Abstract: Complex parts production usually requires a hybrid multistage machining process,which has a variety of errors and accumulates in multiple processes.To ensure the quality of complex product,a method for variation propagation modeling and error source identification based on self-regulation weighted variance propagation network was proposed.This method constructed a self-regular weighted variation propagation network,taking into account the actual error and connection relationship in the processing process.And the weighted LeaderRank sorting algorithm was introduced to extract the key processing characteristics which would lead to the server quality problems in the multistage process.Then,the traveling algorithm was proposed to find the error propagation paths.The ContributionIndex was proposed to identify the error source of the key point to determine which stages need to be controlled firstly.Finally,a representative main bearing cover with multistage machining process was studied,which proved that the method could effectively model the complicated process deviation flow and identified the key weak points and the error sources in the process.

Key words: hybrid multistage machining, variation flow, complex networks, key processing characteristics, error source diagnosis

摘要： 复杂零部件生产通常需要经过混合式多阶段加工,加工过程中的误差种类多样且在多道工序间传递累积。为确保复杂产品的加工质量,提出了一种加权自调节偏差传递网络模型与误差溯源方法。该方法首先基于实际误差信息、质量特征信息与加工工艺构建加权自调节偏差传递网络。通过引入加权LeaderRank排序算法,识别混合式多阶段加工过程中易导致产品质量失控的关键加工特征。然后采用遍历回溯算法识别误差传播路径并提出ContributionIndex指标识别导致关键加工特征质量偏差的关键误差源,确定需要重点监测的加工工序及设备节点。以有代表性的具有多阶段加工过程的主轴承盖为研究对象,验证了该方法能有效建模复杂加工过程偏差流,辨别加工过程中的关键薄弱节点及其误差源。

关键词: 混合式多阶段加工, 偏差流, 复杂网络, 关键加工特征, 误差溯源