| The overall situation of energy consumption and the systerm analysis of energy consumptiom for iron-steel enterprise were reviewed, especially the iron-making process energy consumption. It was pointed out that the feasibility and the necessity of systematic analysis of iron-making process. The energy consumption of the iron-making process in iron and steel enterprises is very large, in its energy consumption, the coke is the most important. The charged coke is influenced by the energy consumption of oke ratios, at the same time, hot air temperature , the quantity of blast ,silicon content in pig iron, the rate of stopped air, the ratio of dregs, the grade of ore and utilily coefficient had on effect on coke ratios. Based on the statistic data of Xinlingang iron and steel company from 1996 to 2005, the prediction of coke ratios is presented with different approaches. According to time series, the forecasting value of coke ratios is made by Box-Jenkis method. The average elative error reached the level of 2.02%. The gray system analysis and the grey forecast were also adopted in this paper to forecast its future coke ratios of December 2003, the average relative error of GM(1,1) model arrived at the level of 0.9352%. In the end of this paper, theory and method of artificial neural networks was introduced to analyse the process energy consumption in iron-steel enterprises. The history of development and application of the artificial neural networks were reviewed. The BP calculation model of neural networks was choosed as objects of study, its topology structure and function of learning and training were explained. The prediction of coke ratios was made by BP artificial neural networks and the prediction value was relatively accurate. Its average relative error is 0.908%. So the application of the three methods was necessary and feasible for the analysis of energy consumption in iron-making process. The prediction results had practical value to research on energy saving.
|
PDF Full Text Request