Data-Driven Based Modeling For The Inventory Problem Of Steel Cold Rolling Lines And Its Software System

Steel cold rolling has such characteristics as a complex process, large variety of products and high cost of high inventory. Each of production stages including pickling rolling, continuous annealing, hot galvanizing and cold galvanizing is concerned with inventory. In order to improve the profits and enhance the competitiveness, it is urgent for steel enterprises to establish a scientific and reasonable inventory strategy.Motivated by a large domestic iron and steel enterprise, this thesis studies the production planning and inventory strategy for each unit in the cold rolling process. To minimize the production inventory costs and the shortage cost, the problem is to optimize production management and inventory control to ensure the inventory level still within a reasonable range. On the basis of deterministic models having been built, considering material loss and uncertainty demand, a more realistic robust inventory model and robust optimization inventory model with the production and inventory capacity constraints are built based on data driven of the plenty of historical data and robust optimization theory. The main research contents are as follows:(1) Analyzing the relationship between cold rolling production and inventory, we refine the cold rolling inventory problem. Combining with the characteristics of cold rolling production, the reason for the inventory and the purpose of the inventory control are studied. Cold rolling inventory optimization factors are put forward.(2) Discuss the structural of cold rolling inventory problem based on data driven. The cold rolling inventory problem with the unit material loss and the uncertainty demand is considered in view of actual background. A large number of actual data in the steel enterprise is collected and processed. We use least-squares fitting to compute the unit material loss rate and statistical methods (i.e., expectation and variance) to analyze the uncertainty demand information to provide data for subsequence modeling.(3) Based on the actual inventory in cold rolling production process, we formulate the cold rolling inventory model that is to minimize the production inventory costs and the shortage cost. A deterministic nonlinear model for cold rolling inventory is established. Through analyzing the structural characteristics of the model, it can be equivalent converted to a linear programming model. Then, in view of the demand uncertainty, the robust optimization inventory model is built. Through constructing auxiliary problems, adding the uncertainty threshold constrains, and using the duality theory, the min-max model is reduced to the mixed integer linear programming model. Further, the robust inventory model with production and inventory capacity constraints are built. Finally, CPLEX optimization software is used to solve the above inventory models based on the actual steel production data and we discuss the results.(4) According to the study of cold rolling inventory problem and the requirements of the actual production in the steel enterprise, we design and develop the cold rolling production inventory performance management decision support system. On the basis of system analysis and design, each function module of the system is programmed. Under the guidance of the statistical optimization, the system has many highly intelligent data processing functions, such as production data statistics, performance indicators allocation and performance appraisal.