To improve the signal-noise ratio of online data identification for SCR flue gas denitration system

this paper applies three representative variable-step least square flitting methods to carry out data filtering for NOxconcentration at inlet and out let of the SCR system and the ammonia spraying amount

on the basis of variable-step self-adaptive filtering method

with different cut-off frequencies.The comparison between the filtering data and the original data shows that

although it costs longer iteration time

the variable-step size method based on a mutual correlation matrix possesses the best stability.Moreover

a slight up-drift occurs to the filtering data under a cut-off frequency of 0.9.At this frequency

the flitting data can also achieve the best general result.