基于贝叶斯网络的制造情景识别

Manufacturing situation recognition method based on Bayesian network

针对制造资源等个体状态及其状态组合复杂多变、个体状态与制造情景间无线性逻辑关系,导致难以直接建立个体状态与制造情景之间的映射,提出基于贝叶斯网络的制造情景识别方法。该方法利用个体状态与制造情景在历史数据中体现的统计学规律,建立了个体状态与制造情景在概率上的因果关系;充分考虑了连续状态与离散状态并存的混合性,通过引入等区间离散法对连续状态进行处理;通过最大后验概率推理确定制造情景,实现了对制造情景的判断。通过案例验证了所提方法的有效性。

Aiming at the problem that the manufacturing resource s individual states and their combinations were complex and variable,and there was no linear logic relationship between individual states and manufacturing situations,which led to the difficulty to directly map individual states to manufacturing situations,a recognition method for manufacturing situation based on Bayesian network was proposed,in which the probability causality between individual states and manufacturing situations was established.The proposed method considered the coexistence of continuous states and discrete states fully,and the equal-interval discrete method was used to deal with continuous states.Through the maximum posterior probability,the manufacturing situation was confirmed.The effectiveness of the proposed method was verified by a case study.