Research On Optimization Methods For Hot-rolling Production Planning Of Round Bar

As an important steel product, the hot-rolling production of round bar has its own characteristics. It is considered that the hot-rolling planning of round bar is a difficult problem, because so many factors need to be taken into consideration. To achieve the lean management for hot-rolling production of round bar, making a reasonable and effective hot-rolling plan is very important in both theory research and practical application. However, the researches on hot-rolling planning of round bar were rare. Further, existing theories and methods cannot be used directly. In this dissertation, considering the characteristics of round bar, and the focus is on the basic properties and solving algorithms for the hot-rolling planning under static and dynamic environment. The main work and contributions of this dissertation are summarized as follows:(1) To reduce material waste in the hot-rolling production, a hot-rolling batch planning problem of round bar from the view of order requirements was researched. A mathematic model was built with two optimization objectives, the minimization of total surplus length and the billet bounce cost. To deal with the complexity of the model, an algorithm based on constraint satisfaction technique was proposed to determine the relationship of rolling unit, billet and round bar. The algorithm was solved by variable selection rules and value selection rules for billet selecting, grouping and sequencing, and constraint propagation for searching-space reducing and rolling-unit partitioning. Meanwhile, BFD heuristic was embedded into the algorithm to optimize the two objectives. The simulation experiments with practical production data demonstrate the feasibility and effectiveness of the model and algorithm.(2) Considering the influence of process conditions and due date on the continuous of hot-rolling production, a mathematical model for batch scheduling problem of round bar with three objectives to minimize setup times, tardiness penalties and steel grade bounce penalties was established. A serial solving strategy for the importance of three different objectives was proposed. A method of variable neighborhood search algorithm embedding with earliest due date first rule and steel grade sorting rule was designed. First, the due date of rolling units were calculated, and an initial solution was generated by constraint satisfaction technology. Then, a variable neighborhood search method was presented to minimize setup times, which was viewed as the primary objective. Further, a hybrid operator was applied in shaking and local search, and the idea of simulated annealing was introduced to take control of the acceptance of new solutions. The Earliest Due Date first rule and steel grade sorting rule were used to minimize tardiness penalties and steel grade bounce penalties. Experiment results show that the model and algorithm are feasible and effective.(3) Considering the dynamic changes of roller life in the hot-rolling production of round bar, the optimization index and constraints about rollers were analyzed. A mathematical model with two objectives to minimize setup times and maximize rollers’continuous rolling length was developed, and a multi-objective variable neighborhood search algorithm based on pareto-optimal was designed. Firstly, billets and rollers were pretreated by the rolling schedules. Then, aiming at the multi-objective optimization features, a selection strategy of base solutions was proposed to guide the search process in the algorithm. An intensification procedure with destruction and construction was employed to enhance the diversity of non-dominated solutions. Meanwhile, rollers were reallocated by rolling schedules; the global optimal set was evaluated and updated by the pareto-optimal strategy. Simulation experiments with practical production data evaluate the feasibility and effectiveness of the model and algorithm.(4) With the consideration of the random factors influence on planning execution, a key-events driven rule based optimization strategy for the rolling planning optimization problem of round bar was presented. To ensure the consistency with the initial plan on productivity and efficiency, a dynamic constraint satisfaction model was established to minimize the difference of planning length between initial plan and rolling plan. Besides, a local repair algorithm based on dynamic constraint satisfaction technique was designed. According to the characteristics of round bar, variable selection rules and value selection rules were proposed to repair the unfeasible rolling units by unscheduled billets. Meanwhile, the initial domain was reduced by constraint propagation, constraint checking and rebound mechanism were employed to eliminate conflict constraints, and the solution was optimized deeply by a local repair strategy. Numerical results demonstrate that the performance of the algorithm is remarkable in both solution quality and solving efficiency.