基于神经网络的发动机模型辨识

为进一步研究发动机的稳态、空载、动态转矩、燃油消耗模型和万有特性,笔者提出了一种基于神经网络的神经元结构模型和多层前馈网络结构模型并分别进行了数学分析.实践研究表明：在发动机的工作过程中，动态工况占66%～80%，并发现发动机的动力性和经济性指标与稳态工况存在着一定的差异.

Study on Model Recognition of Engine on Neural Network

In order to give a further research on steady state and empty carrying dynamic state torque and fuel consuming model and universal characteristics of engine, this text tries to carry on mathematical analysis respectively on a structure of neural model and many layer forward feedback network structure models through neural network and make some quests towards model recognition of engine. Practice clarifies that in engine working process, the dynamic state work condition has occupied 66% - 80%, dynamical and economic capability signs of engine have certain difference between the steady state work condition and the dynamic state work condition.