[4M3-GS-13] AI application: Electric power

[4M3-GS-13-03] Factor Analysis of Nitrogen Oxides Emissions in Coal Fired Power Plant with LiNGAM

〇Tatsuki Saito¹, Koichi Fujiwara¹ (1. Nagoya university)

Keywords:Causal Inference, LiNGAM, Coal Fired Power Plant

Coal has been an important energy source worldwide, and over 30\% of electricity in japan is covered by coal fired power generation; however, its NOx emissions is large since the amount of nitrogen contained in coal is larger than other fossil fuels. Thus, precise control of NOx emissions is required in coal-fired power plant operation. Although the trend of NOx generation can be theoretically calculated, it is difficult to predict real NOx generation because it is affected by complicated factors such as furnace design and the operating conditions. To identify operating factors that affect NOx generation is needed.

LiNGAM is an exploratory causal analysis method, which identifies a causal ordering of variable and their connection strengths without any prior knowledge on causal relationship among variables. In this study, real operation data collected from a coal-fired power plant were analyzed using LiNGAM in order to identify NOx generation factors. The causal relationship between each process variable and NOx generation was estimated by LiNGAM, and their connection strengths on NOx generation was estimated. The results agreed with previous reports about the NOx generation causes. Our analysis demonstrated that LiNGAM is an useful method for identifying important operational factors in processes.