The invention relates to the technical field of gas control, in particular to an
automatic control method for optimizing
combustion of a
hot blast stove based on gas prediction, which comprises the following steps of: estimating
blast furnace gas generation amount, acquiring real-time operation parameters of
blast furnace smelting, establishing a neural
network model, outputting
blast furnace gas prediction generation amount, and correcting the gas generation amount
estimation. Performing
blast furnace gas generation and use prediction, performing feature analysis through
feature screening to obtain key influence factors, performing neural
network model training, predicting the
blast furnace gas by adopting LSTM (
Long Short Term Memory)
time sequence prediction, and outputting the generation amount or consumption amount in the future 30 minutes; the intelligent
combustion of the
hot blast stove comprises a gas flow measurement and calculation model, a vault temperature regulation and control model, a
waste gas temperature regulation and control model and a high-calorific-value
coal gas time-sharing
combustion management and control model. Through precise prediction of gas generation amount, usage amount and equipment and in combination with a combustion control strategy, the energy utilization efficiency is improved, the hot
air temperature is stabilized,
gas consumption is reduced, and environmental
pollution is reduced.