Consumption Prediction Of Energy Media For Typical Operations And Products In Iron And Steel Industry

Iron and steel industry is one of the key industries in energy consumption area in China, and has a large amount of energy consumption and a low-level of resource utilization. Now, iron and steel industry is facing an increasingly grave situation in such a society which has a big demand for energy. Based on the research about the prediction of the product-oriented energy consumption in iron and steel enterprises, three kinds of prediction algorithms are proposed to predict the energy media consumption. Based on these algorithms, two prediction systems of energy media consumption are developed for operations and products in iron and steel enterprises, respectively. The implementation of the developed prediction systems can help to improve the utilization rate of energy media and reduce production cost.The main contents of this thesis can be summarized as follows:(1)This thesis takes the prediction problem of energy medium consumption in production processes in an iron and steel enterprise in China as the research background. Through the research and analysis of the practical problem, the energy consumption prediction problem of typical processes of iron and steel enterprises is derived, and three energy consumption precision problem of hot rolling products is put forward, respectively.(2) Based on the dynamic features of the prediction problem of energy media, three prediction algorithms are developed:the classical methods of linear regression and least square support vector machine, and a new method named the approximate dynamic programming (ADP) algorithm based on parameter models. For the ADP based on parameter models, the gradient method and recursive least squares method are used to obtain the first-order and the second-order parameters, and then the better parameters are used to construct the prediction model. The proposed three algorithms are tested based on practical production data, and the computational results show that the ADP based on parameter models is significantly superior to the classical methods such as the linear regression and the least square support vector machine.(3) For the energy consumption prediction problem for operations in iron and steel enterprises, an energy consumption prediction system is developed. As a whole management system, this system can effectively predict the values of energy consumption of various operations in a period of time in the future, and based on these prediction values the balance analysis of energy media and the key index management can be realized.(4) For the precise measurement problem of energy media consumption in the hot rolling production line, a software system of precise measurement and statistical optimization of the energy media consumption is developed. The main functions of this system are focused on the energy consumption prediction for detailed products in the hot rolling line, and can provide an effective decision-making tool for the precise management of energy media.