【作者】 吴贤国；

【导师】 黄志尧；

【作者基本信息】 浙江大学 ， 控制科学与工程， 2002， 硕士

【摘要】 流化床参数检测是流化床研究中一个急待发展的探索研究领域。虽然有不少学者多年来一直致力于这方面的研究，但由于流化床系统本身的复杂性，使得大多数参数检测方法未能在工业现场得到应用。本文在对国内外大量有关流化床参数检测技术文献资料的深入研究基础上，采用分形技术、模糊信息处理技术对气固流化床从固定床、鼓泡床到湍动床的流型辨识以及颗粒结块故障诊断进行了研究。取得的主要成果如下： 1．在通过FBM(分数布朗运动)数据仿真证明了气固流化床压力波动信号与分数布朗运动是相似的基础上，提出了用分数布朗运动来模拟气固流化床压力波动信号，并采用R／S分析法从信号时间序列中提取出Hurst指数，通过分析信号Hurst指数值对流化床流型和结块故障进行了研究。 2．提出了将分形技术应用于气固流化床流型的分析和辨识的新方法，并构造了相应的特征量即压力波动信号的Hurst指数值。利用R／S分析法对实验数据进行了处理分析，发现信号的Hurst指数值在固定床、鼓泡床、湍动床三种流型下有不同的主分布区间，在此基础上提出了根据Hurst指数值大小进行流型辨识的方法。为了提高辨识准确率，引入了模糊信息处理技术，提出了流型模糊辨识中隶属度函数的确定办法，建立了气固流化床从固定床、鼓泡床到湍动床的流型判别准则，实验验证了该方法有良好的辨识效果，辨识准确率在80％以上。与传统的辨识方法相比，具有简单、客观、适用性强、可以在线实现等优点。 3．通过向床层中加入聚乙烯块状颗粒来模拟气固流化床颗粒结块现象。对结块前后的压力波动信号进行R／S分析后发现，颗粒结块使得信号的Hurst指数值明显降低。在此基础上定义了故障系数来判断颗粒结块故障的发生，实验结果表明相对于Hurst指数值故障系数对颗粒结块现象更加敏感，当故障发生时判断正确率达到80.8％。进一步的分析表明，通过实时调整Hurst指数标准值可以提高判断准确率。该方法与传统的结块故障诊断方法相比，具有快速、实时等优点，因此具有一定的工业实用价值。更多还原

【Abstract】 Parameter measurement of fluidization was important and needed to be solved imminently. Many researchers devoted to it but had gained a little achievement because of the complexity of fluidization. In this thesis,flow regime identification and malfunction diagnosis of fluidization were studied with fractal technology,the presented methods for both flow regime identification and malfunction diagnosis were proved of effectivity. main achievements are as following:1. Fractal Brownian Motion(FBM) was made from Gauss Noise and compared with pressure fluctuation signal of gas-solid fluidization,which demonstrated the similarity between the FBM and the signal. Then,Hurst Exponent was calculated from time series of the signal with R/S method. The exponent was applied in research of the identification and the diagnosis.2. Pressure fluctuation signals of fixed bed,bubbling bed and turbulent bed were analyzed with R/S method. The result showed that the Hurst Exponent of the signal have obvious differences between different flow regime. It increased when flow regime was changed from fixed bed to bubbling bed and more from bubbling bed to turbulent,based on which a new method for identification of flow regime was proposed. To improve the accuracies of identification fuzzy technology was introduced. Experimental results showed that the method was effective and the accuracies can get to above 80%.3. Agglomerate particle was put into the fluidization bed to simulate the agglomeration malfunction. It was found that the Hurst exponent decreased when the malfunction happened. A malfunction coefficient was introduced to diagnose the agglomeration malfunction. Experimental results showed that the malfunction coefficient changed more obviously than Hurst exponent did when malfunction happened,the accuracy of diagnosis was 80.8%.更多还原