| Plenum melting process of Al2O3 is an extremely complex process, where to melt A12O3 out from aluminum-mine in caustic liquor on the technical condition of high-temperature and plenum. The caustic to alumna molar ratio and digester ratio are two important economic and technical indexes, where not only decide the melting efficiency of Al2O3 and soda consume, but also greatly affect the following producing of Al2O3. But the two indexes are lagging in actual producing severely. So it is very significant to make the research of the caustic to alumna molar ratio, digester ratio in the plenum melting process and soft sensing computer system and develop the computer system that forecasts the caustic to alumna molar ratio and digester ratio online for directing producing.The paper analyzes the mechanics of the plenum melting process. Analyze the relationship between elements that affect the caustic to alumna molar ratio and digester ratio. Find the ruling variable using the principle component analyze, conduct data and rectify errors. Built the soft-sensing model of the caustic to alumna molar ratio and digester ratio combining the methods of mechanic analysis, neural network and so on.Based on the data-preconditioning, mechanic module, neural network, table-match function of soft measure, Visual C++ 6.0 including programming technology that faces to the subject, communication technology and database technology developed the center module of data-deriving, the caustic to alumna molar ratio and digester ratio forecast as well as some auxiliary modules such as forecast-data comparison and curve display. It was able to forecast the caustic to alumna molar ratio and digester ratio on-line.The system was debugging and operating on scene and succeeded to put the module function mentioned above into practice. The scene operating condition proved that the system was of reliability and availability and it was able to content the locale-required precision. |
PDF Full Text Request