SIA OpenIR  > 工业控制网络与系统研究室
基于声发射传感器的工业机器人RV减速器故障诊断问题研究
Alternative TitleResearch on Fault Diagnosis Problems of Industrial Robot RV Reducer Based AE Sensors
安海博
Department工业控制网络与系统研究室
Thesis Advisor谈金东 ; 梁炜
Keyword工业机器人 RV减速器 故障诊断 声发射 健康状态评估
Pages129页
Degree Discipline控制理论与控制工程
Degree Name博士
2020-05-29
Degree Grantor中国科学院沈阳自动化研究所
Place of Conferral沈阳
Abstract本文从机理和诊断方法两个方面,开展基于声发射信号的机器人RV减速器故障诊断问题的探索性研究,取得如下创新成果:(1)本文希望通过声发射信号在减速器内部的传播机理的分析和研究,实现对减速器磨损故障的准确监测与评估。针对RV减速器结构复杂、耦合度高等导致的机理分析困难问题,基于前人关于“声发射信号的主分量为曲轴处产生的声发射信号”的结论,以及减速器内部声发射信号的传播随减速器旋转呈现规律性变化的实验发现,首先从RV减速器的运动学机理出发,分析了声发射信号的传播路径及其长度与曲轴随齿盘旋转角度之间的关系,进而得到在减速器内部传播过程中的衰减规律及其趋势演化模型。通过RV减速器的加速退化实验,得到了四种不同故障状态下的声发射信号演化趋势,验证了本文所得声发射信号在RV减速器内部的传播及衰减机理分析的正确性。(2)RV减速器退化趋势分析。碰磨是RV减速器的主要故障来源,也是影响声发射信号传播特性的关键因素。本文在分析减速器运动学原理及声发射信号在减速器内部衰减机理的基础上,进一步对减速器内部零部件碰磨过程和故障状态进行了分析,同时分析了不同碰磨过程以及碰磨过程对声发射信号的影响。在此基础上,针对RV减速器在不同转速、不同负载及不同故障状态下的声发射信号,基于峭度算法对采集到的声发射实验数据进行分析,得到了减速器不同故障状态下的声发射信号变化趋势,验证了碰磨过程以及信号变化趋势分析的正确性。(3)基于隐马尔科夫模型的减速器故障状态估计。针对工业场景下声发射信号存在大量噪声和突发型干扰等问题,提出一种改进式的HMM状态估计模型。首先利用小波变换的高频分量时间分辨率特性和低频分量频率分辨率特性,来自适应的滤除声发射信号的噪声;进而根据RV减速器的摆线针轮运动学模型,建立HMM模型来辨识声发射信号的初始相位;最后,采用改进式的HMM模型,考虑到未来故障状态与当前故障状态的隐马尔科夫关系,将未来时间序列的观测信息引入到HMM估计模型中,通过多数投票策略来估计当前时刻的隐藏故障状态是否存在突发干扰,以此来抑制突发干扰对RV减速器状态估计的影响。实验验证了所提算法的有效性。(4)基于色度的RV减速器健康状态评估。针对RV减速器结构复杂、噪声大所导致的评估不准确且评估时间长等问题,特征差异大,本文提出了一种基于色度的RV减速器健康状态在线评估算法。首先针对RV减速器声发射信号频率范围广、信号特征差异大等特点,使用小波包分解提取不同频段的信号特征并进行标准化变换,解决了现有算法在高频段或信号幅值较低的特征频率处的特征提取难题,缩短了特征提取时间;其次提出基于特征值的色度表征方法,解决了现有算法在对RV减速器声发射信号进行色度变换时灵敏度较低的问题;在此基础上,采用SOM算法辨识减速器状态变化的程度,实现了RV减速器健康状态的评估和可视化表达。(5)基于声发射传感器的RV减速器故障诊断实验平台。本文开发由旋转实验平台、数据采集设备以及数据软件几部分构成的实验平台。该平台采用Huffman编码方法,实现了高速数据采集、数据预处理、数据快速压缩与解压缩等功能,解决了声发射信号采样率高、信号数据量大等数据采集和处理问题;平台集成了多种数据处理算法,能够支撑研究所需的实验测试和算法比对,为RV减速器故障诊断积累了大量的实验数据。
Other AbstractThe innovations are summarized as follows. (1) This work intends to perform the accurate monitoring and evaluation of reducer wear faults by analyzing the propagation mechanism of AE signals within. The mechanism analysis on RV reducer is quite difficult due to its complex and coupling structure. Based on the previous work that the principal components of AE signals are mainly from crankshaft, we investigate the internal acoustic emission signal propagation within the speed reducer in this chapter. Firstly, this research starts from the kinematics mechanism of RV reducer, followed by the analysis of acoustic emission signal propagation path and the relationship between the crankshaft and gear rotation angle, to obtain the attenuation law and trend of evolution model. Through the accelerated degradation tests, the evolution trend of AE signals in four different wear states is predicted, which verifies our RV reducer AE signal propagation and attenuation mechanism model. (2) Degradation trend analysis of acoustic emission signal of RV reducer is presented in this chapter. Collision and wear are the main source of RV reducer faults and also the key factors that determine AE signal propagation. On the basis of analyzing the propagation and attenuation of AE signals in the reducer, we further study how the acoustic emission signals between internal parts evolve over collision and wear, i.e., to analyze the rubbing process and rubbing state between RV reducer parts. Afterwards, the kinematics model of the reducer and the attenuation mechanism of acoustic emission signal in the reducer are analyzed. At the same time, the causes of AE signals generated in different rubbing processes and different working conditions are analyzed. In