连续碳酸化分解过程的混合智能控制  

Hybrid intelligent control of continual carbonation decomposition process

作　　者：胡志坤[1,2] 姜斌[1] 尹林子[2] 陈晓龙[2] 谢永芳[3] 

机构地区：[1]南京航空航天大学自动化学院,江苏南京210016 [2]中南大学物理与电子学院,湖南长沙410083 [3]中南大学信息科学与工程学院,湖南长沙410083

出　　处：《江苏大学学报（自然科学版）》2014年第3期319-323,338,共6页Journal of Jiangsu University：Natural Science Edition

基　　金：国家自然科学基金资助项目(61273159;61105080;60904077);国家"863"计划项目(2014AA041803);中国博士后科学基金资助项目(2012M511752)

摘　　要：为了解决连续碳酸化分解过程中分解率梯度与末槽分解率无法用精确数学模型描述的问题,以控制合适的分解梯度与合格的末槽分解率为目标,将专家控制与预测控制策略相结合,提出连续碳酸化分解过程智能控制模型.根据人工经验建立了专家知识库,一类知识库用来处理工况波动和仪表故障,另一类知识库根据各槽料浆成分信息,给出各个阀门开度的控制输出量,来实现正常工况情况下的分解率的稳定控制;并利用神经网络建立的预测模型预测系统下一时刻分解率输出,用以对专家控制模型输出反馈修正.应用结果表明:该方法有效地克服了大滞后因素的影响,分解过程的优化控制分解率合格率提高了4%,平均分解率提高了0.95%.Due to the long time-delay and complex industrial process of continual carbonation decomposition, the optimal control was difficult to be described by mathematical models. To control the optimal resolution ratio and the last decomposition ratio, the intelligent control system with considerations of expert control and predictive control strategies was proposed for continual carbonation decomposition process of sodium aluminate solutions. The principle knowledge and expert experience was applied to establish ex- pert knowledge bases to deal with fluctuations in operating conditions and equipment faults, and to output the fourth and the fifth control valves to realize stable control in normal operational conditions by slurry composition for every cell. A neural network predicting model was set up to forecast the next output of system and feedback to modify the output of expert control model. The practical results show that eligible ratio of decomposition ratio is increased by 4% with increased average value of decomposition ratio by 0.95%. The influence of long time-delay is conquered effectively, and the process of continual carbona- tion decomposition is optimally controlled by the proposed method.

关 键 词：铝酸钠溶液 连续碳酸化分解 专家控制 预测控制 神经网络 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]