Prediction method of dioxin emission concentration based on PCA and deep forest regression

Municipal solid waste incineration (MSWI) is a widely used technology for reducing, harmless and recycling municipal solid waste. However, dioxin (DXN), a highly toxic pollutant in the exhaust gas of MSWI process, is the main factor that causes the "NIMBY effect" in building incineration power plants. The existing DXN detection method by combining long-period online sampling and off-line testing cannot meet the requirement of direct optimization control of DXN. To solve this problem, the prediction model of DXN emission concentration based on easy-to-measure process variables has become a research hotspot. However, due to the high dimension of process variables and the complex mechanism of DXN generation, adsorption and emission, it is difficult to effectively select input features of DXN model. Therefore, this paper proposes a modeling method based on principal component analysis (PCA) and deep forest regression (DFR). At first, the dioxin emission characteristics of MSWI process are described. Then, a modeling strategy and algorithm including PCA dimension reduction and DFR modeling are proposed. Finally, the effectiveness of the proposed method is verified by using the Benchmark data set and the actual DXN data.