Classification and discrimination system in metallurgical balling process

Abstract

The invention relates to a classification and discrimination system in a metallurgical balling process. The system mainly consists of a cluster and classification module and a discrimination and optimization module, wherein the cluster and classification module comprises a dynamic cluster unit and an SVM classification unit; a group of initial labels are obtained in a dynamic cluster manner and then an SVM classification model is constructed on the basis of the initial labels; the discrimination and optimization module mainly comprises a discrimination unit and a Fisher optimization unit; the discrimination unit discriminates classification results obtained by SVM classification and the last adjacent classification, if the two classification results are consistent, an optimal classification manner can be directly obtained, and the two classification results are inconsistent, the Fisher optimization unit optimizes a basic class by utilizing a Fisher discrimination principle and constructs the SVM classification model by taking the basic class as an initial label; and through repeated SVM classification and Fisher discrimination and optimization, the optimal classification is finally obtained, so that the classification, discrimination and screening for balling samples are completed.

Description

technical field [0001] The invention relates to a classification and discrimination system in the process of metallurgical pelletization, in particular to a classification and discrimination principle based on SVM-Fisher, which optimally classifies the pellet samples produced by metallurgical pelletization, and selects pellets that meet the requirements. group system. Background technique [0002] In the metallurgical industry, pellet production is a process in which finely ground iron concentrates are made into block materials that can meet the needs of blast furnace ironmaking. The iron concentrate is added with a binder according to a certain proportion and mixed evenly, and then rolled into a certain proportion of green balls by a pelletizer, and then dried and roasted to consolidate the pellets. The trend of blast furnace production aimed at low fuel consumption and high yield has been increasing, and the quality requirements for blast furnace charge are getting higher...

AI Technical Summary

Problems solved by technology

[0004] The division based on artificial standards can have 100% reliability for a single index, but when comprehensively evaluating the performance of pellets in the pelletizing test, the number of indicators and the dimension of the sample will expand, and the performance of pellets will be classified artificially is difficult to achieve

Images