Optimized scheme in coal-fired boiler combustion based on information entropy and modified K-prototypes algorithm

• Main Authors: Hui Gu, Hongxia Zhu, Yanfeng Cui, Fengqi Si, Rui Xue, Han Xi, Jiayu Zhang
• Format: Article
• Language: English
• Published: Elsevier 2018-06-01
• Series: Results in Physics
• Online Access: http://www.sciencedirect.com/science/article/pii/S2211379717322337

An integrated combustion optimization scheme is proposed for the combined considering the restriction in coal-fired boiler combustion efficiency and outlet NOx emissions. Continuous attribute discretization and reduction techniques are handled as optimization preparation by E-Cluster and C_RED methods, in which the segmentation numbers don't need to be provided in advance and can be continuously adapted with data characters. In order to obtain results of multi-objections with clustering method for mixed data, a modified K-prototypes algorithm is then proposed. This algorithm can be divided into two stages as K-prototypes algorithm for clustering number self-adaptation and clustering for multi-objective optimization, respectively. Field tests were carried out at a 660 MW coal-fired boiler to provide real data as a case study for controllable attribute discretization and reduction in boiler system and obtaining optimization parameters considering [maxηb,minyNOx] multi-objective rule. Keywords: Boiler combustion efficiency, NOx emissions, Multi-objective optimization, Clustering