基于进化RBF网络的柴油机喷油参数研究  

Parameters of the fuel injection in the diesel engine based on the evolutionary RBF neural network

作　　者：黄强[1] 刘永长[1] 刘会猛[1] 张俊[1] 

机构地区：[1]华中科技大学能源与动力工程学院

出　　处：《华中科技大学学报（自然科学版）》2002年第9期94-96,共3页Journal of Huazhong University of Science and Technology(Natural Science Edition)

基　　金：国家自然科学基金资助项目 (5 9876 0 15 )

摘　　要：基于柴油机燃油喷射电子控制实时性的要求 ,引入非线性最优控制理论来确定RBFNN的参数 ,用全局优化算法———遗传算法来离线求解非线性优化问题 ,用递推最小二乘法 (RLS)在线调整进化RBFNN的输出权值 ,增加系统的抗干扰能力 .同时通过分析输入参数与输出参数的相关性来简化该神经网络 .结果表明 :冷却水温及润滑油温度与喷油参数的相关性最低 。In order to improve the real time injection in the diesel engine, the optimal control theory is introduced to conform the parameters of RBENN. The nonlinear optimization is made by the genetic algorithm. The output weights of EPRBF neural network on line is adjusted according to recursive least square. The correlation between the input and output parameters is analyzed to simplify the structure. The results obtained show that the influence of the temperature of the cooling water and lubricating oil on the fuel injection parameters is smaller. The neural network without them has better approach effect with less iterative times.

关 键 词：喷油参数 柴油机 电控燃油喷射 RBF神经网络 非线性最优控制 冷却水温 润滑油温度 

分 类 号：TK421.42[动力工程及工程热物理—动力机械及工程] TP183[自动化与计算机技术—控制理论与控制工程]