基于压力涨落信号的气固流化床早期颗粒结块故障诊断研究

Study of Diagonosing Incipient Particle Agglomeration in a Gas solid Fluidized Bed using the Pressure Fluctuation Signal

利用压力涨落信号对气固流化床颗粒结块故障进行早期诊断，有着重要的工业应用意义。本文首先从信号产生的物理背景出发，认为压力涨落信号可分解为分形布郎运动和高斯白噪声的加和。结合正交小波变换理论，应用期望最大化算法（ＥＭ），得到了压力涨落信号的自相似参数Ｈ和高斯白噪声强度σＷ的极大似然估计，从而证实了这种分解的可行性。应用参数Ｈ和σＷ对模拟颗粒结块故障的研究表明，流化床床层下部压力涨落信号的Ｈ和σＷ参数均对此故障敏感，本文对结果进行了物理解释。因此，Ｈ和σＷ参数有望应用于对此故障的早期诊断。

Diagnosis of the particle agglomeration in the incipient stage using the pressure fluctuation signal in gas solid fluidized beds has important value to industrial application.In this paper, considering from the physical background, we model the signal as the addition of fractal Brownian motion(FBM) and White Gaussian noise(WGN).With an advancd algorithm based on orthornormal wavelet transform, the self similar parameter H of FBM and the intensity parameter σ w of WGN are robustly estimated and such decomposition of the signal is validated consequently.We study the simulated particle agglomeration with H and σ w and find that both of the two parameters of the pressure fluctuation in the lower portion of the gas solid fluidized bed are sensitive to the inci pient particle agglomeration. Thus, they are promising to diagnose the malfunction.