RESIDUAL FITTING MECHANISM-BASED SIMPLIFIED DEEP FOREST REGRESSION SOFT MEASUREMENT METHOD FOR FURNACE GRATE FURNACE MSWI PROCESS DIOXIN EMISSION

摘要

Provided in the present invention is a residual fitting mechanism-based simplified deep forest regression soft measurement method for furnace grate furnace MSWI process dioxin emission. Toxic pollutant dioxin (DXN) generated in a solid waste incineration process is a key environment index which must be minimized and controlled. Carrying out rapid and accurate soft measurement on the DXN emission concentration is the to priority to reduce the discharge of this type of pollutants. The method comprises: firstly, performing feature selection on a high-dimensional process variable by using mutual information and a significance test; then, constructing a simplified deep forest regression (SDFR) algorithm to learn a nonlinear relationship between the selected process variable and the DXN emission concentration; finally, designing a gradient enhancement strategy on the basis of a residual fitting (REF) mechanism to improve the generalization performance of the layer-by-layer learning process. Compared with other methods, the present method is better in prediction precision and time consumption.