采用神经网络改善闭式循环柴油机的供氧控制

张卫东　 陈国钧　 唐开元　

摘　要：在氧气反馈调节的基础上，不依赖系统模型，借助神经网络构成前馈控制器，以反馈输出引导网络权值及输出的调整，使网络逐步学成前馈补偿功能，并最终在控制中占据主导地位，实现对负荷扰动的补偿。仿真结果表明，采用这一复合控制系统能有效地改善氧气控制的动态特性。
关键词：柴油机； 闭式循环； 氧气控制； 神经网络； 前馈补偿
分类号：TK421.3;TP183　文献标识码：A
文章编号：1000-0909（2001）02-0173-04

Neural Feedforward Control for Closed Cycle Diesel Oxygen Replenishment

ZHANG Wei-dong（College of Power Engine,Naval University of Engineering,Wuhan 430033,China）　
CHEN Guo-jun（College of Power Engine,Naval University of Engineering,Wuhan 430033,China）　
TANG Kai-yuan（College of Power Engine,Naval University of Engineering,Wuhan 430033,China）　

Abstract：The relatively slow response of the oxygen sensor results in the poor dynamic performance of oxygen PID control system which does not meet the dynamic requirements of the closed cycle diesel system.In this paper,the neural network is applied as a feedforward controller.Using the feedback controller output as the training error to train the network,the neural network will learn to compensate the step load changes gradually by minimizing the PID output.Two training methods are discussed.The results of the simulation show that this combining control system has a good control effect in comparison with the PID controller.
Keywords：Diesel； Closed cycle； Oxygen control; Neural networks; Feedforward compensation

参考文献：

［1］HAWLEY J G,ASHROFT S J.Advanced Underwater Power Systems［J］．Proceedings of the Institution of Mechanical Engineers.Part A:Journal of Power and Energy,1994,209(2):37-45.
［2］ZHANG Yu-sheng,QIU Ling.Simulation and Experimental Studies on Closed-Cycle Diesel Engines［C］．In:Proceedings of the International Conference on Internal Combustion Engines,1997.731-735.
［3］LEE M,PARK S.A New Scheme Combining Neural Feedforward Control with Model Predictive Control［J］．AICHE Journal,1992,36(2):193-200．
［4］顾宏中，邬静川.柴油机增压及其性能优化［M］．上海：上海交通大学出版社，1989．
［5］陆燕，杜继红.延迟时间未知的时延系统神经网络补偿控制［J］．清华大学学报，1998，38(9)：67-69．

收稿日期：2000年6月5日

修稿日期：2000年7月26日

出版日期：2001年3月25日