钢铁企业典型产线及产品能源介质消耗预测

发布时间：2018-04-22 06:32

  本文选题：钢铁工业 + 热轧产品　； 参考：《东北大学》2013年硕士论文

【摘要】：钢铁工业是我国能源消耗的重点行业,能源消耗大,资源利用率低,在能源需求日趋紧张的今天,所面临的形势异常严峻。本文通过对钢铁企业典型产线以及产品的能源介质消耗预测问题的研究,提出了三种基于数据驱动的预测算法,来预测未来一段时间的能源消耗值,并分别开发了面向工序和面向产品的钢铁企业能源消耗预测系统,目的在于提高钢铁企业能源利用率,减少生产成本。本文的主要内容如下：(1)以国内某钢铁企业生产过程中面临的能源介质消耗预测问题作为研究背景,通过现场调研,提炼出钢铁企业面向典型工序的能源介质消耗预测问题,并进一步提炼出面向热轧产品的能源介质消耗预测问题。(2)针对能源介质预测问题的动态特点,分别提出了三种预测算法：传统的线性回归算法和最小二乘支持向量机算法,以及基于参数模型的近似动态规划算法。对于基于参数模型的近似动态规划算法,分别采用随机梯度法与递推最小二乘法获取一阶和二阶模型参数,并通过比较两种参数模型的值函数,最终确定预测模型。对三种算法进行了数值实验,试验结果表明基于参数模型的近似动态规划算法要明显优于传统的线性回归算法以及最小二乘支持向量机算法。(3)针对钢铁企业面向工序的能源消耗预测问题,设计并开发了钢铁生产工序能源预测系统,该系统是一个综合性管理系统,可以准确有效预测未来一段时间的各个工序的介质消耗值,实现了介质平衡分析和指标管理功能。(4)针对热轧产品的能源介质消耗精细计量问题,设计并开发了热轧产品能耗精细计量与预测系统,该系统主要针对热轧产线的具体产品做能源介质消耗预测,为精细化的能源管理提供了有效决策支持。
[Abstract]:Iron and steel industry is the key industry of energy consumption in our country. The energy consumption is large and the utilization ratio of resources is low. In this paper, three data-driven prediction algorithms are proposed to predict the energy consumption for a period of time in the future by studying the energy consumption prediction problem of typical production lines and products in iron and steel enterprises. A process oriented and product-oriented energy consumption prediction system for iron and steel enterprises is developed in order to improve the energy utilization ratio and reduce the production cost of iron and steel enterprises. The main contents of this paper are as follows: (1) based on the research background of energy medium consumption prediction in the production process of a domestic iron and steel enterprise, through field investigation, the energy medium consumption prediction problem for typical processes in iron and steel enterprises is extracted. The problem of energy medium consumption prediction for hot rolled products is further refined. According to the dynamic characteristics of the energy medium prediction problem, three prediction algorithms are proposed: the traditional linear regression algorithm and the least square support vector machine algorithm. And approximate dynamic programming algorithm based on parameter model. For the approximate dynamic programming algorithm based on parameter model, random gradient method and recursive least square method are used to obtain the first and second order model parameters, and the prediction model is finally determined by comparing the value functions of the two parameter models. Numerical experiments are carried out on three algorithms. The experimental results show that the approximate dynamic programming algorithm based on parameter model is obviously superior to the traditional linear regression algorithm and least squares support vector machine algorithm. The energy prediction system of iron and steel production process is designed and developed. The system is a comprehensive management system, which can accurately and effectively predict the medium consumption value of each working procedure in the future. The function of medium balance analysis and index management is realized. Aiming at the problem of energy medium consumption fine measurement of hot rolled products, a fine energy consumption measurement and prediction system for hot rolled products is designed and developed. The system mainly makes energy medium consumption prediction for the specific products of hot rolling production line, and provides effective decision support for fine energy management.
【学位授予单位】：东北大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：F426.31;TP311.52
，

本文编号：1786033