基于Bayesian网络的机械加工缺陷溯源方法

Machining error source tracing method based on Bayesian networks

为提高机械加工缺陷原因追溯方法对不确定信息的解释和推理能力,引入Bayesian网络分析方法。给出了利用Bayesian网络进行缺陷溯源的建模方法和推理步骤。一方面,用实际加工状况和缺陷现象作为复合证据;另一方面,利用Bayesian推理确定网络任意原因节点的发生概率及搜索最大概率路径,从而提供了一种针对不确定信息的诊断方法。最后,以转子法兰联接孔的加工缺陷溯源为例,对该方法进行了验证。

To enhance the ability of methods to interpret and infer uncertain information in the source tracing of machining errors, Bayesian Networks was introduced. The modeling method and inference procedures of machining errors source tracing based on Bayesian Networks were presented. On one hand, actual machining conditions and machining error phenomena were taken as complex evidences; on the other hand, by using Bayesian Networks inference, the probability of each cause was obtained and the maximum probability path was searched; and then a diagnosis method for uncertain information was provided. At last, a case study for the machining error source tracing of the connection holes on the rotor flange was reported to verify the proposed approach.