Data-driven automated job shop scheduling optimization considering AGV obstacle avoidance.

The production stage of an automated job shop is closely linked to the automated guided vehicle (AGV), which needs to be planned in an integrated manner to achieve overall optimization. In order to improve the collaboration between the production stages and the AGV operation system, a two-layer scheduling optimization model is proposed for simultaneous decision making of batching problems, job sequences and AGV obstacle avoidance. Under the AGV automatic path seeking mode, this paper adopts a data-driven Bayesian network method to portray the transportation time of AGVs based on the historical operation data to control the uncertainty of the transportation time of AGVs. Meanwhile, a time window is established to control the risk of AGV delay, and a data-driven Bayesian network is constructed to optimize the two-layer scheduling model of automated job shop and AGV. To solve the model, we design an improved particle swarm algorithm combining genetic operators, crossover operators and elite retention operator. The results show that the model in this paper can effectively improve the collaboration between the production stage and AGV operation system within the shop floor, and successfully solve the actual operation scale case to enhance the effectiveness of the production and transportation system.