基于混沌灰狼算法的氧化铝焙烧质量预测与优化控制策略研究

发布时间：2018-07-13 17:38

【摘要】：针对我国铝土矿的特点,生产氧化铝主要采用拜耳法、烧结法和混联法,在拜耳法中焙烧过程是氧化铝生产必不可少也是最后一个过程,对氧化铝产品质量有着非常重要的影响。该过程流程长、检测大滞后、非线性强且工艺机理复杂,设备之间关联严重,工况变化大如原料成份波动频繁、设备结疤等问题,传统的策略难以实现操作优化以实现自动控制和优化运行的生产目的。因此,焙烧过程的重要性及其复杂性,以及现代化工业发展对自动化提出的要求使得焙烧过程控制系统日益得到人们更多的重视。本文以广西某铝厂氧化铝焙烧工序为背景,开展了氧化铝生产焙烧过程质量预测及综合工况优化控制的相关研究,取得了一定的研究成果。本文主要内容归纳如下:(1)首先,通过对氧化铝焙烧过程工艺的机理分析,找到影响质量指标的因素,并分析了焙烧工艺的特点,采用五点三次方法对现场数据进行滤波,并采用相关性分析和箱线图分析方法剔除现场数据的异常值。(2)结合机理分析采用主元分析和相关性分析等数据处理方法,选取了氧化铝质量指标软测量模型输入变量,采用混沌灰狼算法优化在线序贯极限学习机的初始权值和偏差,建立氧化铝质量预测模型。利用工业生产实际数据进行实验验证,本文所建立的软测量模型具有较高的精度和泛化能力,实现了氧化铝质量指标的有效估计,并为焙烧过程的优化控制创造了条件。(3)在上述研究工作基础上,进一步分析了目前焙烧工序操作参数控制的现状和存在的问题。在保证氧化铝质量指标合格的前提下,建立氧化铝焙烧过程综合工况操作优化控制模型,利用带约束的混沌灰狼优化算法进行模型求解而获得优化后的操作参数,即焙烧主炉温度、ID风机功率、下料量等。通过实际生产数据进行仿真实验,结果表明,该优化控制模型效果良好。(4)设计了氧化铝焙烧过程综合工况操作优化控制系统,该系统利用以太网和工业电缆与设备层和控制层进行连接,通过内部协议转换实现仿真数据交互,以实现控制系统的功能化和信息化。实验结果表明,依据氧化铝生产过程中各参数的实时变化,该综合工况优化控制系统可以得到基础控制回路的优化设定值,并将氧化铝质量指标及其相应的关键操作参数控制在目标范围内,保证氧化铝生产焙烧过程的正常运行。
[Abstract]:According to the characteristics of bauxite in China, the alumina production mainly adopts Bayer process, sintering process and mixing method. The calcination process in Bayer process is the necessary and final process of alumina production. It has a very important effect on the quality of alumina products. The process has long process, long detection lag, strong nonlinearity and complex process mechanism, serious correlation between equipments, large variation of working conditions, such as frequent fluctuation of raw material composition, scarring of equipment, etc. The traditional strategy is difficult to realize the operation optimization in order to realize the production purpose of automatic control and optimal operation. Therefore, the importance and complexity of roasting process, as well as the requirement of automation put forward by modern industrial development, make more and more attention to roasting process control system. Based on the alumina roasting process of an aluminum factory in Guangxi, the quality prediction of alumina roasting process and the optimization control of comprehensive working conditions were studied in this paper, and some research results were obtained. The main contents of this paper are summarized as follows: (1) firstly, by analyzing the mechanism of alumina roasting process, the factors affecting quality index are found, and the characteristics of roasting process are analyzed, and the field data are filtered by 5.3 times method. Correlation analysis and box diagram analysis are used to eliminate outliers of field data. (2) combined with mechanism analysis, the input variables of soft sensor model of alumina quality index are selected by using principal component analysis and correlation analysis and other data processing methods. The chaos gray wolf algorithm is used to optimize the initial weights and deviations of the online sequential limit learning machine, and the quality prediction model of alumina is established. The soft sensor model established in this paper has high precision and generalization ability, and realizes the effective estimation of alumina quality index. It also creates conditions for optimal control of roasting process. (3) on the basis of the above work, the present situation and existing problems of operation parameter control in roasting process are further analyzed. On the premise of ensuring the quality index of alumina, the optimal operation control model of alumina roasting process is established, and the optimized operation parameters are obtained by solving the model by using the chaotic gray wolf optimization algorithm with constraints. That is, roasting main furnace temperature and ID fan power, feed volume, etc. The simulation results of actual production data show that the optimal control model is effective. (4) the operation optimization control system of alumina roasting process under integrated operating conditions is designed. The system uses Ethernet and industrial cable to connect with device layer and control layer, and realizes simulation data interaction through internal protocol conversion, so as to realize the function and information of the control system. The experimental results show that according to the real time change of the parameters in the alumina production process, the optimal control system can get the optimal setting value of the basic control loop. The quality index of alumina and its key operating parameters are controlled within the target range to ensure the normal operation of alumina production and roasting process.
【学位授予单位】：广西大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TQ133.1;TP18

【相似文献】