Analytics For The Quality Of BOF Steelmaking Process And Production Cycle Of Contract

How to improve the quality of steel products and how to achieve the effective control of the production cycle are two difficult problems faced by iron and steel enterprises in today intense market competition. The former problem can be solved by operation optimization, process control and some other methods that aim to improve the production process; the latter one can be solved by the effective prediction of the production cycle. They complement each other since they can enhance business efficiency and improve competitiveness from the "quality" and "quantity" point. Basic oxygen furnace (BOF) steelmaking is a very important process in the steel production. In the BOF process, steel making quality is guaranteed by the adding policy of the oxygen and other materials which is finally determined by the smelting state of the converter directly reflected by the trend of carbon and temperature. Therefore, the temperature and carbon content prediction of molten steel is of great significance in improving the steel making quality. On the other hand, all the work of the iron and steel enterprises is commenced surrounding the steel contract and the production capacity, and the management of the complete cycle of the contract is a very important task in the production management. Prediction of contract completion time is helpful to grasp the overall contract process, to guide the institutional arrangement and to adjustment of production plan. In this thesis, analytical studies are conducted on the prediction of the molten steel temperature, carbon content and the complete time of contracts respectively by establishing analytical model based on the practical data of the corporate. The main contents of this thesis includes the following sections:(1)A study on the prediction of the temperature and carbon content of the molten steel is implemented. A least squares support vector machine based on the improved particle swarm optimization algorithm is designed and prediction models for the molten steel temperature, carbon content are established respectively. In the modeling process, multi-stage modeling approaches used so as to realize the dynamic prediction. Interpolation algorithm is used to solve the problem that when the practical data is incomplete. And finally the accuracy of the multi-stage dynamic forecasting model is verified by performing several experiments.(2)A study on the prediction of contract completing time is then conducted under the iron and steel production cycle management background. Substantial contract data is produced by the marvelous categories of steel products and miscellaneous and complicated manufacturing processes. And the focus of this paper is how to predict the complete time of contracts fast and accurately according to contract type and characteristics. In this section, the least squares support vector machine model based on the improved particle swarm optimization algorithm is established by training the historical data with the data analytics method. And the validity of the model is verified by the analysis of several experiments which are conducted on actual data.(3)A steel contract management subsystem is developed based on the actual contract management of a steel to predict the contract completing time and complete the statistics on all kinds of current contracts. The system not only achieves the monitoring and analysis of contract data in production process, but also is compatible with the KPI performance management system. It realized the performance evaluation function by making the contract management index as the performance evaluation of the employee.