Research On Multi-AGV Autonomous Dispatching Technology Based On Internet Of Manufacturing Things

In recent years,under the development trend of industrial Internet of Things and the deep integration of information technology and manufacturing industry,the application field of AGVS has been expanding,and the requirements for automation and intelligence of AGVS are getting higher and higher.Traditional centralized scheduling method is difficult to adapt to this development trend.This paper studies the optimization of task assignment and path planning,the conflict resolution of multi-AGV system in discrete workshop environment based on IoT manufacturing.Firstly,based on the manufacture of IoT,a centralized-distributed hybrid control structure for multi-AGV autonomous dispatching system is proposed.The functional structure of the centralized control system and the functional structure of the carrier with sensing and interaction capabilities are designed respectively.Secondly,aiming at the problem of multi-AGV system task assignment based on automated meter verification workshop model in manufacturing environment,the shop task execution process is analyzed and a dynamic task allocation method based on real-time information interaction is proposed.AGV independently applies to determine the handling tasks to improve the defects of traditional centralized controllers for task assignment and optimize traditional heuristic dynamic scheduling allocation methods.Relying on the information exchange between the AGV and the centralized control system and between the AGVs,reaching the starting point of the higher priority task in the shortest time is the allocation principle,thereby improving the output efficiency of the system.Finally,the effectiveness of the proposed method is verified by simulation experiments.Thirdly,aiming at the problem of AGV independent planning path,a multi-AGV system path planning method based on IoT manufacturing combined with reinforcement learning is proposed.Q value neural network and its parameters are generated using DQN reinforcement learning techniques in the offline phase.During the path planning phase,the AGV senses locations of other AGVs in real time,and uses the neural network to generate the next pre-driving path,which is strong in real-time and saves task execution time.The conflicts that AGV may encounter are classified,and solutions are provided for various types of conflicts to avoid conflicts at run-time.Finally,the scheduling key technology proposed in this paper is applied to the multi-AGV system of a provincial power metering verification center.The scheduling system software is developed and tested.Finally,the effectiveness of the proposed strategy is verified by system simulation and actual operation.