基于DSP和分层时序记忆的齿轮箱故障诊断系统

发布时间：2018-04-06 06:23

本文选题：齿轮箱　切入点：DSP　出处：《中北大学》2011年硕士论文

【摘要】：齿轮箱是机械设备中广泛使用的关键部件,对齿轮箱进行实时监测和故障诊断对于工业生产具有重大的经济和安全意义。本文以齿轮箱为研究对象,使用DSP处理器开发了一种在线监测诊断系统,并将分层时序记忆算法应用于齿轮箱常见故障的诊断识别中,旨在检测该方法在齿轮箱故障诊断上的使用效果。嵌入式故障诊断系统基于嵌入式硬件平台,将信号采集、数据处理与特征提取、故障分类识别集成于一体在嵌入式硬件平台上执行,可以实现自动故障诊断和多种通信方式。本系统以TI公司的DSP—TMS320F2812为核心处理器,并集成了采集和通信模块电路。该系统可以实现模拟量采集、数字量输入输出和转速测量功能,同时也具有以太网通信、GSM网络通信和CAN总线通信功能。为了实现在线数据处理,系统嵌入了基于DSP的FIR数字滤波、FFT功率谱、细化谱分析、希尔伯特包络和小波分析处理算法。最后,故障诊断的特征提取函数也被嵌入到了DSP上以实现齿轮箱故障特征量自动提取。本文介绍了分层时序记忆(HTM)算法的原理和结构,并以齿轮箱故障诊断为实验基础进行了算法测试。实验首先将齿轮箱多个测点所提取出来的特征量进行了融合,并转化为位图格式以满足HTM的输入要求,然后设计了一个HTM区域来进行各工况下的输入位图模式学习。当HTM区域能够对各工况的输入位图产生稳定的稀疏分布表征后,就计算区域的条件概率矩阵来实现故障诊断工作。系统使用VC编写了分层时序记忆的算法程序,结合前端DSP诊断模块成为一个完整的故障诊断系统。实验结果表明本系统能够准确的诊断出故障,而且采用分层时序记忆算法使系统具备在线学习、多传感器融合和实时预测的优点。
[Abstract]:Gearbox is a key component widely used in mechanical equipment. It is of great economic and safety significance for industrial production to monitor and diagnose gearbox in real time.In this paper, a kind of on-line monitoring and diagnosis system is developed by using DSP processor, and the hierarchical sequential memory algorithm is applied to the diagnosis and identification of common faults of the gearbox.The purpose of this paper is to detect the application effect of this method in gearbox fault diagnosis.Embedded fault diagnosis system is based on embedded hardware platform, which integrates signal acquisition, data processing, feature extraction, fault classification and identification on embedded hardware platform. It can realize automatic fault diagnosis and various communication modes.This system takes TI company's DSP-TMS320F2812 as the core processor, and integrates the collection and communication module circuit.The system can realize the functions of analog data acquisition, digital input and output, speed measurement, Ethernet communication, GSM network communication and CAN bus communication.In order to realize on-line data processing, the system embed FIR digital filter power spectrum, thinning spectrum analysis, Hilbert envelope and wavelet analysis algorithm based on DSP.Finally, the feature extraction function of fault diagnosis is embedded into DSP to automatically extract the gearbox fault feature.This paper introduces the principle and structure of hierarchical temporal memory (HTM) algorithm, and tests the algorithm based on gearbox fault diagnosis.In the experiment, the features extracted from several measuring points of the gearbox are fused and converted into bitmap format to meet the input requirements of HTM, and then a HTM region is designed to study the input bitmap mode under various operating conditions.When the HTM region can generate a stable sparse distribution representation of the input bitmap of each condition, the conditional probability matrix of the region is calculated to realize the fault diagnosis.The system uses VC to write the algorithm program of hierarchical sequential memory, combined with front-end DSP diagnosis module to become a complete fault diagnosis system.Experimental results show that the system can accurately diagnose the fault, and the hierarchical sequential memory algorithm makes the system have the advantages of on-line learning, multi-sensor fusion and real-time prediction.
【学位授予单位】：中北大学
【学位级别】：硕士
【学位授予年份】：2011
【分类号】：TH165.3

【引证文献】