【作者】 夏豹；

【导师】 高建强；

【作者基本信息】 华北电力大学 ， 热能工程， 2011， 硕士

【摘要】 目前循环流化床锅炉压力测点都是安装在炉膛壁面上,由于大量物料循环,频繁摩擦撞击测点,使得测点堵塞磨损问题非常严重,造成测量不准确甚至测点损坏。针对这一问题,本文在搭建冷态鼓泡流化床试验台上,将压力测点布置在风室风帽入口的壁面上,通过压力信号采集系统,将采集到的压力信号转变为数字信号存储于计算机中。在试验过程中,通过调整和改变试验的控制条件,采集多种正常工况与扰动工况下的压力波动信号,并对信号进行深入的研究分析,探讨流化床内的压力波动状况。基于Matlab编程,利用自相关函数的傅里叶变换的方法,将采集到的时域信号转变为频域信号,从得到的信号频谱图上可以清楚的看到,风帽入口压力波动与流化床内静床高和表观气速等参数变化密切相关,并且压力波动能量主要表现为低频压力波动,说明在风帽处采集的监测信号能够反映流化床内压力波动状况。接着对信号进行消噪处理,通过对比分析工程滤波法与小波消噪法,结果证明小波消噪法优于工程滤波法,并进一步通过对比选择不同的小波基进行消噪,结果表明利用db2小波基三层分解消噪方法,能达到良好的去噪效果。试验中,模拟了现场流化床中压力信号监测失败、加放料产生扰动、风帽堵塞等扰动工况,利用时域分析、频域分析以及小波包法分析,对比了正常工况与扰动工况下的不同,建立了故障信号的诊断识别的方法。利用时域分析,通过设定置信区间的方法,来判定压力信号是否发生了故障;利用频域分析,对比了正常工况与加放料扰动工况频谱的不同,确定了频谱作为特征向量,实现故障诊断识别;利用小波包分析,将压力信号进行三层分解与重构,对比正常与扰动工况下各个节点功率谱的不同,结果表明信号在结点s30的功率谱p30可以作为特征向量,判定流化床内是否出现风帽堵塞等问题,实现了故障诊断,为现场应用提供理论参考及依据。更多还原

【Abstract】 Currently the boiler pressure measuring points of the circulating fluidized bed are all installed in the furnace walls. Extensive recycling of materials, and frequent frication and clash of measuring points make the plug wear problem at the measuring points very serious, resulting in inaccurate measurements or damage of the measuring points. To solve the problem, in this thesis I arrange pressure measuring points on the walls at the entrance of the wind chamber hood when setting up cold bubbling fluidized bed test rig, then the collected pressure signals are transformed into digital signals with pressure signal acquisition system,and the digital signals are stored in the computer. During the test, I collect pressure fluctuation signals in normal operating conditions and disturbance conditions by adjusting and changing the test’s control conditions and carry out in-depth analysis and research of the signals to probe pressure fluctuation conditions in fluidized bed.Based on Matlab programming and using autocorrelation function of the Fourier transform method, the time domain signals are transformed into frequency domain signals. From the received signal spectrum we can clearly see that the pressure fluctuation at the hood inlet is closely related to some parameters such as static bed height of fluidized bed and superficial gas velocity, in addition pressure wave energy is mainly manifested by low-frequency pressure fluctuations, which show that monitoring signals collected at hood entrance can reflect the status of pressure fluctuation in fluidized bed. Then I denoise the signals,by comparative analysis of project filter method and wavelet denoising method,wavelet denoising method is proved better than project filter method. Further by selective comparison of different wavelet bases for denoising, it is concluded that db2 wavelet three-layer decomposition denosing method can be used to achieve the best denosing effects.During the test, I stimulate some disturbance conditions such as failure of monitoring pressure signals in fluidized bed scene, disturbance generated by placing materials and hood blocking. By using time domain analysis, frequency domain analysis and wavelet packet analysis, I reach different results in normal conditions and disturbance conditions and establish the fault signals’diagnosis and recognition method. I use time domain analysis, by using method of setting confidence interval, to determine whether pressure signals go wrong. I use frequency domain analysis, by comparing different spectrums in normal conditions and placing materials disturbance conditions, to serve spectrum as feature vector and achieve fault diagnosis recognition. By using wavelet packet analysis, carrying out three-layer decomposition of the pressure signals and reconstructing them, and comparing different node powers at each node in normal and disturbance conditions, it is concluded that the signal power spectrum at p30 can serve as feature vector to determine whether hood block arise in circulating fluidized bed, achieve fault diagnosis and provide a theoretical reference and basis for field application.更多还原