【作者】 杨桂林；

【导师】 雷玉勇；

【作者基本信息】 西华大学 ， 机械制造及其自动化， 2006， 硕士

【摘要】 本文通过对水射流技术国内外发展现状的分析研究,提出了磨料水射流切割断面质量控制的研究课题。论文对磨料水射流技术的原理、磨料水射流切割的微观机理进行了理论分析。在研究总结磨料水射流切割工艺特点的基础上,应用人工神经网络理论建立了用于控制磨料水射流切割断面质量的人工神经网络模型。通过实验研究获取大量样本数据,并基于MATALAB编程实现人工神经网络模型的训练。最后在实验室对所建立的人工神经网络模型进行了验证。 本论文的研究目的是建立磨料水射流切割各工艺参数之间的非线性关系,从而为实现磨料水射流精密加工提供理论基础。论文主要研究内容包括: 1、水射流技术国内外发展现状综述。 2、磨料水射流切割的微观机理分析。 3、基于人工神经网络理论,建立磨料水射流切割的BP神经网络模型。 模型的输入输出参数为射流压力、磨料流量、材料厚度、靶距、切割表面的粗糙度和喷嘴的横移速度(进给速度)。 4、磨料水射流切割的实验研究。在实验规划基础上,以不锈钢为切割样本材料,获取实验样本数据。 5、磨料水射流切割人工神经网络模型的计算。应用Levenberg-Marquard(LM)算法,基于MATLAB工具箱提供的人工神经网络工具函数和试验更多还原

【Abstract】 The research topic on control of cutting quality of abrasive water jet (AWJ) is put forward as the result of development of overview of water jet technology abroad. The principle of water jet technology and mechanism of micro-crack using abrasive water jet was analyzed. Based on the characteristics of abrasive water jet cutting process, an artificial neural network (ANN) model for controlling cutting quality of abrasive water jet is established based on ANN theory. The ANN model of AWJ was then trained based on experimental data using MATIAB Tool Box. The trained ANN model for abrasive water jet cutting is verified experimentally.The purpose of this study is to establish the non-linear relationship among the material to be cut, parameters of abrasive water jet cutting process. Thereafter the fundamental for realization of precise cutting is expected as the result of this study. The followingwork is carried out in this thesis.1. Overview of development of abrasive water jet technologyover abroad and in China.2. Analysis of mechanism of micro-crack materials by abrasive water jet.3. A BP ANN model for the abrasive water jet cutting process is built based on ANN theory. Material thickness, water jet working pressure, abrasive flow rate, standoff as well as cutting quality (surface roughness) are selected to be input parameters. The feed rate of water jet cutting is the output parameters of the ANN model.4. Experimental study on abrasive water jet precise cutting. Experiment is designed. Stainless steel is selected to be cutting materials.5. Computation of ANN model of abrasive water jet cutting. The Levenberg-Marquardt algorithm is programmed using MATLAB Tool Box. The ANN model is then trained based on experimental data. After all, the non-linear relationship among the main parameters of abrasive water jet cutting process is established.6. Verification of ANN model for abrasive water jet cutting. Based on trained ANN model, a feed rate can be predicted. After that, numerical control cutting code is programmed. And cutting surface roughness is tested by TR200 roughness instrumentation. It shows that established ANN model is good with experiment.This study has great meaning in abrasive water jet precise cutting application.更多还原