项目名称: 基于动态数据与先验认知混杂驱动的高炉冶炼过程多尺度建模与优化
项目编号: No.61263014
项目类型: 地区科学基金项目
立项/批准年度: 2013
项目学科: 自动化技术、计算机技术
项目作者: 罗世华
作者单位: 江西财经大学
项目金额: 44万元
中文摘要: 建立新型控制系统,即时自动控制喷煤、风量、风温等并据此稳健操控高炉,是高炉能否最终实现闭环控制的关键所在。本项目以后续炉温趋向"理想炉温状态"为控制目标,以专家认知为先导,利用动态数据和尺度分析充分建模,构建喷煤、风量和风温等关键控制变量的在线控制模型。 首先:通过信息冗余、长期均衡等相关统计检验,结合动态主元分析等方法筛选出高炉系统建模的若干关键状态和控制变量,并据此构建高炉系统多变量重构相空间,完成系统非线性特征挖掘并将其用于改进炉温[Si]预报;其次,利用专家认知和截集模糊C均值混杂方法提取高炉节能高效"最佳炉温状态";最后,综合上述两方面的研究,利用动态数驱动和非线性特征量化值确定相关可加非线性系统(ANLS)的模型框架和参数,以解析炉温[Si]与喷煤、风量、风温等变量的非线性耦合关联,建立起高炉冶炼新模型,并研究建模信号尺度变化对模型结构、系统临界动态行为等方面的影响。
中文关键词: 动态数据驱动;先验认知;高炉;多尺度;
英文摘要: Successful instant control of coal injected (PM), cold blast temperature (CFW) and blast volume (FQ) etc. is the key to establish a satisfied blast furnace (BF) clsosed-loop control system. In this issue, after the optimum control centre of BF thermal state is identified and some expert knowledge is used, such control model would be establied based on hybrid integration of dynamic data and multi-scale analysis. Firstly, selects some key state or control variables which affect BF thermal state as input and output variables for new model by using some statistical methods such as information redundancy analysis, long-run equilibrium test and dynamic principal component analysis etc.. Then the multivariate phase space is reconstructed by using such key variables and the inherent characteristics of Mining System can be identified and used to improve the prediction of [Si] by the new topological isomorphism reconstruction space. Secondly, idetifies the optimum control centre of BF thermal state based on expert cognitive and fuzzy cut-set C-means clustering. Thirdly, taking the two aspects, the framework and parameters of the additive nonlinear systems (ANLS) can be determined based on dynamic data-driven and the value of nonlinear characteristics. After that, the nonlinear coupling relationship between the [Si] and PM
英文关键词: dynamic data-driven;prior information;blast furnace;muti-scale;