Study On Modeling And Optimization Algorithm Of Production Realtime Scheduling In Steel Plants

Under the background of promoting manufacturing industry with information technology all over the world,applying intelligent manufacturing technology into steel industry in China and controlling effectively the production process will become one of the important means to reduce production costs,improve production efficiency and enhance the core competitiveness of iron and steel enterprises.In order-driven steel manufacturing mode,planning and scheduling is the core technology which is used for coordinated control of steel manufacturing process.With the application of information system in iron and steel enterprises,the planning and scheduling of steel manufacturing process have been widely concerned.Great progress has been made in related researches.However,from the perspective of practice,there is still a big gap between the theoretical research results and the actual production demand.Therefore,in order to obtain feasible and effective production schedules in the dynamic environment to optimize production logistics and assist decision making of production organization,the aim of this paper is to explore a set of flexible and adaptable modeling and optimization methods for realtime scheduling in steel plants.The main research contents and results are described as follows.Based on the theory of metallurgical process engineering,through analyzing the operating characteristics of the stages and the whole steelmaking production process,a new mode is proposed for decomposing the realtime production scheduling in steel plants as integrated charge batching and casting start time determining,production scheduling and dynamic scheduling for handling disturbances.The procedure of the new scheduling mode is described as follows: the choosing and sequencing of charges in each cast and the casting start time of each cast are first determined by synthetically considering the supply condition of hot metal,the due date of rolling and the production organization characteristics of continuous casting;On this basis,by using the charge as the basic scheduling unit in steel plants,the machine allocations and the starting times and ending times on the allocated machines of each charge are determined.The obtained schedule will be used to guide the actual production organization;Considering the occurrence of random disturbances may cause the initial schedule to become inefficient,in order to maintain the stability and continuity of the production process,dynamic scheduling is needed to adjust the schedule or make a new schedule by using the realtime scheduling information upon the occurrence of disruptions.A hierarchical modeling method and optimization algorithm based on variable neighborhood search(VNS)is proposed for solving the integrated charge batching and casting start time determining(ICBCSD)problem.The number of casts on each continuous caster and the number of charges in each cast are determined in the encoding process of the algorithm.In addition,five neighborhood structures are designed for the iterative optimization of solution.The decoding process is decomposed into two sub-problems,which are charge batching problem and casting start time determining problem.A mixed integer programming model is built for charge batching problem by drawing on the modeling methods for prize collecting multiple traveling salesmen problems,and a VNS algorithm is proposed for solving the model.In addition,a mixed integer programming model is built for casting start time determining problem by drawing on the modeling methods for parallel machine scheduling problems,and the model is solved by CPLEX directly on the basis of designing six valid inequalities.Finally,several experiments are conducted to analyze the performance of the model and algorithm.Experimental results show that the model can always obtain feasible and effective solutions for different production conditions and data.Moreover,the algorithm has good optimization performance because it can obtain sub-optimal solutions in a short time.In order to satisfy the requirements of modern steel plants to produce different steel grades simultaneously and improve the flexibility of scheduling decision,the production scheduling problem considering with stage skipping and adjustable processing time(SCCSA scheduling for short)in steel plants is studied in this paper.Through clarifying the production objectives and constraints related to stage skipping and adjustable processing time,a new SCCSA scheduling model is built.An improved genetic algorithm(GA)is developed for solving the model,which involves:(1)calibration of a GA by comparing and analyzing four encoding methods,two selection operators and three crossover operators,which are effective and widely used in regular hybrid flow shop scheduling problems;(2)development of a quality improvement approach embedded into the GA to further optimize each solution obtained by decoding heuristic;(3)improvement of the GA by introducing a new elitist strategy and a restart strategy to further accelerate the local optimization and avoid premature convergence.Computational experiments based on instances generated from practical production process show that the model can obtain feasible schedules for the production scheduling problems with stage skipping and adjustable processing time in steel plants.In addition,the effectiveness of the quality improvement approach and improved evolutionary strategy for solving SCCSA scheduling problems is demonstrated in the experiments.With respect to dynamic scheduling problems,four kinds of disturbances are first proposed to classify the various uncertainties occurred in the implementation of the production schedule.The effect of any uncertain event on the execution of schedule can be represented by one combination of the four kinds of disturbances.In order to realize the optimal decision of realtime production scheduling in steel plants,a predictive–reactive scheduling strategy in periodic rolling mode is designed by combining the results of ICBCSD study and SCCSA scheduling study.The rescheduling method for handling disturbances is the key technology in the dynamic scheduling strategy.Therefore,a mathematic programming model is built for the rescheduling problem.Moreover,a hierarchical rescheduling method is proposed to guarantee the realtime performance and optimization performance of rescheduling process.The hierarchical rescheduling method contains two aspects: local rescheduling(LRS for short)and rescheduling for the charges in those casts that have already been processed in the production process(RSS for short).LRS is used to allocate a processing machine and determine the processing time on this machine for the next operation of each charge that are being processed in the production process.Several scheduling rules are designed for LRS in the paper.RSS is used to make a new schedule for all the charges in those casts that have already been processed in the production process.In this paper,a hybrid algorithm combining a genetic algorithm with a local search(GALS for short)is proposed for RSS.By using the schedule obtained in the application experiment of SCCSA scheduling as the initial schedule,simulation experiments are carried out by means of setting production disturbances.The experimental results show that the dynamic scheduling model and algorithm can obtain executable and effective reschedule for different kinds of disturbance.In addition,several experiments are also conducted to analyze the performance of GALS.The experimental results show that different neighborhood structures defined for infeasible solutions and feasible solutions according to the production characteristics of continuous caster and the hybrid strategy that the local search is embedded into the genetic algorithm are both effective for solving rescheduling problems in steel plants.Finally,on the basis of the above research results of modeling methods and optimization algorithms,a production realtime scheduling system is designed and developed for the vanadium-extracting and steelmaking plant of panzhihua iron and steel corporation(SPPISC for short).The system is composed of data interface module,modeling module,data analysis module,logistics tracking module,ICBCSD module,SCCSA scheduling module,rescheduling module,plan simulation module,plan evaluation module and human-computer interaction module.The system can exchange data with the MES system of SPPISC by using the data interface module.System application tests on practical production data of SPPISC show that system decision has a better performance than manual decision in reducing the cost of continuous casting in casts,shortening the flow time of the whole production process and maintaining the stability of processing time of charges.In conclusion,based on the new proposed mode of production realtime scheduling in steel plants,the proposed mathematical models of ICBCSD,SCCSA scheduling and dynamic scheduling provide a set of new methods for realtime optimization of production logistics in steel plants.Moreover,the developed realtime scheduling system provides a scientific and effective assistant decision-making tool for the production organization arrangement of steel plants.