【作者】 陈金明；

【导师】 滕国库；

【作者基本信息】 大连海事大学 ， 计算机应用技术， 2008， 硕士

【摘要】 本文主要介绍了球磨机料位测控系统的设计方案,该系统主要完成对球磨机振声信号的采集、滤波、希尔伯特变换、下采样、包络信号提取等处理,并找出其能够表征料位信息的特征信号,建立特征信号与料位之间的映射关系。球磨机是一种十分重要的制粉设备,广泛应用于火力发电厂等工业场所,球磨机内的料位信息是关系到其运行状态和工作效率的重要参数,因此准确测量料位信息是非常重要的。本文的研究意义也正在于此。本系统是以ARM9系列的开发平台作为主要硬件平台,再外接电容传声器和触摸屏等设备来实现整个硬件系统地搭建。系统首先对振声信号进行采集,然后通过信号放大、数字滤波后进入微控制器S3C2410,经过S3C2410的一系列处理后得出振声信号中与料位信息相关联的特征信号,并以此信号为基础将料位信息显示出来。在整个软件系统搭建过程中,主要介绍了Linux嵌入式操作系统以及QT图形系统的移植方法。在信号处理过程中,主要应用了希尔伯特变换、小波理论、BP神经网络和下采样定理等几种算法。系统首先通过希尔伯特变换提取球磨机振声信号的包络,然后对包络信号进行下采样处理,对处理后的信号应用小波变换的多尺度重构技术进行滤波从而提取特征向量,最后用BP神经网络在特征向量与球磨机料位之间建立映射关系,从而实现对球磨机的料位检测。更多还原

【Abstract】 The design and realization of the embedded system to measure and control the material level of ball mill are introduced in this article. Its main work is that pick up the ball mill’s acoustical sound and deal with it by filtered and Hilbert transformed, and then find the characteristic vector and built the relationship between the characteristic and the material location.Ball mill is an important equipment be used to coal pulverization in the coal-fired power plant. The material location is a very important parameter and it influences the ball mill’s working efficiency. So the measurement of the material location is very important. It is the meaning of this article.The arm9 embedded development equipment is the main hardware in the embedded system, it is connects with the microphone and the touch screen to built a full system. First the system picked up the acoustical sound, dual with it by enlarge, digital filter and then put it in S3C2410, and dual with by S3C2410,we can find the characteristic vector, and use it to display the material location When built the software system, the Linux embedded OS and QT graph system transplant method are introduced. When duel with the single, the Hilbert transform, Wavelet transform theory, BP neural network theory and down sampling theory are introduced. First, system picks up the envelope of the acoustical sound. Second the envelope is processed by down sampling. Finally we using wavelet transform extract characteristic vector and using BP neural network we make a mapping between the characteristic and the material location actualize the material location recognition.更多还原