基于动态概率分布的复杂流程工业混合过程鲁棒优化方法研究

项目名称： 基于动态概率分布的复杂流程工业混合过程鲁棒优化方法研究

项目编号： No.61203136

项目类型： 青年科学基金项目

立项/批准年度： 2013

项目学科： 自动化学科

项目作者： 孔玲爽

作者单位： 湖南工业大学

项目金额： 23万元

中文摘要： 混合过程是冶金、化工、钢铁等许多复杂流程工业中一道非常重要的工序，其混合操作的合理性不仅直接关系到下一个工序生产的稳定，而且对整个企业资源能源的消耗及最终产品质量产量的提高都有重要影响。复杂混合过程普遍具有生产用料复杂、工艺指标多、强耦合、非线性、在线检测困难和过程机理不明确等复杂特点，特别是多种混合原料质量参数表现出不同程度的随机性，严重影响了传统优化方法的实施效果。本项目结合随机规划、鲁棒优化和预测控制的思想，研究一种基于随机参数概率分布函数的复杂混合过程鲁棒优化方法，给出基于实时趋势预测和随机补偿策略的随机参数动态概率分布统计方法，建立融合质量预测模型和鲁棒评价指标的鲁棒随机优化模型，提出一种基于高效随机仿真技术的两层智能鲁棒优化算法，形成一套能有效克服参数随机性影响的工业过程鲁棒优化理论和方法，研究成果的工业应用可有效提升我国流程工业的生产水平。

中文关键词： 混合过程；不确定；时间序列；随机优化；鲁棒实时优化

英文摘要： The blending process is a very important procedure of the production in many complex process industries such as metallurgy, steel and chemical industry, which not only directly relates to the production stability of downstream processes but also make great influence on the energy consumption and the productivity of whole enterprise. However, a lot of complexity，such as complex materials, various kind of indices, strong coupling, nonlinearity,great difficulty of online detection and unknown process mechanism and so on, generally exists in these blending processes, so as to it is very difficult for classical methods to effectively realize the optimization control of blending processes. In the project, based on stochastic programming, robust optimization and predictive control, the robust optimization method with dynamic probability distribution is studied for the blending operation of complex process industry. The statistical method of the dynamic probability distribution is proposed by combining trend prediction and random compensation strategy. The robust stochastic optimization model with stochastic parameters, quality prediction model and robust evaluation indices is constructed. And the two-level intelligent algorithm with an efficient stochastic simulation technique is put forward to effectively solve it. F

英文关键词： blending process；uncertainty；time series；stochastic optimization；robust real-time optimization