基于BP神经网络的PID控制器在温控系统中的应用

Application of PID Controller Based on BP Neural Network in Temperature Control System

工业中的温控对象普遍具有大时滞、非线性和时变性等特点，采用传统的PID控制器难以取得满意的效果。基于BP神经网络的PID控制器具有逼近任意非线性函数的能力，能实现对PID控制器的参数Kp，KI，KD的实时在线整定，使系统具有更好的鲁棒性和自适应性，其输出也可以通过在线调整达到预期的控制精度，适用于温控系统。实验结果表明，该控制器具有抗干扰能力强、鲁棒性好等特点。

In the industrial practice, it is difficult to achieve the satisfactory results with traditional PID temperature controller for those plants which are characteristics of large time lag, nonlinearity and time variation. A PID controller based on BP neural net- work has the capability in approaching any nonlinear functions, and can real-timely realize the online setting of the PID controller＇s parametersKp, KI, KD.With the optimized parameters, the control system can be of better robustness and self-adaptability, and the out- put of the system can reach the expected control precision through the online tuning. Thus it is suitable for temperature control. The experimental result shows that the controller is of a strong anti-jamming capability and a better robustness in application.