基于神经网络的多变量模糊自整定PID控制器

针对脉冲ＴＩＧ焊动态过程控制提出了一种基于ＢＰ神经网络的多变量模糊自整定ＰＩＤ控制器。该控制器利用神经网络在线学习具有多变量耦合，非线性及不确定性的复杂的焊接动态过程的控制规则，实现ＰＩＤ参数的自动整定。仿真实验结果表明该控制器不仅具有模糊控制的简单，有效的非线性控制作用。还具备了神经网络的学习与适应能力，以及ＰＩＤ控制的普遍适用性。

A Multivariable Fuzzy Self-Tuning PID Controller Based on Neural Network

A multivariable fuzzy self-tuning PID controller absed on BP neural network is presented for the dynamic process control of pulse TIG welding.The controller can learn,on line,the control rules of the complex dynamic process with multivariable coupling,nonlinear and uncertainty,so that the PID parameters are tuned automatically.The results of experiment show that the controller possesses not only the simplicity and the nonlinear control ability of fuzzy control,but also the learning and adaptive functions by using neural network.Furthermore,the controller is as widely adaptive as PID regulation.