AN APPROACH OF USING COMPUTER IN BLAST FURNACE CONDITION PREDICTING
Shen Xing (Institute of Chemical Metallurgy, Chinese Academy of Sciences)
An attempt is made to introduce the concept of artificial intelligence to the process control of a blast furnace. A PROLOG program is used for predicting the process condition and providing operational guidance. The identification of the furnace condition is made on the basis of careful analysis of the gathered data, in which a series of calculations and comparisons may be involved. The results of these calculations support the furnace condition recognition according to the rules of the empirical knowledge. A set of standard values is defined by applying statistical tests to the normal condition of the process, that plays an important role in the judgement for the process condition. At the same time, the online data is compared with their former values for estimating the trend of variations. More attention is paid on some important irregular conditions of the blast furnace are paid more attention. The methodology in expressing the experienced knowledge of the blast furnace is the main problem in the establishment of the knowledge base, which should be open to the users for modification and knowledge accumulation. Some of the imitation of the thinking process of the blast furnace operators are made. The results of the calculations and data analysis are expressed in the form of facts and are linked to the rules for the judgement of the furnace condition. The furnace condition is fully identified as normal or certain trouble being present. Suggestions on the operational acts are given as well.