A new model based on improved PSO and BP to predict silicon content in hot metal

Abstract

In the smelting process of blast furnace, maintaining the temperature at an acceptable level is the key to ensure the smelting at a good level. The hot metal Silicon content is not only the indication of the blast furnace thermal state and its changes, but also the significant indicator for assessing the blast furnace's stability and the quality of iron. Therefore, as the core content of automatic control of blast furnace, it is crucial to create a model to predict Silicon content of hot metal. Using the online data of a steel company's blast furnace, a new model based on improved Particle Swarm Optimization (PSO) and Back-propagation (BP) is proposed to predict Silicon content of hot metal in blast furnace. As a new bionic algorithm, the improved PSO has gained very good performance in some classical optimization problems. Its properties such as fast searching, global searching have well conquered the long convergence speed and premature problem, which are the main deficiencies of BP algorithm. Compared with pure BP algorithm and basic PSO, experimental results show the model proposed has good performance in predicting Silicon content of hot metal.