基于黑箱模型和RBF-PID的HCCI发动机燃烧正时控制

由于均质充气压缩点燃(HCCI)发动机缺少直接控制其燃烧的手段,导致HCCI发动机的燃烧正时控制成为HCCI发动机的研究热点。以HCCI发动机进气歧管的温度和压力、燃油当量比、转速以及进气门关闭正时为输入,利用BP神经网络建立用于估计HCCI发动机燃烧正时的黑箱模型。在此模型基础上,以进气门关闭正时为控制量设计了PID控制器,并利用径向基神经网络对其参数进行整定,以实现对燃烧正时的反馈控制。实验结果表明,BP神经网络估计模型对HCCI发动机燃烧正时的估计误差小于0.4(CAD),能实现准确的估计;此外,与传统的PID控制器相比,设计的RBF-PID控制器在超调量、调节时间以及抗干扰性等性能方面均有改善。

Combustion timing control of homogeneous charge compression ignition engine based on RBF PID and black box model

Combustion timing control is a research hot of homogeneous charge compression ignition (HCCI) due to the lack of direct actuator of combustion. A black box model based on BP neural network is proposed to estimate the combustion timing with the temperature and pressure of gas at intake manifold, fuel equivalence ratio, speed and intake valve closing timing as the system inputs. A PID controller which is adjusted by RBF neural network is designed to control combustion timing through controlling the intake valve closing timing. Results from numerical experiments show that the BP neural network model can estimate combustion timing accurately with error less than 0.4(CAD). Compared with the traditional PID controller, the proposed controller has a better performance with respect to the overshoot, settling time, and the robustness.