基于声发射的流化床故障诊断方法研究

发布时间：2018-05-18 17:39

本文选题：气固流化床 + 声发射　；参考：《内蒙古科技大学》2015年硕士论文

【摘要】：化学工业是国民经济的支柱产业，化工生产合成聚合物中产量最大的是聚乙烯，其生产方式分为高压法、中压法、低压法三种。低压生产中有溶液法、淤浆法和气相法，相对比较，气相法生产具有明显优势，得到广泛应用。气相法生产聚乙烯主要依靠气固流化床反应器，其系统较为复杂，其工作环境常常具有高温、高压、高湿度等特点，在长期的运行当中极易发生故障，若不能及时发现，将会造成设备损坏而停产的危险，因此流化床在线故障诊断就显得非常的重要。本文研究的主要内容是声发射技术应用于流化床故障的测量。针对化工生产中气固流化床风帽故障影响流化质量问题，借助声发射技术进行定位测量，通过均匀安装在流化床分布板下方的声发射传感器，进行气固流化床内固体颗粒撞击分布板的声信号采集，再对该信号进行多尺度小波包分解，找出特征频段能量值和能量分布差异，提出一种能够快速准确地检测出风帽故障位置和故障类型的方法，并完成实验，验证其方法的可行性。此外，流化床内流化粒子受原料质量、催化剂等因素影响会发生局部高温而产生结块，本文应用声发射传感器采集流化床内颗粒撞击器壁的声信号，将检测的数据经经验模态分解与能量信息熵分析进行故障特征向量提取，，然后将提取的特征向量输入到支持向量机中进行分类训练，完成结块故障的分类。在实验室冷模流化床实验平台模拟了风帽故障与动结块故障，完成了信号采集以及故障识别算法的验证，实验表明声发射技术用于流化床故障诊断具有良好的效果。
[Abstract]:The chemical industry is the pillar industry of the national economy. The largest production of synthetic polymers in chemical production is polyethylene, which is divided into three production modes: high pressure method, medium pressure method and low pressure method. There are solution method, slurry method and gas phase method in low pressure production. Compared with gas phase method, gas phase method has obvious advantages and is widely used. The gas-solid fluidized bed reactor is the main way to produce polyethylene in the gas phase process. Its system is complex, and its working environment is usually characterized by high temperature, high pressure and high humidity. It is easy to break down in the long run, if it can not be found in time, Because of the danger of equipment damage and shutdown, it is very important to diagnose the fault in fluidized bed. The main content of this paper is the application of acoustic emission technology in fault measurement of fluidized bed. In view of the problem that the air-solid fluidized bed air cap failure affects the fluidization quality in chemical production, the acoustic emission sensor, which is uniformly installed under the fluidized bed distribution plate, is measured by means of acoustic emission technology. The acoustic signal of solid particles impinging on the distributing plate in a gas-solid fluidized bed is collected, and then the signal is decomposed with multi-scale wavelet packet to find out the difference of energy value and energy distribution in characteristic frequency band. A method to detect the fault location and type of the hood quickly and accurately is proposed, and the experiment is completed to verify the feasibility of the method. In addition, fluidized particles in fluidized bed are affected by raw material quality, catalyst and other factors to produce caking. In this paper, acoustic emission sensors are used to collect acoustic signals of particle impingement on the wall of fluidized bed. The detected data are extracted by empirical mode decomposition and energy information entropy analysis. Then the extracted feature vectors are input into support vector machine for classification training to complete the classification of caking faults. The fault of air cap and dynamic caking is simulated on the laboratory cold model fluidized bed experiment platform. The signal collection and fault identification algorithm are verified. The experiment shows that acoustic emission technology has good effect in fault diagnosis of fluidized bed.
【学位授予单位】：内蒙古科技大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TQ051.13

【参考文献】