压力容器焊接质量控制中的数据挖掘方法及其应用研究

发布时间：2018-08-02 21:06

【摘要】：目前,焊接在压力容器生产制造过程中的地位越来越重要,为了确保压力容器的质量,实行焊接质量的控制和管理显得尤为关键。当前计算机技术以及数据库技术的飞速发展,现代制造业在生产过程中已经产生海量数据,很显然,传统的统计方法在此陷入了困境,已经不能满足当今时代的发展需要。本文针对现代制造业中的海量数据无法提取知识的问题,结合某公司的实际项目,将数据挖掘技术应用于制造行业中的焊接质量控制当中。本文在前人研究的基础上,重点研究了压力容器焊接质量控制的工作,研究内容归纳如下:首先,针对压力容器焊接质量不合格的现象,从质量管理5M1E(操作者、机器设备、原材料、工艺方法、环境和测量)等多个方面,全过程、多方位、多角度的来分析影响压力容器焊接质量的因素,缩小焊接质量异常范围。其次,本文提出了基于特征选择与决策树C5.0相结合的分类方法,首先采用特征选择算法对大量的特征进行降维处理,将降维后的焊接数据,利用决策树C5.0算法构建焊接分类模型,从而找出影响焊接质量不合格的影响因素。将这些重要因素看作"质量控制点"并严格控制。最后,将决策树算法与神经网络算法和Logistic回归算法进行比较,实验表明,针对本文中研究的问题,决策树方法的准确性要优于神经网络算法和Logistic回归算法,并将决策树分析的结果应用于公司的实际焊接工作中,焊接问题得到了明显的改善,其理念和指导思想在压力容器制造行业中具有—定的推广应用价值。
[Abstract]:At present, welding is becoming more and more important in the manufacturing process of pressure vessels. In order to ensure the quality of pressure vessels, the control and management of welding quality is particularly critical. With the rapid development of computer technology and database technology, the modern manufacturing industry has produced massive data in the process of production. Obviously, the traditional statistical method is in a dilemma and can not meet the needs of the development of the times. Aiming at the problem that mass data can not extract knowledge in modern manufacturing industry, this paper applies data mining technology to welding quality control in manufacturing industry, combining with the actual project of a company. On the basis of previous studies, this paper focuses on the work of welding quality control of pressure vessels. The research contents are summarized as follows: first, aiming at the phenomenon of nonconformity of welding quality of pressure vessels, the quality management 5M1E (operator, machine equipment), Raw materials, process methods, environment and measurement, etc., the whole process, multi-direction, multi-angle to analyze the factors that affect the welding quality of pressure vessels, and reduce the abnormal range of welding quality. Secondly, this paper proposes a classification method based on feature selection and decision tree C5.0. Firstly, the feature selection algorithm is used to reduce the dimension of a large number of features. The decision tree C5.0 algorithm is used to construct the welding classification model, so as to find out the influencing factors that affect the welding quality. Regard these important factors as "quality control points" and strictly control them. Finally, the decision tree algorithm is compared with the neural network algorithm and the Logistic regression algorithm. The experimental results show that the accuracy of the decision tree algorithm is better than that of the neural network algorithm and the Logistic regression algorithm. The result of decision tree analysis is applied to the actual welding work of the company, and the welding problem is obviously improved. Its concept and guiding ideology have a definite value of popularization and application in the pressure vessel manufacturing industry.
【学位授予单位】：天津工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TG457.5

【参考文献】