基于神经网络辨识改进的自适应PID控制器

针对工业生产过程具有时变、非线性、不确定和难以建立精确数学模型的特性，提出了用神经网络辨识被控对象，并针对常规PID控制器存在的设计复杂、计算量大、控制精度较差和参数值实时更新复杂等问题，提出采用基于BP网络辨识改进的自适应PID控制器。通过仿真与实验表明，神经网络辨识模型能较好的辨识被控对象的输出特性，同时该控制器的控制性能优于传统的PID控制器。

Adaptive PID Control Based on the Improved Identification of Neural Network

During industrial producing process,it usually has time-varying、nonlinearity、 uncertainty、difficulty in establishing the precise mathematical model,controlled plant is identified by neural network in this paper.And specially for the problems of conventional PID controller,it is complexity of design、big calculated amount、low-precision control accuracy、complixity in real time update the parameter value.The paper purposes a predictive mathematical model of controlled plant established by neural network.The simulation and the expertment indicates that model identification of neural network is able to identify the output of controlled plant better and the control performance of this controller is better than the traditional PID controller.