基于神经网络的直线电磁阀式怠速系统预测控制方法研究

基于汽油发动机怠速系统的非线性、时变性和不确定性等特点,构建了神经网络与预测算法相结合的控制系统。利用预测控制算法的滚动优化和反馈校正的特性,采用神经网络建立系统的动态模型作为预测控制器的预测模型;提供怠速系统的开环输入输出数据离线训练神经网络,再在线对神经网络模型的权值和阀值进行调整,获得精确的预测模型,实现了对怠速系统的自适应控制。仿真结果表明,这种方法有效地提高了发动机怠速系统的控制精度、可靠性和转速的稳定性。

Predictive Control of the Idle Speed System of a Linear Electromagnetic Valve Based on Neural Networks

In accordance with the features of the gasoline engines idle speed system,such as nonlinearity,randomicity and uncertainty,a control system that integrates a neural network and a predictive algorithm is proposed. Using the characteristic of predictive control system′s roll optimization and feedback emendation,we first take the system′s dynamic model which is established based on a neural network as the predictive controller′s model. Then, we utilize the input and output data of the system to train the neural network offline,and adjust the weights and threshold value online in order to realize the idle speed system′s adaptive control. The simulation result indicates that the algorithm effectively improves the control precision,reliability and stability of the engine speed.