abstract |
Disclosed is a dioxin emission concentration prediction method based on the hybrid integration of a random forest and a gradient boosting tree. The method comprises: firstly, performing random sampling of training samples and input features on DXN modeling data with a small sample high-dimensional characteristic, so as to generate a training subset; then, establishing, on the basis of the training subset, J RF-based DXN sub-models; then, performing iteration I times on each RF-based DXN sub-model, and constructing J×I GBDT-based DXN sub-models; and finally, combining predicted outputs of the RF-based DXN sub-models and the GBDT-based DXN sub-models by using a simple average weighting method, and obtaining a final output. By using a DXN prediction model construction method integrating RF and GBDT, the online DXN prediction precision can be improved, the operation optimization of MSWI process operation parameters is facilitated, and the economic benefits of enterprises are improved. |