基于分解率预报的连续碳酸化分过程模糊专家控制技术及应用

【作者】 何卉；

【导师】 桂卫华；

【作者基本信息】 中南大学 ， 控制理论与控制工程， 2007， 硕士

【摘要】 铝酸钠溶液的连续碳酸化分解是烧结法氧化铝生产中的关键工序之一，其分解率的控制对于提高产品的产量和质量，减少环境污染具有非常重要的意义。本文在深入分析连续碳酸化分解过程工艺的基础上，提出了分解率优化控制系统的结构，包括分解率优化设定模块、分解率智能预报模块和分解率模糊专家控制模块。在此基础上，详细阐述了分解率智能控制策略。分解率智能预报模块采用神经网络和专家系统，对难以实时测量的分解率进行智能建模。该模型由数据采集与处理模块、决策支持模块、预报模块、在线校正模块以及效果评价模块组成。在很大程度上解决了关键变量测量困难的问题，为系统的优化控制奠定了良好的基础。分解率模糊专家控制模块采用模糊控制技术结合专家控制技术，实现了分解率的优化控制。模糊专家控制模块包括三个回路：进料量控制回路、4~#槽通气量控制回路和5~#槽通气量控制回路。根据供料槽液位的改变，采用专家控制策略，改变进料阀门的开度，保证进料量的稳定，构成进料量控制回路；同时，根据分解率的预报值与设定值之间的偏差，采用模糊专家控制策略，改变4~#槽、5~#槽通气阀的开度，保证分解率的合格。本文提出的模糊专家智能控制策略在中铝氧化铝厂的连续碳酸化分解工序的实际生产中得到了应用。现场运行情况表明，这种控制策略符合工业实际，建立的软件系统稳定可靠，有效的控制了分解率，提高了产品产量、质量，创造了显著的经济效益。更多还原

【Abstract】 Continual carbonation decomposition (CCD) of sodium aluminates solutions is one of the key processes in the process of alumina production by sintering process. The control of the resolution ratio is significant for improving the output and quality of product, decreasing air pollution.Based on analysis of the mechanic theory and characteristic of CCD, the structure of control system is established and the intelligent control strategy of resolution ratio is presented. The system consists of three modules, such as optimal setting model of resolution ratio, intelligent prediction model of resolution ratio and fuzzy expert control model of resolution ratio.Based on neural network and expert system, the intelligent prediction model of resolution ratio is developed, which consists of five modules, such as data collection and pretreatment, decision-making support, prediction, on-line modification and effect estimate. It has applied in the prediction of resolution ratio, which is hard to get timely. It has proved that the prediction model solves the problem of the measurement difficulty of the key variables to a great extent. It also provides good foundation for optimal control of the system.Combining fuzzy control and expert system, the fuzzy expert control model of resolution ratio is used for optimal control of resolution ratio. The control model has three loops, which are inlet amount loop, dioxide carbon flow loop of 4~# tank and dioxide carbon flow loop of 5~# tank. According to the variation in liquid level of gravity-feed tank, expert control strategy is used to adjust the inlet valve opening to ensure the steady inlet amount. At the same time, according to the error between target and prediction, fuzzy-expert control strategy is used to adjust dioxide carbon valve opening to ensure the proper resolution ratio.The intelligent control based on fuzzy-expert strategy is used in the CCD process in Zhongzhou Aluminum Company. The validityof this strategy is proved by the running results. The software is stable and reliable; the resolution ratio is control effectively and the output and quality of product are improved.更多还原