基于贝叶斯网络推理的数控机床元动作装配情景异常识别

Anomaly recognition of meta-action assembly scenario of CNC machine tools based on Bayesian network

针对机床的机械故障频发且装配因素难以识别的问题,提出了基于贝叶斯网络的机床装配情景异常推理识别方法。以机械零部件多尺度运动分析为切入点,建立了机床功能-元动作的多尺度映射模型,利用故障模式及影响分析(Failure Mode and Effect Analysis,FMEA)方法建立了机床元动作单元关键装配情景构成模型。基于装配情景构成模型建立了元动作单元装配情景的贝叶斯网络结构,利用证据推理法实现了元动作单元装配情景异常概率的智能推理。以蜗轮转动元动作单元为例,构建了蜗轮转动单元装配情景初始贝叶斯网络,获取了蜗轮转动元动作输出的异常概率(由装配因素引起)为2.35%;以蜗轮转动故障为起点进行了贝叶斯网络反向推理,识别出导致蜗轮转动故障的各装配情景异常概率。元动作装配情景的异常识别为实现机床故障装配因素的追溯提供理论依据。

In this article,due to frequent machine failures and difficulty in identifying assembly factors,the method of assembly-scenario anomaly recognition of machine tools based on the Bayesian network is proposed.For the machine tool,the multiscale mapping model of function-meta action is set up based on the analysis on the multi-scale action of parts and components.The key-assembly-scenario-composition model of the machine tool's meta-action unit is built by with the help of FMEA(Failure Mode and Effect Analysis).Based on this model,the Bayesian network structure of the assembly scenario for the meta-action unit is worked out.The method of evidence reasoning is used to realize intelligent reasoning of the abnormal probability of the meta-action unit assembly scenario.With the worm gear's rotating meta-action unit as the example,the initial Bayesian network of the rotating unit assembly scenario is constructed,and the abnormal probability(caused by the assembly factors)of the rotating metaaction output is 2.35%.The Bayesian network is subject to reverse reasoning based on the worm gear's rotating failure;then,the abnormal probability of each assembly scenario that causes the rotating failure is identified.The abnormal recognition of the metaaction assembly scenario provides theoretical basis for tracing failure-induced assembly factors for machine tools.