Study on the parameter discretization algorithm of furnace flame image based on rough set theory
Rongbao Chen, Wuyong Ma, Benxian Xiao, Zipei Cao · 2016
The combustion parameters were described as the interval number because of the pulsating flame, but it could not eliminate the data redundancy, complex relationship among the multiple attributes and interval data continuous uncertainty. Based on rough set theory, the paper proposed an interval number discretization algorithm. First, based on the multi-attribute decision making algorithm, a combustion interval decision sample table was formed. Second, the distance and similarity relationship were used to quantify the interval parameters of the combustion attribute objects, and the similarity threshold value was defined to determine the similarity relationship between the combustion parameters. The paper defined the concept of rough entropy and calculated the upper and lower rough entropy to obtain the optimal similarity threshold value. Then the similarity matrix among interval objects was determined and used minimal discretization intervals to obtain the final discretization result. The experiments show that the discretization result reduces the redundancy and complexity of combustion data, and improves the confidence of combustion data.