神经网络PID参数自整定的应用研究

PID控制是工业控制中应用最为广泛的控制方法,但在实际应用中,其参数整定仍未得到较好的解决.本文把神经网络技术应用在PID控制中,采用BP网络对被控对象在线辨识,利用神经元自适应PID在线调整参数,构造神经网络PID自整定控制器.通过在实际交流变频调速系统上的实验表明,当突然加、减负载时,神经网络PID控制与传统PID控制相比,具有恢复时间短、超调量小等特点.

Application Research on Self-adjusting of PID Parameters Based on Neural Networks

PID controllers are most commonly used in industries,particularly in the process industries.However,up to now,there are no satisfactory solutions about the adjustment of PID parameters.NN(Neural Network) technique is used in PID control because it could change parameters itself on line.Experiments of AC speed-adjustable system with variable frequency show that the restoration time of rotational speed in neural network control system is much shorter than that in PID control system and the overshooting during the adjustment period is also weakened when the loads are suddenly increased or decreased.