项目名称: 基于支持向量机和群智能的煤制甲醇合成过程建模及优化方法研究
项目编号: No.21366017
项目类型: 地区科学基金项目
立项/批准年度: 2014
项目学科: 化学工程及工业化学
项目作者: 王建国
作者单位: 内蒙古科技大学
项目金额: 25万元
中文摘要: 用支持向量机和群智能优化方法进行煤制甲醇合成过程的数学建模及优化研究,探索甲醇合成过程中工艺参数与甲醇产率之间的机理性问题。本项目以甲醇合成数据中的组分含量、温度、压力、氢碳比、空速等过程工艺参数为输入,甲醇产率为输出,针对变量之间的时变性和非线性,研究具有在线学习能力的支持向量机动态建模方法;针对变量之间的强耦合性,研究基于支持向量机的规则抽取方法;最后结合动态模型和提取的规则,研究基于多目标粒子群的工艺参数优化方法,实现生产过程中对甲醇产率的预测、监测及参数优化。此研究对揭示甲醇合成生产规律、提高生产效率等方面有着重要的科学意义和推广价值,为化工过程的建模及优化提供了一条新途径, 同时也拓展了数据挖掘应用的新领域。
中文关键词: 醇合成过程;数据降噪;增量建模;规则抽取;群智能优化
英文摘要: Using support vector machines and swarm intelligence optimization methods to build and optimize the mathematical model in coal methanol synthesis process, and discover the mechanism problem of the relationship between the process parameters and methanol productivity. This project treat the methanol synthesis data: constituent content, temperature, pressure, hydrogen-carbon ratio and airspeed parameters as the input, and methanol product rate as the output. Aiming at the time-varying and nolinear of varibles, develop a support vector machine dynamic modelling method with the online learning ability; aiming at the strong coupling of varibles, develop the extraction rules method based on support vector machines for the dynamic model; finally, with the dynamic model and extracted rules, we develop a particle swarm based process parameters optimization method, achieve to predict and monitor the methanol product rate, optimize the parameters in the manufacture process. The research is of great significance and generalization in understanding of the production law, improving the production efficiency. It provide a new approach for chemical process modeling and optimization, extend a new field of data mining application.
英文关键词: methanol synthesis process;data noise reduction;incremental modeling;;rule extraction;swarm intelligence optimization