CAJ | 학술논문

针对汽油机怠速工况的非线性、时变性和不确定性，传统PID控制难以获得理想控制效果的问题，提出一种基于模糊神经网络的PID控制方法，将模糊控制、神经网络与PID控制相结合，给出了BP神经网络模型，采用3层前向网络，动态BP算法，利用神经网络的自学习和自适应能力，实时调整网络的权值，改变PID控制器的控制参数，整定出一组适用于PID控制的kp、ki、kd参数，实现汽油机怠速PID控制的自适应和智能化控制．实验结果表明，采用BP神经网络整定的PID控制，控制响应快、鲁棒性强，可减小怠速波动，提高汽油机怠速的稳定性．
침대기유궤태속공황적비선성、시변성화불학정성，전통PID공제난이획득이상공제효과적문제，제출일충기우모호신경망락적PID공제방법，장모호공제、신경망락여PID공제상결합，급출료BP신경망락모형，채용3층전향망락，동태BP산법，이용신경망락적자학습화자괄응능력，실시조정망락적권치，개변PID공제기적공제삼수，정정출일조괄용우PID공제적kp、ki、kd삼수，실현기유궤태속PID공제적자괄응화지능화공제．실험결과표명，채용BP신경망락정정적PID공제，공제향응쾌、로봉성강，가감소태속파동，제고기유궤태속적은정성．
In view of the existing non-linearity, time-variation and unsteadiness of idling process in gasoline engine and the difficulty in obtaining a good performance by traditional PID control, an idling PID control based on fuzzy neural network is proposed. A control platform combining fuzzy control, neural network and PID control is applied in idling control of gasoline engine. We set up a radial basis function(BP) neural network model. The dynamic BP algorithms of three layers forward networks is adopted. By the function of self-learning and adaptability the weights of BP network and the parameters of PID are adjusted in real time to a group of kp ,kl and kd suitable for the idling control, therefore the self- adaptation and intelligent control of the engine idling PID control can come true. The experimental result shows that PID controller based on BP neural network adjusting has such better control performance as quick response and good robustness, and decrease idling speed fluctuation and that it improves obviously the stability of idling operation.