Characteristics Analysis of HRSG Drum Level Oscillation Signal Based on EMD

Drum level noise makes influence on level control accuracy of heat recovery steam generators (HRSG) in gas-steam combined cycle. To improve the accuracy and robustness of control system, the nonlinear characteristics of level signals are exploited. Empirical mode decomposition (EMD) is an adaptive frequency band dividing method of signal. Through EMD, practical drum level signal is decomposed into a series of IMFs and their spectrum is analyzed. Each IMF component covers specific frequency band. Useful IMFs are selected to reconstruct the true drum level signal without corresponding disturbance. The true level signal is immune to HRSG operating conditions and noise. Under pure level noise and two types circumstances, the experimental results of level signals show the proposed method is reasonable to extract true level signal and deal with nonlinear level signal of HRSG