经验模式分解法结合相关积分法的离心风机喘振先兆辨识

Surge Synptom Identification of Centrifugal Blowers by EMD in Connection with Correlation Integrals

针对离心风机的喘振问题,在研究相关积分算式参考距离取值原则的基础上,提出了一种经验模式分解法(EMD)结合相关积分法的喘振先兆辨识方法。利用EMD对信号的自适应分解特性,将测量信号分解为有限的几个本征模态函数(IMF),通过与喘振对应的IMF分量的相空间重构,从而实现系统喘振工况动力学特性的再现。在此基础上,根据滑动窗计算相关积分值得到相关积分曲线,以清晰反应喘振的起始时刻和其发展变换的过程。实验数据分析表明,喘振所对应的IMF分量可清楚反映出稳态向喘振过渡时风机出口压力的脉动,与直接采用相关积分法相比,喘振前相关积分值下降幅度更大,更早时刻就反映出了喘振先兆信息,其有效性被现场监测仪器所验证。

Focusing on the surging problem of centrifugal blowers and based on research concerning reference distance value taking principles in correlation integral formulas, a new identification method of surge symptoms is presented, inwhich empirical model decomposition （EMD） is used in connection with the correlation integration method. The measured signal is decomposed into a finite number of intrinsic mode functions（IMF） by making use of EMD＇s self-adaptive decomposing property. Then, via reconstrncing the phase space with surge related IMF components, the dynamical characteristics of the system＇ s state of surge can be reproduced. The correlated integration curve is then obtained witht ehc orrelated integral values calculated by the sliding windows, clearly reflecting the starting time, incipience and the development process of surge. An analysis of experimental data demonstrates that components, corresponding to the surge, can distinclty reflect fluctuations of the centrifugal blower＇ s outlet during the transition period from stable to surging state. Surging symptoms are herewith reflected earlier and the the IMF pressure drop-off of the amplitude of the correlation integral value prior to surge are larger than those obtained with the integral method. Its effectiveness has been vindicated by on site monitoring instruments.