摘要为了对碳钢焊条焊接熔敷金属的合金元素进行预测，建立了焊条成分预测遗传算法神经网络模型及BP神经网络模型，通过25组数据对模型进行训练，5组数据对模型进行预测验证，研究分析得出以下结论：E4303碳钢焊条中，依据焊芯中的C含量、中碳锰铁中C含量、还原钛铁矿中C含量，对熔敷金属中C含量进行预测，平均相对误差为10.73%；依据焊芯中Mn含量、中碳锰铁中Mn含量，对熔敷金属中Mn含量进行预测，平均相对误差为6.80%，可以满足实际工作预测要求。表明所建立的遗传算法BP神经网络模型是有效的，能够根据焊条合金元素的含量对熔敷金属合金含量进行预测。68100

毕业论文关键词：碳钢焊条  BP神经网络  遗传算法  合金元素预测

毕业设计（论文）外文摘要

Title  Based on genetic algorithms and neural network prediction carbon steel electrode alloying elements                                                          

Abstract

In order to prediction the alloying elements of carbon steel electrode deposited metal,we established electrode component prediction genetic algorithm neural network model and BP neural network model,the model through 25 sets of data for training, five sets of data to predict and validate.we can research and analysis the following conclusions：According to carbon steel electrode welding the C content in the core,the C content in the medium carbon ferromanganese, the C content of ilmenite, forecast for C content in deposited metal，the average relative error is 10.73% ；the Mn content in the core, the Mn content in the medium carbon ferromanganese,forecast for Mn content in deposited metal，the average relative error is 6.80%, to meet the requirements of practical work prediction.That the established genetic algorithm neural network model and BP neural network model is valid, according to the content of alloying elements on the electrode deposited metal alloy content for prediction.

Keywords： Carbon steel electrode     BP neural network    Gentic Algorithms          Alloying elements forecast

目录

1  绪论 1

1.1  前言 1

1.2  焊接材料研究概况 1

1.3  神经网络在材料中的应用概况 2

1.4  遗传算法在材料中的应用概况 4

1.5  本课题研究内容 5

2  碳钢焊条合金元素BP神经网络预测模型 6

2.1酸性碳钢焊条配方成分特点分析 6

2.2  BP神经网络模型建立 10

2.3  BP神经网络模型的训练 10

2.4  BP神经网络模型的预测 16

3  遗传算法优化BP网络 18

3.1  遗传算法优化的优点 18

3.2  遗传算法优化的具体方法 18

3.3  遗传算法优化BP网络训练 20

3.4  遗传算法优化BP网络预测 23

4  结论 26

致谢 27