Abstract: To adress the problem that the emission concentration of municipal solid waste incineration (MSWI) is difficult to control efficiently due to the lack of dust prediction model, this paper constructs a prediction model of dust concentration in MSWI process based on Takagi-Sugeno (TS) fuzzy forest regression (TSFFR). First, the feature random sampling strategy was used to generate a set number of training subsets for the original dust data set. Second, a TS decision tree (TSDT) sub-model was constructed for each training subset, which included a two-layer structure of feature screening based on non-leaf nodes in the screening layer and TS reasoning based on leaf nodes in the fuzzy reasoning layer. The former part of the latter part updated the parameters with standard normal distribution and uniform distribution strategy, and the latter part initialized the weight strategy using prior knowledge to update the parameters. Finally, the TSFFR model was obtained by integrating all sub-models in a pseudo-inverse manner to predict the dust emission concentration. The effectiveness is verified using industrial dust data from an MSWI factory in Beijing.