山梨醇结晶温控系统的研究  

Reseach on Crystallizing Process Temperature System of the Sorbitol Production

作　　者：郑晓茜[1] 赵方[1] 邵帅飞 

机构地区：[1]郑州职业技术学院电气电子工程系,郑州450121 [2]机械工业第六设计研究院,郑州450007

出　　处：《食品工业》2012年第7期127-130,共4页The Food Industry

摘　　要：山梨醇广泛应用于食品行业,针对山梨醇结晶温控系统存在时变大滞后特性,研究了一种融合遗传算法和神经网络的PID控制器。该控制算法采用RBF网络在线辨识被控对象,先利用遗传算法优化神经网络的初始权值,再结合神经网络所具有的自学习和任意非线性表达能力,用BP神经网络自整定PID参数,对结晶过程温度进行控制。仿真结果表明:该控制器提高了系统的控制性能,具有很强的自适应性和鲁棒性,满足山梨醇生产的要求。Sorbitol is widely used in the food industry,as the time delay and inaccurate model exist in the crystallizing process temperature system of the sorbitol production.A new type controller based on genetic algorithm and neural network was presented.RBF neural network was utilized to identify the controlled plant on line,and the genetic algorithm was utilized to optimize the initial weights of BP neural network,combing with the self-learning ability and describing any non-linear ability of neural network.A BP neural network was utilized to achieve PID parameters self-adjustment to control the temperature of crystallizing process.Finally,the system had been simulated and compared with the traditional method.The simulation results showed that the effectiveness of this controller was superior to the traditional control algorithm;it had higher adaptability and robustness and could satisfy the requirement of sorbitol production.

关 键 词：BP神经网络 RBF辨识 遗传算法 PID 参数自整定 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]