Prediction Model Of Pellets Quality Based On Data And It’s Application

As the development of iron and steel industry, the materials into furnace are required to be better and better, which improves the development of pellet production. While the main purpose is to produce pellet of high quality. And Grate-kiln is a new process of pellet, which is utilized widely all over the world for its superiority on pellet production. But it’s a complex process including heat radiation, conduction, convection and various kinds of chemical reaction, in which the parameters are coupled each other. In additon, there are some uncertain factors such as the faluctuation of pressure and calorific value, which make the producting process being a large lag, time-varying and strongly coupled system. So it’s difficult to established the mechanism model of the process, which make it unfavorable for the automation of pellet production. And now most pellet plants rely on manual operation by experience, which is inaccurate.As a rusult of this, quality prediction model of manufactured pellet is established in this paper to determine the relationship between quality indicators and parameters which is important to improve the quality and convenient to direct the production, and at the same time some problems are solved for traditional BP algorithm, such as conversing slowly and diverging easily and so on, which increases the hit rate.At first, the development of grate-kiln and current status of study on modeling are analyzed. And then the technology process of Grate-kiln and quality index of manufactured pellet are presented in this paper. And as followed, some common methods of modeling are listed in detail including Neural Network, Genetic Algorithm and Particle Swarm Optimization. Besides, BP, GA-BP, PSO-BP are focused. And next, the quality predictiong models are established according to this three algorithms respectively and simulated by MATLAB to verify its accuracy and compare their advantages and disadvantages. At last, a quality prediction system of Grate-kiln is desighed and developed based on VS2008, MATLAB and Microsoft Access, which is proved to be useful to improve the quality of pellet.