基于广义回归神经网络的发动机动力性、经济性预测

Prediction of the Dynamic and Economic Performance of the Engine Based on General Regression Neural Network

神经网络是当前最主要的智能控制技术 ,它是模拟人脑的结构以及对信息的记忆和处理功能 ,具有擅长从输入输出数据中学习有用的知识的特性。发动机性能预测是根据发动机结构参数和运转参数来估算推测发动机的各种性能指标 ,因此可以利用神经网络的学习性的特点 ,借助神经网络 ,将各种影响汽油机燃烧过程的主要参数对汽油机的非线性影响以网络模型的形式表示出来。本文讨论了如何抛开数学建模的方式 ,选用广义回归神经网络进行发动机动力性、经济性的预测 ,并应用了MATLAB软件工具箱编程 ,给出一个两缸电控汽油发动机的动力性、经济性预测模型的实例。

The neural network is the most important artifical intelligence technology at present. It simulates the structure of human brain and its function of dealing with or memorizing the information. The neural network is adept at learning from the input and output data and draws out an useful information. The prediction of the engine performance means using the structure and operation parameters of the engine to predict the engine performances. So we can use the learning characteristic of the neural network to show how the primary parameters of the engine combustion process affect the engine performance nonlinearly, and describe them with a network model. This article discusses how to choose the general regression neural network(GRNN) and uses it to predict the dynamic and economic performance of the engine instead of physical model. Then it uses the programs on MATLAB to give a prediction model of the dynamic and economic performance of a two-cylinder engine.