the experiments, the variation trend of AE signal is achieved and the rubbing process model under different working conditions with various loads and wear states are verified, which proves our proposed trend analysis model correctness. (3) Hidden Markov Model (HMM) based fault state estimation of reducer is presented in this chapter. An improved HMM state estimation model is proposed, which could cope with the issues such as the presence of noises and disturbances accompanied with collected AE signals in industrial scenarios. Firstly, the wavelet packet decomposition (WPD) is performed (that has high frequency component temporal resolution and low frequency component frequency resolution) to adaptively filter AE signal noises. Then HMM model is adopted to identify the initial phase of AE signal using the cycloidal pinwheel kinematics model of RV reducer. Finally, an improved HMM model is developed, which fully considers the observation information of the future time series, and introduces the observation sequence into HMM estimation model to predict the hidden wear state through voting strategies. In this way, the impact of noises and disturbances on the state estimation of RV reducer could be suppressed. The experimental results and analysis verify our proposed method. (4) RV reducer health evaluation method is designed based on improved chromatic algorithm in this chapter. RV reducer wear evolves over time, and the degree of wear is closely related to reducer health conditions. Thus it is necessary to evaluate the degree of RV reducer wear. Firstly, the wavelet packet decomposition is used to extract the signal features of different frequency bands followed by standardized transformation, in order to solve feature extraction failures in high-frequency band or the parts with lower intensity values. Secondly, the improved chromaticity processing method is developed to solve the problem of low sensitivity when using Gabor to perform chromaticity transformation on AE signals. Eventually, self-organizing map (SOM) is adopted to identify the degree of health evolution. In the experiments, RV reducer evaluation results and the corresponding visualized representations are presented, which verify our method effectiveness. (5) RV reducer fault detection platform using AE is introduced in this chapter. The experimental platform mainly consists of rotation mechanics, data acquisition hardware and data analyzers. The Huffman coding method is adopted to collect, preprocess, compress and decompress AE measurements with high frequency. Also, the platform is integrated with numerous data processing algorithms, which facilitates various tests and comparisons. The large volume of AE data is collected on this platform for AE fault diagnosis and health assessment.
Language中文
Contribution Rank1
Document Type学位论文
Identifierhttp://ir.sia.cn/handle/173321/27173
Collection工业控制网络与系统研究室
Affiliation中国科学院沈阳自动化研究所
Recommended Citation
GB/T 7714
安海博. 基于声发射传感器的工业机器人RV减速器故障诊断问题研究[D]. 沈阳. 中国科学院沈阳自动化研究所,2020.
Files in This Item:
File Name/Size DocType Version Access License
基于声发射传感器的工业机器人RV减速器故(7649KB)学位论文 开放获取CC BY-NC-SAApplication Full Text
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[安海博]'s Articles
Baidu academic
Similar articles in Baidu academic
[安海博]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[安海博]'s Articles
Terms of Use
No data!
Social Bookmark/Share
All comments (0)
No comment.
 

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.