摘 要

内燃机是世界上应用范围最广的动力机械，近年来对汽车发动机性能及排放的研究层出不穷，但由于内燃机燃烧过程较为复杂，研究难以高效进行。人工神经网络是科学家们参照人脑运行方式而逐渐研究发展出来的学科，随着人工神经网络的发展，衍生出当今应用最广泛的BP神经网络。本文利用BP神经网络，在避免高难度实验、复杂数学推理的情况下对某增压直喷发动机性能及排放做出了预测。本文主要研究内容如下：

首先，本文论述了国内外发动机性能及排放预测模型的研究状况；介绍了人工神经网络的发展历史及相关研究。介绍了BP神经网络的基本原理及基本结构，介绍了建立BP神经网络模型用到的函数功能及其具体意义。论述了MATLAB实现BP神经网络的基本方法。说明了用遗传算法优化BP神经网络的目的和意义。

其次，通过对研究课题和实验数据的分析，确定了BP神经网络的基本结构及参数的设置方法。研究了BP神经网络训练参数的设定方法及训练过程；分析了遗传算法的基本要素，研究了遗传算法的基本流程，分析了遗传算法基本参数的设置方法，研究了用遗传算法优化BP神经网络的具体步骤，介绍了遗传算法的重要函数。

最后，本文评价了BP神经网络模型的可行性；分析BP神经网络模型对该增压直喷发动机油耗及NO_X排放预测的结果，解释了误差产生的原因。分析了遗传算法优化该BP神经网络模型的结果，对比了遗传算法优化前后BP神经网络模型的预测误差，分析了误差产生的原因及优化效果。

研究结果表明：所构建的BP神经网络具有一定的可靠性，能够对该增压直喷发动机的性能及排放做出误差允许范围内的预测；遗传算法优化BP神经网络后误差下降明显，优化效果良好。

关键词：内燃机；BP神经网络；遗传算法；排放

Abstract

The internal combustion engine is the most widely used power machine in the world. In recent years, the research on the performance and emission of automotive engines has been continuously developed. However, due to the complexity of the internal combustion engine combustion process, the research is difficult to carry out efficiently. Artificial neural network is a subject that scientists have gradually studied and developed by referring to the way of human brain operation. With the development of artificial neural network, the most widely used BP neural network is derived. This paper makes use of BP neural network to predict the performance and emission of a supercharged direct injection engine under the condition of avoiding difficult experiments and complex mathematical reasoning. The main research contents of this article are as follows:

Firstly, this paper discusses the research status of engine performance and emission prediction models at home and abroad, and introduces the history of artificial neural network development and related research. The basic principle and basic structure of BP neural network are introduced. The function functions and specific meanings used in establishing BP neural network model are introduced. The basic method of implementing BP neural network by MATLAB is discussed. It explains the purpose and significance of optimizing BP neural network with genetic algorithm.

Secondly, through the analysis of research topics and experimental data, the basic structure of BP neural network and the setting method of parameters are determined. The method of setting training parameters and training process of BP neural network are studied. The basic elements of genetic algorithm are analyzed. The basic flow of genetic algorithm is studied. The basic parameter setting method of genetic algorithm is analyzed. The BP neural network optimized by genetic algorithm is studied. The specific steps introduced the important functions of the genetic algorithm.

Finally, this paper evaluates the feasibility of the BP neural network model; analyzes the results of the BP neural network model for predicting the fuel consumption and NOX emissions of the turbocharged direct injection engine, and explains the causes of the error. This paper analyzes the results of genetic algorithm optimization of the BP neural network model, compares the prediction error of the BP neural network model before and after the optimization of the genetic algorithm, analyzes the causes of error and the optimization effect.

The results show that the BP neural network constructed has certain reliability and can predict the performance and emissions of the supercharged direct injection engine within the allowable range of error. After the BP neural network is optimized by the genetic algorithm, the error decreases significantly. good.

Keywords :combustion engine; BP neural network; genetic algorithm; emission

目 录

摘 要 I

Abstract II

第1章 绪论 1

1.1 选题背景及意义 1

1.2 国内外研究现状 1

1.2.1 发动机性能及排放的研究进展 1

1.2.2 人工神经网络的发展及研究 2

1.3 研究内容及技术路线 4

1.3.1 研究内容 4

1.3.2 技术路线 5

第2章 BP神经网络 6

2.1 BP神经网络概述 6

2.1.1 BP神经网络的基本原理 6

2.1.2 BP神经网络的MATLAB实现 7

2.2 BP神经网络的网络结构 8

2.3 用遗传算法优化BP神经网络 9

2.4 本章小结 9

第3章 模型建立及运用 10

3.1 实验设备及过程 10

3.2 BP神经网络模型结构的确定 11

3.2.1 模型层数的确定 11

3.2.2 输入层、输出层及隐含层神经元数的确定 11

3.3 BP神经网络运行过程 11

3.3.1 训练样本的确定 11

3.3.2 传递函数的选择 12

3.3.3 网络的训练 13

3.4 遗传算法优化BP神经网络的过程 14

3.4.1 遗传算法的基本要素 14

3.4.2 遗传算法流程 15

3.4.3 遗传算法的优化过程 16

3.5 本章小结 18

第4章 结果分析 19

4.1 BP神经网络模型输出结果及分析 19

4.1.1 BP神经网络的评价 19

4.1.2 BP神经网络预测结果 21

4.2 遗传算法优化BP神经网络的结果 26

4.3本章小结 28

第5章 结论 29

以上是毕业论文大纲或资料介绍，该课题完整毕业论文、开题报告、任务书、程序设计、图纸设计等资料请添加微信获取，微信号：bysjorg。

相关图片展示：