Multi-scale sample entropy analysis of furnace process in pulverized coal boiler

摘要

A feature extraction method based on Multi-scale sample entropy is applied to the nonlinear analysis of furnace pressure in a pulverized coal boiler. Firstly, multi-scale sample entropy characteristics of two typical chaos time series were studied, which revealed the natural nonlinear characteristics of the two signals accurately, and then based on this, the furnace pressure signals in three stable operation conditions including high, middle and low load were analyzed, which were collected from the DCS system in a 300MW pulverized coal boiler. The results indicated that the different uncertainty degrees of furnace process in three stable operation conditions were reflected by multi-scale sample entropy characteristics of furnace pressure signals in multiple scales. The higher uncertainty degree of furnace process in stable high load condition was shown as higher value and changing rate of sample entropy at large scales; the lower uncertainty degree of furnace process in stable low load condition was represented as lower value and changing rate of sample entropy at large scales; the uncertainty degree of stable middle load condition is between high and low load conditions. Multi-scale sample entropy analysis of the furnace pressure is helpful for further understanding the nonlinear characteristics of furnace process, and providing new methods and thoughts for the performance optimization of a pulverized coal boiler.