基于发动机模型的神经网络点火控制器

针对YN30QNE发动机,根据发动机平均值原理,建立了发动机平均值模型。在均值模型的基础上,建立了外负荷控制器模型,根据发动机的准维燃烧模型建立了发动机的点火提前角控制模型,并制取了点火提前角的初始MAP图。根据模型测得的试验数据利用神经网络自学习的能力,采用BP神经网络的改进LM优化控制方法,对点火提前角进行训练,神经网络根据训练过后得到的权值和阀值,达到发动机点火控制的目的。最后,搭建神经网络点火系统的Simulink对其进行仿真验证。

Ignition Controler of Neural Network Based on Engine Model

As for the engine YN30QNE, the mean-value model of the engine was set up, based on the theory of the mean-value of the engine. And on the basis of the mean-value model, the externalloading controller model was founded. Furthermore, based on the theory of the adjust dimension burning model of the engine, the angle of the advance ignition controlling model was set up, and the initial MAP figures of the angle of the advance ignition were obtained. Moreover, on the basis of data tested from the models as well as using the ability of self-study of the neutral network, the angle of the advance ignition can be trained with the method of the improved LM optimizing control of the BP neutral network. And the weighted values and the threshold values can be obtained by the neutral network from the training in order to control the ignition of the engine. Finally, the simulation results were verified by setting up the Simulink model of ignition system of the neutral network.