温室温度控制系统的神经网络PID控制

建立温室温度控制系统的数学模型。针对温室温度控制系统存在的大滞后、大惯性等问题,考虑到常规PID控制器自适应能力差、鲁棒性不强等缺陷，提出采用将具有较强的自组织、自学习和自适应能力的径向基神经网络与常规PID相结合构成RBF-PID控制策略，自适应调整PID控制器的参数。在该控制策略中，采用RBF神经网络辨识器实现温度控制系统的Jacobian矩阵信息在线辨识，对 RBF-PID控制器控制参数在线自整定。研究结果表明：RBF-PID控制器可使温室温度控制系统动态响应快、鲁棒性强、稳态精度高、超调量小、抗扰动能力强，具有良好的控制效果。

Neural networks based on PID control for greenhouse temperature

A mathematical model of greenhouse temperature was established. Confronted with problem of greenhouse temperature control existed in conventional PID controller such as big inertia, big lag, bad adaptive ability and robustness, and other defects, a kind of intelligent PID controller based on RBF neural network with adaptive ability and self-leaning and self-organization was proposed to adjusted the parameters of PID controller. It identified the Jacohian matrix of feedback system by the RBF neural network and acquired on-line tuning the parameters of PID controller. The experimental results demonstrated that a high performance was achieve with little overshoot, steady-state precision and disturbance rejection ability.