HRSG drum level sloshing model and level trend signal extraction based on non-linear filter

Abstract

To improve drum level control system accuracy of heat recovery steam generators (HRSG) in gas-steam combined cycle, the drum level sloshing thermaldynamic model is constructed The uncertainty factors influenced the drum water level model and its nonlinear features were analyzed. The factors influences the impulse characteristic of water level signal are described. The pressure drop is the main factor caused water level nonlinear features. The time-frequency distribution method (Hibert-Huang transform) is applied to reveal interfered components characteristics. The HHT spectrum with high time-frequency resolution represent level signal local features, which shows main interference component exists in very low frequency band (less than 0.02 Hz). The Alpha statistical model is used to describe drum level pulse noise. The low-pass Myraid nonlinear filter is designed to mitigate the effects of impulse noise, The experiment shows the proposed method is reasonable to process non-linear level signal of HRSG and gains good performance. The feedwater controller based on this drum level model can control the water level safely.