Study On Modeling Method Of Magnesite Grade Classification

Magnesium is an important industrial raw material.In recent years,with the development of industry,the demand for raw materials of magnesium is growing rapidly.Magnesite is one of the main sources of magnesium,the traditional mineral processing and mining methods lead to the rational development of magnesite resources have been destroyed,the quality of magnesite ore declining,high-quality magnesite become less and less,Low-grade magnesite waste a serious,or even discarded.High-quality magnesium products for the increasingly high demand for magnesite,especially in refractory materials.At present,the original standard specified magnesite magnesite has been very rare,making the importance of magnesite grade classification become more and more important.Near Infrared Spectroscopy(NIRS)is an efficient and rapid modern analytical technique,it combines the latest research achievements of computer science,spectroscopy and chemometrics,which has been widely used in many fields with its unique advantages.Near-infrared spectroscopy uses different types of H-containing in magnesite have different absorption characteristics to determine the composition of magnesite and its content.Extreme learning machine is a single-hidden layer feedforward neural networks,which runs fast and have fine generalization performance.This paper presents a binary classification model of magnesite based on near infrared spectroscopy combined with PCA-ELM.In this paper,near-infrared spectroscopy(NIRS)data of magnesite samples were collected from the magnesite of Dashiqiao,Yingkou,Liaoning Province.The principal component analysis(PCA)was used to reduce the dimension of the magnesite ore,and establish the binary classification mathematical model of magnesite grade classification based on the extreme learning machine(ELM)algorithm.The selected ELM model is proposed on the basis of the original algorithm model.The simulation results show that the classification accuracy of the improved model is further improved.A multi-classification algorithm model of magnesite grade SAE-ELM neural network based on stack autoencoder network combined with extreme learning machine is proposed to further demand for the classification of magnesite ore on the market.Compared with the traditional chemical method,artificial method,the SAE-ELM magnesite grade multi-classification model can achieve better effect in speed and accuracy.It can achieve cost reduction and efficiency improvement in the magnesite ore grade detection,which provides a new path for the magnesite grade classification.