基于软测量的醋酸精馏过程串级预测控制策略

A cascade predictive control strategy based on soft sensing of acetic acid distillation process

针对醋酸精馏控制中,产品成分无法在线检测并且对产品质量采用温度间接控制存在控制精度低的问题,提出一种基于在线更新小波核函数极限学习机软测量的DMc预测控制策略,其中,在线更新的小波核函数极限学习机软测量实现了塔底醋酸浓度的实时检测,仿真结果表明,在线更新模型的预测精度比离线模型提高52%。将上述在线更新的软测量应用于塔底醋酸浓度闭环预测控制系统中,实现对塔底醋酸浓度的直接质量控制,该系统采用DMc作为醋酸浓度控制器,其输出量作为再沸器蒸汽流量控制器的设定值,与再沸器流量控制构成串级调节系统。控制系统仿真结果表明,该软仪表具有良好的在线预测性能,预测控制系统控制精度高、可以实现产品质量的卡边控制。

For distillation products are hard to be measured online and the performance of indirectly temperature control can＇t reach a high level, a Dynamic Matrix Control （DMC） strategy based on online update wavelet kernel function extreme learning machine soft-sensor is proposed. The online update soft-sensor is used to obtain the concentration of acetic acid in the bottom of the column. According to the results of the simulation, the online model prediction accuracy is 52 % higher than outline model. Then the soft-sensor is used to obtain the concentration of acetic acid in the closed predictive control system. DMC is used as the master controller for the acetic acid concentration. A PID controller is used as a slave controller for the reboiler vapor flow. The two controllers constituted a cascade control system. According to the results of the simulation, the predictive ability of the soft-sensor is good, and can be used to realize the bounder control of concentration of acetic acid in the bottom of the column.