论文部分内容阅读
针对叶片加工过程中质量精度不高的问题,提出了基于动态Bayesian网络的叶片加工质量监控与溯源方法。利用动态Bayesian网络建立起叶片加工工序间的相互联系,实现对整个加工过程的控制。基于Bayesian网络对影响加工工序的因素集建立因果联系,采用多元统计过程控制中的T2控制图完成对各工序影响因素集的监控。进行误差溯源时,根据Bayesian网络建立的因果关系对失控样本的T2统计量依据原因变量进行误差分解,并构建各分解变量的控制限,将其作为误差源判定的条件。通过对某叶片加工过程的仿真,验证了所提方法的有效性。
Aiming at the problem of low quality precision in the process of blade processing, a method of quality control and traceability of blade processing based on dynamic Bayesian network is proposed. The use of dynamic Bayesian network to establish the process of blade processing between the interconnection, to achieve the entire process control. Based on the Bayesian network, the set of causal factors is established for the factors that affect the process. The T2 control chart in the multivariate statistical process control is used to monitor the set of influencing factors in each process. According to the causality established by Bayesian network, T2 statistic of uncontrolled samples is decomposed according to the causal factors, and the control limit of each decomposing variable is constructed, which is used as the condition of error source judgment. Through the simulation of a blade processing process, the effectiveness of the proposed method is verified.