【作者】 陈松实；

【导师】 宋蜇存； 王春雨；

【作者基本信息】 东北林业大学 ， 控制理论与控制工程， 2002， 硕士

【摘要】 本文旨在设计一种基于木工机床等复杂控制对象监控器。针对某种木工机床，应用小波理论及逆向建模方法，对信号的去噪及故障诊断进行计算机仿真，并提取有用的状态特征量，现场进行实验分析，从而在硬软件设计及系统实现上进行了深入浅出的研究。 在木工机床的故障检测与诊断领域中对于平稳随机信号的处理，采用傅立叶变换可以具有足够的精度，但傅立叶变换难以全面反映非平稳过程的时变特性信号。为解决这一问题，我们引进小波变换以其强有力的时频分析能力解决了这一难题。 我们首先使用MathWorks公司的MATLAB工具软件，利用其强大且方便的数据处理能力及小波工具箱进行了计算机仿真。从而在理论上证明了：1．小波理论对平稳信号及非平稳信号的消噪处理比传统的傅立叶分析更优越。2．应用小波理论能将时域和频域结合起来描述观察信号的时频联合特征，构成信号的时频谱。在信号处理过程中，尤其是对非平稳信号的处理中的任意时刻附近的瞬变信号的分析，都具有很强的分辨能力。然后，我们在现场对砂光机进行实验，利用振动及红外线温度传感器现场实时采集信号，通过A/D采样卡送入计算机中，采用逆向建模的新方法来消除传感器零漂及A/D转换的非线性，再以数据文件形式保存，然后在MATLAB中调用数据文件，应用小波函数及工具箱对其进行处理分析，经过大量的实验及数据分析发现在不同的故障状态下，特征量有明显变化。从而对现场故障能做出准确判断。更多还原

【Abstract】 This is paper aims at designing a kind of monitor controller for complicated control objects on woodworking machine tool. We apply wavelet theory and inverse modeling to some woodworking machine tool and put up computer simulation for wiping off yawp and fault diagnosis. Then we distill some useful character for analyzing on the spot, thereby the designing of software and hardware and the relevant system implement are also researched deeply.Applying FTP transformation may acquire adequate precision for dealing with steady variable signal in fault diagnosis of woodworking machine tool field. But FTP transformation can’t completely reflect time change characteristic of signal in unsteady process. For solving this problem we adopt wavelet transformation to resolve this problem because of its strong time-frequency analysis characteristic.First we use MATLAB tool software from Math Works company to put up computer simulation because of its strong convenient data handling capability and wavelet toolbox. Thereby on theory prove: 1. Wavelet theory more excellent in wiping off yawp of steady signal and unsteady signal than conventional FTP theory. 2. We can unite time field with frequency to describe time-frequency characteristic of signal and compose time-frequency chart of signal. In disposal of signal process, especially in disposal of unsteady signal process its ability in analyzing the transient signal of neighboring any time is very strong. And then we put up experiments for grinding machinery on the spot. First we collect signals by shake and infrared temperature sensor, transfer them to computer by A/D instrument, dispel zero excursion of sensor and in-linearity of A/D transformation, save them in the form of data file. Then transfer them to MATLAB, analyze and dispose by wavelet function and toolbox. By lot of experiments and data analysis we find distinct change of the character in different fault states. Consequently distinguish accurately the different fault states on the spot. 7更多还原