Research On Resource Matching Evaluation And Dynamic Task Planning For Production-logistics-integrated System In Autonomous Workshop

The current manufacturing industry has exhibited characteristics such as product personalization,dynamic customer demands,and complex production operations.The traditional centralized production operation mode is no longer able to meet the requirements of workshop production flexibility and agility in response to constantly changing internal and external factors,which can lead to problems such as low operating efficiency and inadequate dynamic response capabilities.The extensive interconnection and autonomous collaboration of production elements within the manufacturing workshop have become the key to solving the above problems and moving towards intelligent manufacturing,resulting in bottleneck problems such as data exchange and collaborative computing,manufacturing resource evaluation and selection,and real-time response to production processes.This paper takes the customized production-oriented autonomous workshop as the research object,considering the interdependent operation relationship between production and logistics within the workshop,constructs an autonomous workshop production-logistics-integrated system featuring manufacturing resource state awareness and collaborative decision-making.Through the interconnection and empowerment of various production elements,and the intelligence based on their autonomous capabilities,manufacturing resources can collaboratively complete manufacturing tasks in a distributed manner.The main research work includes:Firstly,in response to the problem of large-scale data transmission and computational pressure at the bottom layer of the interconnection environment in autonomous workshop,which can lead to delayed production decision-making,this paper considers offloading some computational tasks to the edge for processing.It constructs an autonomous workshop Industrial Internet-of-Things(IIoT)architecture based on the cloud-edge-end collaboration and designs edge nodes for production,logistics,warehouses,and operators.Based on this,a cloudedge-end collaborative multi-task offloading problem model is established,and an improved particle swarm optimization algorithm based on penalty functions is used to formulate the computation resource allocation strategy.The allocation of massive computation tasks is balanced among the cloud center,edge nodes,and terminal devices.The results show that as the computational task volume,number of terminals,and transmission rate between edge nodes and cloud center gradually increase,the proposed strategy in this paper is lower than the Local,Edge C,Cloud C,and RAN strategies,effectively improving the efficiency of data task computation in the IIoT environment of the workshop.Secondly,to address the issue of increased decision-making complexity and uncontrollability caused by considering the complete set of manufacturing resources in production-logistics task planning,this study focused on the manufacturing resource matching and candidate evaluation method for specific production-logistics processes.Based on the decomposition of multi-resource collaborative production tasks at the part level,a temporal production-logistics task information model and a corresponding manufacturing resource information model were constructed.Using the triangular fuzzy number-TOPSIS/VIKOR method,a multi-attribute decision evaluation method for "task-resource" matching was established to obtain the best candidate set and weight of manufacturing resources matched with specific production-logistics processes.The results show that the triangular fuzzy numberTOPSIS method and the triangular fuzzy number-VIKOR method established in this paper have consistent selection and ranking of manufacturing resource candidate sets,which can serve as effective input conditions for subsequent dynamic production-logistics task planning,providing support for reducing the complexity and improving the efficiency of dynamic task planning.Thirdly,in response to the insufficient responsiveness to dynamic user demand changes and production process disturbances in current manufacturing workshop production planning,this study considers the autonomy of manufacturing resources in autonomous workshop and proposes a real-time data-driven production-logistics dynamic task planning method.Based on real-time manufacturing data acquisition,on the one hand,a real-time data-driven autonomous workshop production-logistics dynamic task planning model was established with the objectives of maximizing resource utilization and minimizing completion time,and solved using reinforcement learning algorithms.On the other hand,based on the above dynamic task planning results and historical manufacturing correlations,a manufacturing resource autonomous collaboration network model was established based on complex network theory,and the corresponding topology and engineering indicators were designed to analyze the network characteristics,providing feedback information and decision-making basis for optimizing manufacturing resource matching evaluation and dynamic task planning.The results show that compared with the GA-based centralized production-logistics planning method,the proposed distributed production-logistics dynamic task planning based on reinforcement learning reduces the maximum completion time by 4.3%,shortens the waiting time for all orders by 16.3%,and improves resource utilization by 172.5%.Finally,based on the research on the above key technologies,a self-contained workshop production-logistics integrated system platform for customized product manufacturing was developed using the Python language and My SQL database.The system was validated through application testing in the automotive synchronizer manufacturing workshop,demonstrating the feasibility and effectiveness of the proposed models and methods in this paper.