Research On Performance Feedback Based Material Allocation Problem In Steel Enterprise

Based on iron and steel production process,this thesis adds the idea of prediction control to the steel material matching process,and studies the performance feedback of the material allocation problem.In contrast to the open-loop structure of traditional materials allocation research,this thesis introduces the analytics of key performanceindex asfeedback to the material matching optimization model,which constitutes a closed-loop structure and guides the adjustment of the material allocation plan.The application of this research improves the material utilization rate and the order on-time delivery capacity,which are the key performance index of iron and steel enterprise.In this thesis,under demand orders and cold-rolling unit load capacity are two key factors,which effect the delivery time,are to be predicted.The forecasting models based on the decision tree and the time series methods are formulated respectively.Based on the results of the prediction,an integer programming model of slab reallocation that considered under demand orders is formulated.And animproveddifferential evolution algorithm is designed to solve the problem.Finally,the slabreallocation and performance management system is developed.The main researchcontents are as follows:(1)Theproblem of under demanding orders prediction is studied.Based on the production process of hot rolling and cold rolling,the main factors that may cause materialloss are analyzed,and the forecasting model of under demanding orders is formulated according to the type of the products.Firstly,the original data are decoded,and then the processed data are solved by the improved decision tree model.The experimental results show that the improved decision tree model has better coverage ability than the traditional random forest and gradient boosting tree,and it can meet the requirement of accuracy of the problem.(2)The problem of cold-rolling unit load capacity prediction is studied.The forecasting model of ARMA considered production delay is formulated for the load capacityof each cold-rolling unit in a future period of time,and the input time series is smoothed.The experimental results show that the time series model is effective and the accuracy of the prediction can be controlled within an acceptable range.(3)The problem of slab reallocation with under demanding orders is studied.To reduce the risk of matching inaccuracies and improve the robustness of the model,this thesis formulates an interval programming model for slab reallocation considering under demanding orders,which is equivalently transformed into a deterministic model,and an improved differential evolution algorithm is designed to solve it.A mutation strategy and the repair strategy is developed according to the encoding method and constraints of the solution.The validity of the algorithm is verified by numerical experiments.(4)Based on the above model and algorithm,the slab reallocation and performance system is developed.The system can realize the functions of under demanding orders prediction,cold-rolling unit load capacity prediction,automatic slab reallocation and so on,which improves the working efficiency of the planners and the level of informatization in iron and steel enterprises.