参数,构造初始的多维特征空间。运用马田系统中的正交表和信噪比进行特征降维,得到精简特征空间。接下来使用偏二叉树方法构建支持向量机多分类模型。通过实验数据进行模型验证,结果表明该方法可以实现滚动轴承故障模式分类。     

基于声发射参数分析的滑动轴承故障诊断方法研究

本文基于声发射产生机理,对基于声发射参数分析法进行滑动轴承故障诊断方法进行理论和实验研究。首先,通过汽轮机发电机组模拟转子实验台模拟了滑动轴承三种润滑状态,通过实验台以及设计的实验方案,利用声发射采集设备对不同润滑状态的声发射信号进行采集。其次,针对采集到的不同润滑状态声发射信号,对其能量均值以及功率谱熵均值进行计算,提出了基于声发射能量均值和功率谱熵均值的散度指标的滑动轴承润滑状态诊断方法,并利用这种方法对模拟信号进行诊断,同时将其与单一能量参数分析法进行对比,发现能量参数分析法不能很好的反映出滑动轴承的三种润滑状态,而文中所提的采用多参数结合的指标诊断方法具有更好的信号适应性以及更高的区分度。     

基于坐标变换的径向主动磁悬浮轴承容错控制

针对安装3个位移传感器的六极径向主动磁轴承,研究了基于坐标变换的位移传感器容错控制算法;在电流分配矩阵重构法的基础上,提出了基于坐标变换的执行器容错控制算法;并将这两种算法结合,提出了基于坐标变换的位移传感器和执行器容错控制算法;最后,在一个两自由度径向主动磁轴承实验台上对该容错控制算法进行了实验研究.实验结果表明,采用新容错控制算法可以最多在1个传感器故障和3个励磁线圈同时故障的情况下实现转子的稳定悬浮.     

足式机器人腿部倾角传感器信号处理研究

足式机器人在自主行走时,一般通过倾角传感器来测量腿部转动角度计算足端位置,然而目前足式机器人腿部倾角传感器测量时易受噪声干扰、温度等因素的影响,导致测量精度低,足端位置估计不准确.针对以上问题,提出新的倾角传感器信号处理方法,首先利用卡尔曼滤波方法对倾角传感器输出信号进行滤波预处理,然后把滤波信号和倾角传感器输出温度值作为建立的双输入单输出RBF神经网络模型的输入变量,采用蚁群聚类算法的并行寻优特征和自适应调整挥发系数方法来确定RBF神经网络基函数位置.实验结果表明,提出的算法能很好地滤除倾角传感器信号中的噪声,实现了倾角信号的温度补偿,测量误差能够控制在0.75％以内,具有实际运用价值.     

基于LMD和GNN-Adaboost的滚动轴承故障严重程度识别

提出一种基于局部均值分解（Local Mean Decomposition,LMD）和遗传神经网络自适应增强（Genetic Neural Network Adaptive Boosting,GNN-Adaboost）的滚动轴承损伤程度识别方法。通过LMD方法将轴承振动信号分解为若干个瞬时频率有物理意义的乘积函数（Production Function,PF）,对能反映信号主要特征的PF提取能量矩,结合原始振动信号的时域特征参数（方差、偏度、峭度）,组成故障严重程度识别特征参数矩阵。将基于LMD方法的特征参数矩阵作为GNN-Adaboost方法的输入向量,对不同载荷与转速工况下的轴承进行故障严重程度识别。结果表明,基于LMD和GNN-Adaboost的方法能够有效提高轴承故障严重程度识别准确率,对滚动轴承等关键旋转部件的故障识别与定位具有重要意义。     

基于自动搜峰和shannon熵的车辆轴承多普勒畸变故障声信号校正研究

针对传声器采集的运动声源信号存在多普勒畸变问题,提出一种基于自动搜峰和shannon熵的滚动轴承多普勒畸变故障声信号校正方法。首先对所采集的声音信号进行短时傅里叶(STFT)时频分析;然后利用自动搜峰方法进行瞬时频率估计,设置shannon熵来提高瞬时频率估计精度,并得到拟合的瞬时频率曲线,进而得到信号重采样时间点;最后对原信号进行时域重采样,从而使畸变信号得以矫正。通过仿真和动态滚动轴承内外圈故障声信号的实验验证了此种方法的可行性。