Energy Consumption Prediction Of Iron And Steel Enterprises Base On The Interval Neural Networks

Our country has a lot of basic industry, iron and steel industry is the most important nuclear industry of them. The production is more as the capacity promote by year, meanwhile, there must be much energy consume. The energy consumption of per ton steel of our country is much more than the average of the world per some reason. The energy consumption of per ton steel of the emphasis corporate is 10% more than the advanced standard in the word. It's so tight that we need to make the lower consume for iron and steel that it will impact our development of country.The goal of this subject is about the energy conservation of iron and steel, do the analysis and forecast confirm to the energy conservation requirement of our country. The solution not only has the practicability for the iron and steel industry and the other business, but also has a economic value and social meaning.Series of work start around the nuclear goal of iron and steel energy consumption, the specific work is as follows:Base on the material flow model of the iron and steel industry, start the work from the benchmark material flow, make each analysis according to the situation. Deeply research about the flow direction of the material flow of Fe have been made, further analysis for the impact of material with Fe which diverge the benchmark material flow to the energy consumption of per ton steel. The conclusion show that the industry will benefit if there is material which be added to the process from the outside, and the later the process is, the more benefit we will get. Opposite, if the material is outflow of the process or go back to the early process to do the second machine, the consume will be more, and the later process is. the consumption will be more.This article choose the interval neural network as the model during the energy consumption prediction of iron and steel industry, the material flow with Fe is the typical material which impact the comprehensive energy consumption of per ton steel, and it will be chose as the import. Base on the dimension reduction of principal component analysis, a lot of variable which have a influence of the comprehensive energy consumption of per ton steel have been reduce to 3 new comprehensive variable. The data will be given to the model as the import after the reduction of dimensionality and shift-out redundancy so that the complexity of the prediction model could be improved. Interval neural network is superior to the BP neural network to analysis and forecast, and this method can get the better and fully predictive effect.