Abstract: Material kitting is an important guarantee and prerequisite for the efficient and orderly conduct of the assembly process of complex products. With the continuous improvement of digital and network capabilities in assembly workshops, the traditional material kitting mode in assembly processes is undergoing a shift from product-level kitting to process-level kitting. This shift poses higher demands for the accuracy of material demand forecasting. Due to the extensive manual operations involved in the assembly process, and the frequent occurrence of mixed-flow co-linearity between developmental models and batch production models, traditional simulation methods face challenges in predicting process completion times, consequently impacting material demand forecasting. To better support production scheduling in the assembly process and ensure the orderly execution of complex assembly tasks involving multiple varieties and small batches, a method for predicting the process-level kitting time of assembly materials for a specific type of anti-aircraft missile complex product based on digital twins is proposed. This involves constructing a digital twin model tailored to the assembly process of complex products, providing a simulation environment for the time consumption of tasks in each stage of the assembly kitting process. Based on the simulation results data of the assembly kitting process and combined with real-time production data of assembly tasks, design a material process kitting time prediction algorithm model. By capturing the relationship between workshop status and manual efficiency through historical production feature data, enhance the accuracy of predicting the kitting time for pending assembly tasks. Finally, an application demonstration is conducted in a specific anti-aircraft missile assembly workshop to validate the effectiveness of the proposed method.

Key words: material kitting time, digital twin workshop, virtual simulation, prediction model, assembly workshop