离散制造车间的数字孪生仿真参数修正方法

Parameter Correction Method of Digital Twinning Simulation in Discrete Manufacturing Workshop

为了提高数字孪生模型的准确度,提出了一种离散制造车间的数字孪生仿真参数修正方法。根据数据驱动仿真参数修正的方式,将数字孪生模型的仿真参数划分为静态属性、动态属性和性能属性3类,设计了一种基于深度学习的时间序列预测算法—DF-LSTM用于表征性能属性。在复杂离散制造车间的仿真模型基础上,用时间序列预测算法的预测结果作为仿真模型的性能属性值,以实时数据驱动仿真模型的动态属性和性能属性的更新,实现了由仿真模型向数字孪生模型的转变。开发了装配车间的数字孪生系统,实现了装配车间的可视化监控和数字孪生模型的在线运行,最终实验验证了方法的可行性。

In order to improve the accuracy of digital twinning model,this paper presents a digital twinning simulation parameter correction method for discrete manufacturing workshop.According to the updating method of simulation parameters driven by data,the simulation parameters of digital twin model were divided into static properties,dynamic properties and performance properties.A time series prediction algorithm based on deep learning,DF-LSTM,was designed to characterize performance properties.Based on the simulation model of the complex discrete manufacturing shop,the output of the time series prediction algorithm was used as the performance properties of the simulation model,and the real-time data was used to drive the update of the dynamic and performance properties of the simulation model,which realized the transformation from the simulation model to the digital twin model.A digital twin system for the assembly shop was developed,the visual monitoring of the assembly shop and the online operation of the digital twinning model were realized.Finally,the feasibility of the method was verified by experiments.