Oxygen And Iron And Steel Enterprises To Predict Optimal Scheduling Model

With the development of industrial technology, information and automation control technology are gradually applied in metallurgical process of iron and steel complex. Some enterprises establish EMS to monitor and control energy-flow and energy utilization of the manufacturing process. Oxygen is an important energy medium which is indispensable in ironmaking and steelmaking process. Oxygen system is composed of production, storage and consumption units. Optimal operation of it is an effective way of energy saving and emission reduction. In this paper we established prediction and scheduling models of the oxygen system to forecast the oxygen comsumption with artificial intelligence methods, and then guide dispatching and operation with objective optimization models and unit models, to decrease oxygen diffuse quantity, save energy cost and enhance working efficiency.To predict the oxygen system running trend and consumption of each consumer and offer production scheduling support for data. In this paper, we established prediction models of oxygen system based on support vector machine (SVM) and least squares support vector machine (LSSVM) with the actual data of iron and steel complex. With the methods of data extraction, cleaning, normalization preprocessing and optimization selection model parameter to get prediction models with different fitting precision. Comparing with these models, we found that the nonlinear regression improved models of SVM and LSSVM are more suitable to predict consumption of the oxygen system. And the predition accuracy of the improved LSSVM predition model is higher, arithmetic speed is faster than others, suitable for widespread use of online prediction in iron and steel complex oxygen system.To regulate the balance of oxygen system efficiently, we established target optimization convex programming models which include minimum diffuse quantity of oxygen, minimum operation cost and maximum comprehensive income of air separation system, and established production, storage and consumption unit models as main constraints based on the prediction models. At the same time, make out dynamic scheduling rules on different production process of iron and steel complex for regulate scheduling operation, ensure oxygen system dynamic balance and reduce the diffuse of oxygen.The aplication shows that the predction and scheduling models is helpful for production operation and oxygen dispatch. During the study phase from15-19h (5hours) on November14,2011, the MSE between forecast data and real data of oxygen consumption about1#BF is507409.49m3, MAE is579.91m3and the MAPE is less than0.282with application of the prediction model in the oxygen system of an iron and steel enterprise. The MSE of2#BF is19440.79m3, MAE is79.00m3and the MAPE is0.028. The MSE between forecast data and actual data of oxygen consumption on LD is5464.8m3, MAE is62.4m3and the MAPE is0.003. The diffuse quantity of the oxygen system decreased about58636m3with application of the scheduling model, and liquid oxygen for external supply increased7.31tons, In conclusion, the LSSVM prediction model and the objective optimization scheduling model of iron and steel complex’s oxygen system that can forecast the running trends accurately, and schedule with the prediction result. It’s good for maintain dynamic balance of the oxygen system, reduce the diffuse quantity of oxygen and energy costs effectively, increase productivity and revenue.