【作者】 孟庆宇；

【导师】 贾明兴；

【作者基本信息】 东北大学 ， 控制理论与控制工程， 2009， 硕士

【摘要】 随着高温工业技术的进步和发展,对于其辅助耐火材料质量的要求也越来越高。尤其是在钢铁冶金行业大量应用的烧结耐火砖,其质量和寿命直接关系到钢铁冶金行业的安全生产和经济效益。耐火砖生产工艺已经比较成熟,而耐火砖的内部缺陷检测技术水平却一直没有太大的发展,至今为止,绝大多数耐火砖生产厂家的检测方法比较落后。本文采用敲击法来检测耐火砖内部质量缺陷。通过耐火砖的敲击声音自动判断耐火砖内部是否存在缺陷,并进一步区分缺陷类别。为了检测耐火砖内部缺陷,需要从敲击声音信号中获得有用的数据信息。首先设计了FIR滤波器,滤掉信号中的现场噪声,然后采用短时能量和短时平均过零率双门限比较法将真正的敲击信号提取出来。为了从信号中得到耐火砖内部特征的信息,采用Welch方法估计其功率谱。将功率谱峰值点的频率和幅值信息提取出来,作为后期处理的特征数据。上述处理方法具有运算速度快,存储要求低的特点。以特征数据作为进一步分析的基础数据,本文提出基于Fisher判别式分析的耐火砖内部缺陷检测方法,通过分析发现Fisher判别式分析方法在耐火砖内部缺陷检测应用存在不足,也就是正常类主导了特征值的分解,使缺陷类之间区分不明显。针对上述情况提出采用PCA-FDA混合模型方法对耐火砖内部缺陷进行检测,利用现场采集的数据进行仿真研究,得到较好的缺陷检测和诊断效果,说明该方法可以应用到实际检测系统中。数字信号处理器(DSP)有强大信号处理能力,完全可以胜任在线检测任务。本文设计了一个基于TMS230VC5509A DSP的耐火砖内部缺陷检测系统。根据检测实际需要选择器件,确定模块与功能划分,并设计各个模块的电路接口,完成原理图与PCB版图的绘制以及硬件制作。将理论研究阶段的各个处理算法编写成C语言子程序,设计主程序使系统在无操作系统下合理调度各个子程序,并利用键盘和LCD显示屏进行人机交互。耐火砖内部缺陷检测系统运行稳定、快速、操作简单、功能完备,完全满足耐火砖内部缺陷在线检测的需要。更多还原

【Abstract】 Along with the progress and development of the high temperature industrial technology, quality of the refractory is also getting higher and higher, especially the refractory brick has been applied massively in ferrous metallurgy industry, its quality and life relate to the safety and economic efficiency of ferrous metallurgy industry. The technique of refractory brick has formed, but the technique of testing the interior quality fault of the refractory brick has not been developed, until now, the overwhelming refractory diagnosis method is backward. In this paper striking method is adopted to testing the interior quality fault of the refractory brick, according to refractory brick striking sound it is judged if there are defects and differentiate the type of the fault.In order to diagnosis the interior quality fault of the refractory brick, it needs to obtain the useful data information from the rap sound signal. Firstly, it uses finite impulse response (FIR) digital filtering technique. On the other hand, striking sound is a series of interval signal, adopting double threshold comparison method to judge the start and ending points of striking sound, then to get the extracted signals. Extracted signals are time-domain signal without obvious regulation and it cannot be used to detect defects. Appling Welch method to estimate the power spectrum and then detecting by spectrum data, nevertheless, peak points are energy concentrations in one signal, and there are correlation within peak points of the same spectrum. We can extract peak points to be used characteristic data in refractory brick interior quality fault diagnosis. Above methods have the characteristics that computes rapidly and meet real-time demand of online detection.This paper puts forward and experiences an effective method of interior defect diagnosis of refractory brick based on the fisher discriminate analysis which uses characteristic data. The author discovered the lack of fisher discriminate by analyzing method through it is used, it is normal class which has led characteristic value decomposition, it caused the differentiate which is not obviously between the flaw class. In view of the above situation it proposed the interior defect diagnosis method of refractory brick based on PCA-FDA mixed model. After these, the author took test on simulation of these means, and gained better defect test and diagnose result, explained that the method may apply in the actual system.Digital signal processor (DSP) has the formidable ability of signal processing, it may be competent to the online detection definitely. The author chooses the component according to the actual need, definite the module and the function, designs the electric circuit connection of each module, completes the schematic diagram and the PCB, as well as make the hardware. Compiles the C language based on the processing algorithm during the theory research stage, designs main program to make the system dispatching each subroutine program reasonably under the no-operation system. Operator can connect the system by the keyboard and LCD. The interior defect diagnosis system of refractory brick has the characteristics of stable, fast, operating simply, function completed and meeting the demand of online detection completely.更多还原