Research On Scheduling Problems In Iron And Steel System Based On Artificial Bee Colony Algorithm

Iron and steel enterprises are the important pillar industries of national economy in China.Efficient optimization methods play important roles for iron and steel production schedules.The hybrid flowshop scheduling problem(HFS)is the classical scheduling problem in iron and steel processes.Based on the investigation of the three key processes of iron and steel production system,i.e.,ironmaking,steelmaking casting,and hot rolling,in this study,the mixed integer programming mathematical models are designed for each type of HFS problems.The knowledges and the structure features of these HFS problems are also investigated.The hybrid algorithm framework based on the artificial bee colony(ABC)is utilized to solve the HFS problems.In addition,several optimization and scheduling methods and heuristics are included into the hybrid algorithm framework.The main contributions of this study are as follows.1.For the molten iron scheduling process of the Baosteel Ltd,first,the hybrid flexible flowshop scheduling problem with dynamic operation skipping(HFF-D)is proposed.The mathematical model of HFF-D problem is also investigated.The two-level vector based encoding mechanism is presented,where,in the first evolution phase,the permutation-based representation is used,which can enhance the searching efficiency,while in the second evolution phase,the complex encoding mechanism is utilized,which can enhance the searching accuracy.The flexible decoding approach is proposed,which is based on the reference weight and therefore can solve the searching dead zone problem occurred by using the permutation-based representation mechanism.A well-designed job right-shift heuristic is developed,which can improve the given schedules and decrease the earliness objective value.Two-level based neighborhood structure,which combines the skipping neighborhood and scheduling neighborhood structures,is investigated to enhance the global search ability of the proposed algorithm.Furthermore,an enhanced local search approach,which is based on iterated greedy(IG),is designed to enhance the local search ability.Finally,comparative evaluations based on the molten iron scheduling problems show the performance of the proposed algorithm.2.For the steelmaking casting process,first,the HFS problem with continuous process constraints in the last process stage is proposed.The mathematical model is presented,and a hybrid ABC algorithm is developed to solve the considered problems.Then,the permutation based encoding mechanism is introduced into the algorithm.The decoding method combining the forward and backward heuristics is developed to solve the preventive maintenance constrains.The forward method is used to decrease the tardiness objective value,while the backward approach is introduced to decrease the average sojourn objective value.In addition,five neighborhood structures,i.e.,pair swap,forward insertion,reverse,multiple swap and multiple insertion,are presented.Based on a two-vector-table structure,a well-designed self-adaptive neighborhood structure selection mechanism is proposed to further balance the global and local search abilities.Finally,comparative evaluations based on the steelmaking casting process show that the proposed algorithm is better than the best performing algorithms from the literature.3.For the hot rolling scheduling process,with considering the problem features,the problem is modelled as an HFS problem with limited buffer constraints.A hybrid algorithm framework combining ABC and tabu search(TS)is developed.In the framework,the permutation-based encoding method is introduced.A novel decoding method is embedded to tackle the limited buffer constraints,which guarantees the efficiency and feasibility of the schedules.Then,a TS-based self-adaptive neighborhood strategy is adopted to increase the global search ability.Furthermore,a well-designed TS-based local search is developed to enhance the local search ability of the employed bees and onlookers.Finally,comparative evaluations based on the hot rolling process show the efficiency and effectiveness of the proposed algorithm.4.For the steelmaking casting rescheduling problem,the mathematical model is designed,which considers two disruptions,i.e.,the random machine breakdown and processing variation events.The hybrid algorithm framework combining ABC and IG is proposed.In the framework,a novel encoding method,which combines the machine assignment vector and scheduling vector,is designed.Meanwhile,an efficient decoding approach,which combines the forward and backward approaches,is developed to make the solution feasible and efficient.The forward method is used to decrease the tardiness objective value,while the backward approach is introduced to decrease the penalties of the earliness,average sojourn time,and cast breaks.Several heuristics such as processing time delay,cast break erasing,and right shift approaches are developed,which can improve the solution quality significantly.Then,based on an external archive,an efficient machine assignment neighborhood is developed to enhance the global search ability.In addition,an IG-based local search mechanism is embedded in the proposed algorithm to enhance its exploitation ability further.Finally,comparative evaluations based on the steelmaking casting rescheduling process show that the proposed algorithm is better than the best performing algorithms from the literature.