基于BP神经网络的改进增量式PID暖通控制器设计

为了克服传统PID控制在暖通空调系统应用中超调量大、控制精度低的缺陷,提出了一种基于BP神经网络的PID控制器设计方法.利用BP神经网络具有很强的学习能力、任意逼近非线性能力、自适应性和鲁棒性等特点,将BP神经网络与PID控制结合,实现了PID的3个控制参数的在线自整定.仿真结果表明,该方法可以显著改善系统的动态性能和控制精度,实现了PID控制参数的在线动态调整,避免了由于系统模型和结构参数变化导致的控制效果不稳定.

Design of BP Neural Network PID controller to HVAC system

Aiming at solving the existing problems that the traditional PID controller applied to heating ventilation and air conditioning system has such as large overshoot and low accuracy,an improved PID controller designing approach based on BP（Back Propagation） Neural Network are proposed.The solution takes advantage of the characteristics of BP Neural Network,such as strong learning ability,arbitrary nonlinear approximation ability,self-adaptability and robustness,combines the BP Neural Network algorithm and PID control algorithm,and realizes the on-line self-tuning of three PID parameters.The simulation results demonstrate that the method significantly improves system＇s dynamic performance and control accuracy,ensures the dynamic adjustment of control parameters,and avoids the unstable control effects caused by system model and structure parameters changing.