A Mechanical Job-shop Task Scheduling Support System Based On Reinforcement Learning For Energy Saving And Optimization

Manufacturing industry is considered as a high energy consumption industry,reducing its energy consumption should be motivated not only by the increasingoperating expenses of the manufacturing systems but also by a proportional reductionof the production of greenhouse gases. With the global energy shortage, globalwarming as well as the globally implement of relevant laws and regulations, themanufacturing industry had to reduce energy consumption (especially electricenergy).Therefore, reducing energy consumption is regarded as one of the strategies toimprove sustainable manufacturing.Mechanical workshops are conceived as manufacturing systems which arecomposed of machine tools, They are used for machining task and consume lots ofenergy. Through the reasonable scheduling of production tasks in the productionprocess, the energy consumption can be saved for the production process ofmechanical workshop. Therefore, the paper studied the energy saving and optimizationproblem of machining task in mechanical workshop, and presented a mechanicaljob-shop scheduling support system based on Q-learning for energy saving andoptimization. The study includes the following sections:First, the impact on energy consumption from mechanical shop scheduling isanalyzed, and the mechanical job-shop scheduling method frame based on Q-learningfor energy saving and optimization was proposed. On this basis, the architecture,functional modules and databases of the system was designed.Then, some key technology of the system was studied, which including the energysaving and optimization algorithm based on reinforcement learning and interactiveGantt chart controls. The former studied including the state space, action strategy,penalty function, algorithm flow as well as the learning environment; the latter studiedthe design and implementation of interactive Gantt chart controls and interactivealgorithm.Finally, on the basis of the above study, mechanical job-shop scheduling supportsystem for energy saving and optimization was developed by SQL Server2005andVS2005development tools. A case study was implemented to compare the makespan,energy consumption and weighted target value between before and after optimization,which showed the effectiveness of the system.