射流喷丸表面质量预测研究

发布时间：2018-01-19 20:47

本文关键词： 表面质量喷丸强化人工神经网络 Matlab VB　出处：《黑龙江科技大学》2015年硕士论文　论文类型：学位论文

【摘要】：现代科技发展越来越快,对生活质量的要求越来越高,工业部门对零构件的性能也提高了要求。机械产品质量的好坏代表着一个国家综合国力的强弱,喷丸有利于改进机械零件表面的性能,因此对机械零构件表面质量优劣的研究就尤为重要。本文选择2A11铝合金为进行射流喷丸强化试验的材料,研究起主要作用的工艺参数——喷丸压力、靶距和喷嘴移动速度对零件表面质量的影响情况。采集对喷丸效果起主要衡量作用的表面粗糙度、显微维氏硬度和残余应力等表面质量指标数据;以试验数据为依据,应用标准BP神经网络算法建立表面粗糙度预测模型,并与回归分析法和小波包方法模型的预测结果进行了分析对比,表明神经网络的预测效果更优;应用LM前馈神经网络和Elman反馈神经网络建立射流喷丸表面质量预测模型,对比两种方法预测所得数据,得出效果最优预测模型。在此基础上,采用VB和Matlab混合编程的方法搭建人机交互界面,用VB语言实现用户界面设计,Matlab软件实现复杂运算,通过二者的有机结合,实现了射流喷丸表面质量预测系统的软件开发。研究表明,相对其他几种模型,采用Elman反馈神经网络建立的预测模型训练时间更短,预测精度更高,更能满足工业上对喷丸表面质量预测精度的要求。据此建立的表面质量预测软件系统界面简洁,操作方便,具有较大的使用价值。
[Abstract]:With the rapid development of modern science and technology, the quality of life is becoming more and more demanding, and the performance of the industrial sector has been improved. The quality of mechanical products represents the strength of a country's comprehensive national strength. Shot peening is beneficial to improve the surface performance of mechanical parts, so it is particularly important to study the surface quality of mechanical components. In this paper, 2A11 aluminum alloy is selected as the material for jet shot peening test. The main technological parameters, shot peening pressure, target distance and nozzle moving speed, which affect the surface quality of the parts, are studied. The surface roughness, which plays an important role in measuring the shot peening effect, is collected. Surface quality data such as microhardness and residual stress; Based on the experimental data, the prediction model of surface roughness is established by using the standard BP neural network algorithm, and compared with the prediction results of regression analysis and wavelet packet method. It shows that the prediction effect of neural network is better than that of neural network. LM feedforward neural network and Elman feedback neural network are used to establish the prediction model of jet shot peening surface quality. Using VB and Matlab mixed programming method to build the human-computer interface, using VB language to realize the user interface design and Matlab software to achieve complex operation, through the organic combination of the two. The software development of jet shot peening surface quality prediction system is realized. The research shows that compared with other models, the prediction model established by Elman feedback neural network has shorter training time and higher prediction accuracy. The software system is simple in interface, convenient in operation and valuable in use.
【学位授予单位】：黑龙江科技大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TG668

【相似文献】