基于bp神经网络的铝合金变质研究—本科毕业论文.docVIP

摘 要 炉前对铝合金变质效果的预测，是提高铸件质量，降低废品率，实现生产监控的重要环节。 本文主要介绍人工神经网络（Artificial Neural Network,ANN）在铸造铝合金变质效果预测的应用研究工作，主要研究基于反向传播（Back Propagation）BP神经网络在课题中的应用。 论文在进行基于人工神经网络（ANN）的铝合金变质效果研究过程中，主要进行以下工作：介绍和分析国内外对铝合金变质效果的研究现状；简要说明铝合金变质的热分析理论；人工神经网络（ANN）概述；BP网络介绍和算法选择；决定神经网络的输入、输出变量以及网络训练样本、测试样本的选取和数据归一化处理；比较两种BP网络改进算法对网络训练结果的影响，然后选择最优值；检验新的预测模型对于测试数据的应用效果；提出预测模型的实际应用存在的问题和后续的研究工作的见解。 关键词：铝合金变质；人工神经网络；BP算法 Abstract The prediction of the aluminum alloy modification effect before furnace is an important link to improve casting quality and reduce scrap rate in production monitoring. This paper presents my work chiefly on the research of the prediction of casting alumin- um alloy modification effect based on Artificial Neural Network, and I largely do my research on the application of Back-Propagation neural network in this task. This paper mainly introduced the work as the following in the the prediction of casting aluminum alloy modification effect based on Artificial Neural Network: to introduce and to a- nalysis domestic and foreign research status of aluminum alloy modification level; to briefly describe the thermal analysis theory of aluminum alloy modification; to briefly introduce AN- N; to introduce popular Back-Propagation network and to choose the learning algorithm; to decide the input variables of the artificial neural network, to decide the training simples and to normalize the data; to compare the training results of two BP networks with different improv- ed BP algorithm and to choose the best; to test the new module with the testing samples; to put forward the problems existing in the practical application of the model and the improvem- ent to further study. Keywords: aluminum alloy modification; artificial neural network; BP algorithm 目 录 摘 要 I Abstract II 1 绪论 1 1.1 课题的研究意义 1 1.2 课题研究背景和现状 1 1.3 课题主要内容 1 2 铝合金变质检测方法综述 3 2.1 铝合金变质的传统测量方法 3 2.1.1 断口观察法 3 2.1.2 电导率法 3 2.1.3 金相法 4 2.1.4