基于人工智能的加工过程质量诊断与调整研究

发布时间：2018-10-21 11:24

【摘要】：产品质量形成并贯穿于整个产品生命周期,是企业参于市场竞争、赖以生存和发展的基础,而加工过程中的产品质量是产品最终质量的基石。随着“世界级质量”的提出,市场、客户、企业对质量的要求不断提高,传统的质量控制已不能满足其需求。加工过程中的产品质量出现异常,只有很少一部分时间用于质量监测和控制,而80%的时间都用来判断异常的来源和调整引起异常的因素,所以为了满足加工过程全面质量管理的需求,应用于加工质量的诊断技术和系统成为众多学者和企业新的研究热点。本文在前人质量控制研究的基础上,提出了基于改进BP神经网络的控制图模式识别,同时对加工过程中的质量诊断与调整进行了相关研究。主要研究包括三方面的内容： (1)提出一种基于激励函数参数可调和动态阈值的改进BP神经网络控制图模式识别算法,并优化Monte Carlo工序数据模拟方法,使样本数据更具与实际生产数据相同的质量特性。根据改进后的网络参数迭代公式,将预处理后的样本数据作为输入对该神经网络识别器进行训练,训练结果用于生产过程的控制图模式识别。改进BP神经网络识别器的拓扑结构简单,在保证识别精度的前提下,提高识别速度,改善神经网络的泛化能力。最后,通过计算机模拟验证该算法的可行性。 (2)提出基于故障树分析的加工过程质量诊断与调整方法。首先,对加工过程与质量相关的规则进行编码和产生式知识表示,系统自动生成以控制图异常模式为顶事件的故障树,通过故障树分析获取引起质量波动的主要异常因素子集。然后,以各质量因素特征值的监测结果作为规则匹配依据,进行专家系统的自动推理和人工辅助推理。最后,系统针对控制图异常模式诊断结果进行质量调整,将最优调整方案及时反馈给技术人员改进生产。质量诊断与调整专家系统有助于生产人员快速诊断质量异常因素、实施质量调整方案,大大缩短产品生产周期。 (3)开发了车削加工质量诊断与调整专家系统。采用VB6.0开发环境和SQL Server 2003数据库软件,建立专家系统知识库和各功能模块的人机交互界面,实现加工过程的控制图模式识别、控制图异常模式的质量诊断与调整、专家知识库的维护和机器自学习功能。
[Abstract]:Product quality forms and runs through the whole product life cycle. It is the basis for enterprises to participate in market competition and survive and develop. The product quality in the process of processing is the cornerstone of final product quality. With the development of "world-class quality", the demand of market, customer and enterprise for quality has been improved, the traditional quality control can not meet its demand. In the process of manufacturing, the quality of the product is abnormal, only a small part of the time is spent on quality monitoring and control, while 80% of the time is spent on judging the source of the anomaly and adjusting the factors causing it. Therefore, in order to meet the requirement of total quality management in machining process, the diagnostic technology and system applied to machining quality has become a new research hotspot for many scholars and enterprises. Based on the previous research on quality control, this paper puts forward the control chart pattern recognition based on improved BP neural network, and studies the quality diagnosis and adjustment in the process of machining. The main contents are as follows: (1) an improved BP neural network control chart pattern recognition algorithm based on the harmonic dynamic threshold of excitation function parameters is proposed, and the simulation method of Monte Carlo process data is optimized. Make the sample data have the same quality characteristics as the actual production data. According to the improved iterative formula of network parameters, the preprocessed sample data is used as input to train the neural network recognizer, and the training result is used for pattern recognition of control chart in production process. The topology structure of the improved BP neural network recognizer is simple, the recognition speed is improved and the generalization ability of the neural network is improved under the premise of ensuring the recognition accuracy. Finally, the feasibility of the algorithm is verified by computer simulation. (2) A fault tree analysis based process quality diagnosis and adjustment method is proposed. Firstly, the rules related to the quality of the machining process are coded and the knowledge is expressed productively. The system automatically generates the fault tree with the exception pattern of the control diagram as the top event, and obtains the subset of the main abnormal factors that cause the quality fluctuation through the fault tree analysis. Then, the monitoring results of each quality factor's characteristic value are taken as the basis of rule matching, and the expert system's automatic reasoning and artificial assistant reasoning are carried out. Finally, the system adjusts the quality of the abnormal pattern diagnosis result of the control chart, and feedback the optimal adjustment scheme to the technicians to improve the production in time. The quality diagnosis and adjustment expert system is helpful for the production personnel to quickly diagnose the abnormal quality factors, implement the quality adjustment scheme, and greatly shorten the production cycle. (3) an expert system for the diagnosis and adjustment of turning quality is developed. By using VB6.0 development environment and SQL Server 2003 database software, the expert system knowledge base and man-machine interaction interface of each functional module are established, and the control chart pattern recognition and the quality diagnosis and adjustment of the abnormal control chart pattern are realized. Expert knowledge base maintenance and machine self-learning function.
【学位授予单位】：山东大学
【学位级别】：硕士
【学位授予年份】：2011
【分类号】：TH161.5;TP311.52

【参考文献】