基于ELM遗传算法的氧化铝焙烧过程智能建模与控制系统研究
发布时间：2018-10-10 07:43
【摘要】：近十多年来,我国的工业发展取得了长足的进步,其中冶金工业的发展,对国家经济、社会的快速成长和国防科技建设的提升起到了极大的促进作用。氧化铝作为生产金属铝的原料,在铝冶炼工业中具有举足轻重的地位。目前,拜耳法是我国生产氧化铝所采用的主要方法之一,在该工艺过程中,氧化铝焙烧过程是影响氧化铝质量、生产能耗和生产成本的重要工段之一。利用智能化方法对焙烧过程进行建模,利用合适的算法进行焙烧的参数优化和控制研究是氧化铝生产工业技术创新的一个方向,是提高氧化铝质量的有效途径。本文以气态悬浮焙烧炉工艺为基础,采用改进粒子群(PSO)优化极限学习机算法(ELM)对氧化铝焙烧进行预测建模,利用遗传算法(GA)完成氧化铝焙烧工况参数的优化,设计基于DCS的氧化铝焙烧过程控制系统,通过BP神经网络PID控制器实现焙烧关键参数的精确控制,主要内容有:(1)针对焙烧过程建模困难的问题,分别采用BP神经网络、标准ELM和改进PSO优化ELM建立焙烧温度预测模型,对比发现,采用改进PSO优化方法相较于BPNN和标准ELM方法,在预测精度和泛化性能方面均有明显优势。(2)针对焙烧过程参数耦合严重,工况波动频繁的问题,利用遗传算法,建立氧化铝焙烧工况优化模型。以实际生产正常工况状态下焙烧温度稳定值(1070℃)为控制目标,寻找对焙烧温度影响较大的操作参数在技术指标范围内的最优组合,并以此为基础,建立优化工况数据库,在生产过程中,控制系统根据监控到的焙烧温度与设定值之间的偏差,从优化工况数据库中寻找最优工况组合,指导对应控制变量的实时调整,使得生产过程处于最优状态,避免人工设定的主观性和生产过程的误操作,减少不必要的能耗,稳定焙烧温度,提高氧化铝质量。(3)针对氧化铝焙烧过程自动化水平不足、生产和管理工作不完善的现状,设计基于DCS的氧化铝焙烧过程控制系统。采用BP神经网络PID控制器实现过程操作参数控制,以及生产过程的监控,合理配置生产资料,以提高生产效率,降低企业生产成本。(4)以氧化铝焙烧温度为例,设计氧化铝焙烧过程优化控制系统。在高级过程控制系统仿真平台上构建对象模型、虚拟执行机构、基础控制回路,进行仿真实验,结果表明系统可以很好的跟踪焙烧温度的设定值,验证了控制系统的可行性。
[Abstract]:In recent ten years, great progress has been made in China's industrial development, among which the development of metallurgical industry has played a great role in promoting the rapid growth of national economy and society and the promotion of national defense science and technology construction. Alumina, as the raw material of aluminum production, plays an important role in aluminum smelting industry. At present, Bayer process is one of the main methods of alumina production in China. In this process, alumina roasting process is one of the important sections which affect the quality, energy consumption and production cost of alumina. It is a direction of technological innovation in alumina production industry to model the roasting process by intelligent method and to optimize and control the roasting parameters by using the appropriate algorithm. It is an effective way to improve the quality of alumina. Based on the technology of gaseous suspension roaster, the prediction modeling of alumina roasting is carried out by using the improved particle swarm (PSO) optimization extreme learning machine (ELM) algorithm, and the optimization of operating conditions parameters of alumina roasting is accomplished by genetic algorithm (GA). The control system of alumina roasting process based on DCS is designed, and the precise control of the key parameters of roasting is realized by BP neural network PID controller. The main contents are as follows: (1) aiming at the difficult modeling problem of roasting process, BP neural network is adopted respectively. Standard ELM and modified PSO optimization ELM are used to establish the calcination temperature prediction model. The comparison of the improved PSO optimization method with BPNN and standard ELM method is made. It has obvious advantages in prediction accuracy and generalization performance. (2) aiming at the problems of severe coupling of calcination process parameters and frequent fluctuation of operating conditions, the optimization model of alumina roasting condition is established by genetic algorithm. Taking the stable value of calcination temperature (1070 鈩，
本文编号：2261205
[Abstract]:In recent ten years, great progress has been made in China's industrial development, among which the development of metallurgical industry has played a great role in promoting the rapid growth of national economy and society and the promotion of national defense science and technology construction. Alumina, as the raw material of aluminum production, plays an important role in aluminum smelting industry. At present, Bayer process is one of the main methods of alumina production in China. In this process, alumina roasting process is one of the important sections which affect the quality, energy consumption and production cost of alumina. It is a direction of technological innovation in alumina production industry to model the roasting process by intelligent method and to optimize and control the roasting parameters by using the appropriate algorithm. It is an effective way to improve the quality of alumina. Based on the technology of gaseous suspension roaster, the prediction modeling of alumina roasting is carried out by using the improved particle swarm (PSO) optimization extreme learning machine (ELM) algorithm, and the optimization of operating conditions parameters of alumina roasting is accomplished by genetic algorithm (GA). The control system of alumina roasting process based on DCS is designed, and the precise control of the key parameters of roasting is realized by BP neural network PID controller. The main contents are as follows: (1) aiming at the difficult modeling problem of roasting process, BP neural network is adopted respectively. Standard ELM and modified PSO optimization ELM are used to establish the calcination temperature prediction model. The comparison of the improved PSO optimization method with BPNN and standard ELM method is made. It has obvious advantages in prediction accuracy and generalization performance. (2) aiming at the problems of severe coupling of calcination process parameters and frequent fluctuation of operating conditions, the optimization model of alumina roasting condition is established by genetic algorithm. Taking the stable value of calcination temperature (1070 鈩，
本文编号：2261205
本文链接：https://www.wllwen.com/kejilunwen/huagong/2261205.html
