SIA OpenIR  > 数字工厂研究室
钢铁企业副产煤气的产消量预测及优化调度方法研究
Alternative TitleResearch on byproduct gas generation and consumption forecasting and optimization scheduling in iron and steel plant
孙雪莹
Department数字工厂研究室
Thesis Advisor胡静涛
Keyword副产煤气系统 预测 极限学习机 优化调度 不确定性
Pages106页
Degree Discipline检测技术与自动化装置
Degree Name博士
2018-11-28
Degree Grantor中国科学院沈阳自动化研究所
Place of Conferral沈阳
Abstract

钢铁工业是现代化建设的基础产业,对国民经济的发展起到重要的作用。同时,钢铁企业也具有高耗能和高排放的特点。钢铁企业的节能减排是实现国家绿色可持续发展的迫切需求。副产煤气是钢铁生产过程中产生的重要副产能源,煤气的充分利用对降低钢铁企业的能耗具有十分重要的意义。本研究以钢铁企业副产煤气系统为研究对象,对副产煤气的产消量的预测和优化调度进行了深入研究。考虑钢铁企业副产煤气系统工艺复杂、产消量波动较大的特点,本文对副产煤气的产消量点预测和区间预测方法,以及考虑预测不确定性的副产煤气优化调度策略展开了相关理论和实践研究,具体内容如下:(1)考虑工况变化的副产煤气产消量点预测方法研究。针对副产煤气的产消量受工况变化影响较大,工况发生变化导致预测模型精度降低的问题,提出了基于自适应遗忘因子极限学习机的煤气产消量点预测方法。通过预测误差的在线反馈调整模型参数,提高了模型的对工况变化的适应能力。工业实际数据的实验表明所提出的方法提高了煤气产消量的预测精度,其结果为副产煤气的优化调度提供了基础。同时,针对副产煤气的产消量数据存在较多缺失,影响预测效果的问题,提出了基于改进模糊C均值的数据填补方法,采用极限学习机和粒子群算法对模糊C均值的参数选取进行优化,提高了数据填补的精度。(2)考虑预测偏差的副产煤气产消量区间预测方法研究。针对现有的预测区间评价指标的不足,提出了考虑预测偏差的极限学习机区间预测模型。该模型利用极限学习机直接预测区间的上限和下限。考虑区间偏差信息,给出了新的预测区间的综合评价指标,并将其作为极限学习机的优化准则。此外,为了提高模型的训练速度和精度,提出基于预训练的两步训练方法。与现有的区间预测方法相比,所提出的模型结构简单,训练速度快,构建的预测区间质量高,更能适应副产煤气的产消量在线预测和调度的需求。(3)考虑不确定性的副产煤气系统模糊优化调度方法研究。针对副产煤气的产消量预测存在不确定性的问题,分析了产消量的不确定性对副产煤气系统优化调度的影响,基于模糊机会约束规划方法构建了副产煤气系统优化调度模型。利用副产煤气的点预测和区间预测信息,将副产煤气的产消量表达为模糊变量,引入置信度,将模型中的模糊约束转化为确定性的约束,从而使副产煤气模糊优化模型转化为可以求解的确定性的混合整数线性规划模型。进一步地,为了评估不同置信度水平下得到的调度方案的风险,首次提出了副产煤气系统风险成本计算方法。为调度人员选取合适的置信度水平提供了重要的参考依据。钢铁企业的实际案例表明所提出的煤气系统模糊优化调度方法与现有的确定性方法相比,能够有效地处理煤气预测的不确定问题,降低煤气不足和放散的风险,为调度人员提供更加全面的调度方案。

