多温区电加热炉的PID神经网络控制  

PID Neural Network Control for Electric Heating Furnace with Multiple Temperature-zone

作　　者：袁朝辉[1] 张慧[1] 

机构地区：[1]西北工业大学自动化学院,陕西西安710072

出　　处：《计算机仿真》2010年第12期176-179,185,共5页Computer Simulation

摘　　要：研究温度控制对提高产品品质意义较大。多温区电加热炉具有强耦合、时变性等特点,针对多温区电加热炉要求动态品质好、控制精度高等特点,采用将PID控制融合入神经元网络的方法,比例元、积分元和微分元的存在,使得PID神经网络的控制系统响应快、超调小、无静差。采用改进的反向传播学习算法,在权值调整算法式中加入阻尼项,提高了算法收敛速度和学习速率,又不易产生振荡。在MATLAB环境下,用S函数编写算法进行仿真。仿真结果表明,PID神经网络达到较好的解耦效果,提高系统的抗干扰能力,改善了系统的动态品质,使多温区电加热炉温度控制系统的性能得到改善。Because multiply temperature-zone electric furnaces have the characteristics of strong coupling,time-varying,and demand good dynamic quality and high control precision,the method of integrating PID which introduces proportion,integral and derivation into neural network control ensures the furnace to track the input signal.Improved back-propagation learning algorithm adds damping to weight adjustment function to improve the convergence speed and learning rate,and this algorithm is also not prone to produce oscillations.In MATLAB environment,simulink simulation is presented with S-function.Simulation results show that PID Neural Network achieves better decoupling result,improves the anti-jamming capability and dynamic quality of the system,limits the steady-state error and meets the precision needs of the control system.

关 键 词：神经元网络 改进的反向传播学习算法 多温区电加热炉 炉温控制系统 

分 类 号：TP13[自动化与计算机技术—控制理论与控制工程]