Research on the optimizing control technology based on fuzzy-neural network for hydrogen-fueled engines
Research on the optimizing control technology based on fuzzy-neural network for hydrogen-fueled engines
复制标题
DOI：
10.1016/j.ijhydene.2006.02.027
复制
发表时间：
2006-12
影响因子：
7.2
通讯作者：
Zhenzhong Yang;Li-jun Wang;S. Xiong;Jingding Li
中科院分区：
文献类型：
--
作者：
Zhenzhong Yang;Li-jun Wang;S. Xiong;Jingding Li
氢燃料被认为是未来最有前途的汽车替代燃料之一。氢燃料发动机的优化控制技术是提高氢燃料发动机各方面性能的关键。本文在试验的基础上，提出了氢燃料发动机的最优控制策略。在此基础上，建立了以点火提前角、喷油提前角和喷油持续期为控制变量，以输出功率和燃油消耗率为性能指标函数的氢燃料发动机优化控制模型。在此基础上，巧妙地采用模糊神经网络（Fuzzy-Neural Network，FNN）系统对优化控制模型进行优化控制律的求解，建立了由FNN控制器和点火时刻自适应控制器组成的串联控制系统，实现了点火时刻的开环或闭环步进调节。最后，将改进后的模糊神经网络系统的控制量计算结果与实验结果进行了对比，对比结果表明，最大绝对误差小于1°CA，最大相对误差小于5%，均方误差为0.381°CA。因此，用模糊神经网络系统进行计算是切实可行的，令人满意的。
Hydrogen fuel is regarded as one of the most promising alternative fuels for automobiles in future. Technology of the optimum control on hydrogen-fueled engines is a key to improve its performances in every respect. In this paper, based on the experiments, the policy of optimum control on hydrogen-fueled engines has been shown. Moreover, a new optimizing control model on hydrogen-fueled engines has been constructed, in which ignition timing, injection timing and injection duration were separately selected as control variation, and output power and rate of fuel consumption are separately chosen as performance index function. Further, a new method ingeniously adopting fuzzy-neural network (FNN) system to calculate the optimizing control laws for the optimizing control model has been designed, and a series connection control system has been set up, which is composed of FNN controllers combined with an adaptive controller for ignition timing to realize open-loop or closed-loop control pattern with stepping regulation of ignition timing. Last, the calculated results of the control variations with an improved FNN system were contrasted with experimental results; the contrasting result has shown that the maximum absolute error is less than 1°CA, the maximum relative error is less than 5%, mean square error is 0.381°CA. Therefore, the calculation method with FNN system is practical and satisfactory.