The invention provides a method for predicting the permeability of blast furnaces. The method includes acquiring historical data of operation databases of the blast furnaces; analyzing the historical data and preprocessing the historical data; selecting the historical data to obtain actual data meeting production requirements; acquiring influence factors for the permeability of the blast furnaces in current time periods and sequencing weights of the influence factors according to contribution of importance; establishing corresponding data sets according to permeability influence factor weight sequencing results; classifying influence factors of the data sets according to permeability parameters and computing centers of the influence factors in each category; building a plurality of permeability prediction sub-models on the basis of black box model building; inputting adjusted original fuel quality parameters or operation control parameters into the corresponding permeability prediction sub-models according to requirements and corresponding categorical data; dynamically updating model parameters by the aid of the latest data according to the data acquisition frequencies of the operation databases of the blast furnaces. The method has the advantages that permeability models can be dynamically updated, and accordingly the prediction accuracy can be improved.