Research On Prediction Model Of Physical Performance Of Pellets

As one of the artificial rich ores, iron ore pellet has become an indispensable metallurgical material because of its particular metallurgy characters. As the blast furnace material, the pellet should meet following three-aspect standard:physical performance, chemical characteristics and metallurgical characters. Physical performance is an important indicator of its quality and consists of compressive strength, the drum index and abrasion index, screening index.Generally, physical performance of the products is examined after they were produced. As the access to get quality data for long time, it is not beneficial to the quality control of the pellet. Therefore, the research of the prediction model for the physical performance of pellets can be early to understand the quality status of pellets, which are targeted to adjust the process parameters to ensure the quality of pellets.Based on the research of the grate-kiln product technics and a detailed analysis of the factors affecting the compressive strength,the bentonite content, the temperature of preheat one and two, preheating time, the temperature of head and end of kiln, calcination time and cooling rate are the input of the model. The mathematical method based on neural network prediction model of compressive strength is established in the paper.According to the consolidation mechanism of pellet and R.Jbatterham’s quality model, relying on actual data and documentation, the unknown model parameters are determined, so a prediction model of the mechanism of abrasion index is established in the paper. And based on the documentation, a mathematical method based on neural network prediction model of drum index is established in the paper too. This paper is based on the project of Ansteel, for analyzing the result of simulation, the models are proven to be efficient. Production practice is instructed by the models of prediction of physical performance of the paper.