Method for predicting mixing ratio of upper furnace charge of blast furnace based on linear regression

The invention provides a blast furnace upper furnace charge mixing ratio prediction method based on linear regression, which comprises the following steps of: representing a furnace charge mixing ratio as a weighted sum of furnace charge operation parameter characteristics, establishing a hypothesis model, and determining a loss function to quantify a difference between an actual value and a predicted value of the mixing ratio; and fitting the hypothesis model, updating parameters, calculating a minimum loss mean value, obtaining an optimal prediction model, and selecting parameters of the optimal model to obtain a prediction mixing ratio. According to the method, linear hypothesis is carried out through furnace charge operation parameters, a hypothesis model of the furnace charge mixing ratio is obtained, training is carried out through measured furnace charge operation parameters in blast furnace simulation experiment data and the corresponding mixing ratio, an optimal coefficient solution is sought through