Other Abstract

Iron and steel industry is a pillar industry of national economy. During its high economy-consuming, energy conservation and emission reduction of iron and steel enterprises is of great significance to the country for achieving green development. In the iron and steel production process, by-product gas is produced and consumed by many reactors and it plays an important role as a secondary energy source. The efficient use of by-product gas is crucial to the energy conservation and consumption reduction of iron and steel enterprises. Therefore, by considering the complexity and large fluctuation of by-product gas system, our research focus on by-product gas production and consumption prediction, both in point prediction and interval prediction. On the other hand, we are also engaged in scheduling optimization in view of prediction uncertainty. The specific contents of relevant theoretical and practical studies are as follows: (1) Research on byproduct gas generation and consumption forecasting method considering the change of working conditions. By-product gas production and consumption can be greatly affected by the change of working conditions. The accuracy of the prediction model can be reduced when working conditions change. To solve this problem, an online prediction method based on adaptive forgetting factor extreme learning machine is proposed. Parameters of the model are adjusted by the online feedback of the prediction error. This leads to high adaptability to the change of working conditions of the prediction model. Experiments on industrial data show that the proposed method improves the prediction accuracy of gas production and consumption. Besides, data missing is inevitable in the by-product gas flow data, which can affect prediction effect. To solve the problem, the missing values imputation method based on fuzzy C-means clustering is introduced to our work. Parameters of FCM are optimized by the particle swarm optimization method. During the optimization process, extreme learning machine is employed to compute reference values. By combining the supervised and the unsupervised learning methods, data cleaning can be performed. (2) Research on for byproduct gas generation and consumption interval prediction method considering the prediction deviation. To quantify the uncertainties of byproduct gas forecasting, an interval prediction model based on twin extreme learning machine is proposed. As a result, the upper and lower bounds of the model of the interval are directly predicted. Considering the interval deviation information, a comprehensive evaluation index of the new prediction interval is given and used as the optimization criterion of the twin extreme learning machine. In addition, in order to improve the training speed and accuracy of the model, a two-step training method based on pre-training is proposed. Compared with the traditional interval prediction method, the proposed model has advantages in simple structure, fast training speed, and high-quality prediction interval. The proposed approach is more suitable for the online prediction and scheduling of the by-product gas production and consumption. (3) Research on fuzzy optimal scheduling method for byproduct gas system considering the uncertainty. To solve the problem of uncertainty in the production and consumption forecast of by-product gas, the influence of the uncertainty of production and consumption on the optimal scheduling of the by-product gas system is analyzed, and a new by-product scheduling model is constructed based on the fuzzy chance constrained programming method. By taking the point prediction and interval prediction information into consideration, the production and consumption of by-product gas can be expressed as a fuzzy variable. Confidence is introduced to transform the fuzzy constraint to a deterministic constraint. Afterwards, mixed integer linear programming is employed to solve the problem. Furthermore, in order to evaluate the risk of the scheduling scheme obtained at different confidence levels, the risk cost measurement method of the by-product gas system is proposed for the first time. It provides an important reference for the dispatcher to select the appropriate level of confidence. The actual case of the iron and steel enterprise shows that the proposed fuzzy optimization scheduling method of the gas system can effectively deal with uncertainties of gas prediction. The proposed method provides an approach to reduce the risk of gas shortage and emission, and also provides a more comprehensive scheduling solution for dispatchers.

Language中文
Contribution Rank1
Document Type学位论文
Identifierhttp://ir.sia.cn/handle/173321/23636
Collection数字工厂研究室
Recommended Citation
GB/T 7714
孙雪莹. 钢铁企业副产煤气的产消量预测及优化调度方法研究[D]. 沈阳. 中国科学院沈阳自动化研究所,2018.
Files in This Item:
File Name/Size DocType Version Access License
钢铁企业副产煤气的产消量预测及优化调度方(2039KB)学位论文 开放获取CC BY-NC-SAApplication Full Text
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[孙雪莹]'s Articles
Baidu academic
Similar articles in Baidu academic
[孙雪莹]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[孙雪莹]'s Articles
Terms of Use
No data!
Social Bookmark/Share
All comments (0)
No comment.
 

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.