Two-sided Matching Theory Based Research On Open-order Coil Allocation Problem In The Steel Industry

In order to improve the open-order coil utilization,this thesis studies the open-order coil allocation problem under the framework of two-sided matching theory,combining with data analytics and optimization technology.This research is not only an exploration of data analytics and optimization techniques,but also a specific application and extension of two-sided matching theory in the steel industry,which contributes to improve open-order coil utilization and the ability of on-time delivery,so as to increase the profit of enterprise and improve market competitiveness.With the difference of traditional coil allocation problem,a coil-order two-sided matching model with consideration of order satisfaction and coil satisfaction is formulated,in which the predicted order completion time is added as a key indicator of coil and order satisfaction.To obtain high quality initial solutions,an order satisfaction preference based heuristics as well as a history matching relationship based heuristics is proposed.Then,an improved differential evolution algorithm based on the matching stable theory is designed to generate near-optimal solutions for the problem in a short time.Finally,based on the proposed models and algorithms,a decision support system is developed.The main contents of this thesis are summarized as follow.(1)The order completion time prediction problem is firstly transformed into a prediction problem of order processing time in cold rolling lines.Then,based on the analysis of the key process parameters in cold rolling lines,three kinds of key processing parameters including order information data,order process data and quality management data are selected as input data.A DE-MK-LSSVM algorithm is developed to build the prediction model.Considering the problem involves physical factors as well as man-made factors,an input variable re-selection strategy is proposed and introduced into the algorithm to filter the useless input variables and improve prediction precision.(2)For the open-order coil allocation problem,a coil-order two-sided matching model is formulated to allocate the open-order coils to orders as well as to decide the allocation proportion,with the consideration of coil-matching satisfaction,coil-utilization satisfaction,order-matching satisfaction and order-capacity satisfaction.The forecast completion time of order is also introduced into the model as coil matching-satisfaction and order-capacity satisfaction to reduce the uncertain risk.Finally,the effectiveness of the model is tested by CPLEX.(3)Due to the fact that the traditional optimization solver cannot effectively solve the practical instances of open-order coil allocation problem,an improved differential evolution algorithm based on the matching stable theory is developed.Firstly,by defining a Euclidean distance to measure the matching quality,a history matching relationship based heuristics is designed.Then,based on G-S delay accept mechanism,a matching satisfaction preference based heuristics algorithm is designed.Finally,an improved differential evolution algorithm is proposed to obtain near-optimal solutions for the large scale instances of the problem.In the algorithm,a non-integer with modulo mutation operation and a gene segment crossover operation are designed.In addition,based on matching stable theory,gene-value-searching strategy is proposed to improve the algorithm.The performance of the proposed algorithm is evaluated by comparing it with CPLEX.The computational results show that the proposed improved differential evolution is effective.(4)Based on the practical operations and requirements of the coil-order allocation process,a decision support system that embeds all proposed models and algorithms is designed and developed.The application of this system can significantly improve the work efficiency of planners and makes the coil allocation decisions more flexible.