基于信息融合的建模方法及其优化  

Modeling and Optimizing of Fuel Injection System of Diesel Engine Based on Neural Networks

作　　者：黄强[1] 宾鸿赞[1] 高世伦[1] 刘永长[1] 

机构地区：[1]华中科技大学能源与动力工程学院,湖北武汉430074

出　　处：《柴油机设计与制造》2005年第4期16-19,共4页Design and Manufacture of Diesel Engine

摘　　要：建立基于神经网络技术的柴油机燃油喷射系统简化模型,然后利用进化编程和遗传算法分别对该模型进行结构和参数的优化。优化结果表明:进化编程算法可以使网络逼近精度足够高;而基于信息基准(AIC)判据的遗传算法可以获得网络逼近精度和结构复杂程度之间的折中。建立的模型结构简单,更适合柴油机这类高速系统的实时控制[1]。A model for fuel injection system of diesel engine is built based on neural networks. Then the evolution programming and genetic algorithm are used to optimize the structure and parameters of the model. The evolution programming can get higher approach precision of the networks. On the other hand, the genetic algorithm based on the AIC criterion can be a trade off between the approach precision and the structure complicacy. The model with simpler structure built by it is beneficial to the real time control of high speed system such as diesel engine.

关 键 词：柴油机 燃油喷射系统 神经网络 优化 建模 

分 类 号：TK423[动力工程及工程热物理—动力机械及工程]