Development of PM Collection Efficiency Prediction Model Based on Deep Learning to Improve WESP PM Collection Performance

Wet electrostatic precipitators (WESPs) are commonly used in thermal power plants and industrial facilities where most particulate matter (PM) is generated. To ensure the efficient operation and maintenance of WESPs, it is important to develop technology that can accurately predict their PM collection efficiency. Although recent studies mainly focused on predicting the concentration of PM in the atmosphere, studies that predict the PM collection efficiency of WESPs installed in PM-generating facilities are insufficient. This paper proposes a new approach to predict PM collection efficiency using a deep learning model-based fine PM collection efficiency prediction model developed by collecting experimental data on various types of PM collection in WESPs. The developed deep learning model achieved excellent results with an R2 score of 0.964 and a mean absolute error of 0.747. Therefore, this paper proposes using deep learning to predict PM collection efficiency as a means to improve and maintain the PM collection performance of WESPs.