Method and system for thermal state prediction of coal blending and coking based on machine learning

Abstract

The invention discloses a method and system for predicting the thermal state of coal blending and coking based on machine learning. The method includes: constructing a derivative index of a single type of coal according to the original index of a single type of coal on the coal blending list; Clustering model, the class center value of a single coal is obtained from the original index and derived index of a single coal; the membership degree matrix calculation is performed on the class center value of a single coal to obtain the derived characteristics of a single coal; The derived features are linearly summed to obtain the derived features of the blended coal; the derived features of the blended coal are normalized, and the support vector regression model completed through machine learning training is used to obtain the CSR value of the blended coal formed by the coal blending list. Through the technical proposal of the invention, it is organically combined with the coking mechanism of coal blending to improve the accuracy and precision of coke quality prediction, improve the production quality of coke at the same time, reduce costs, and overcome the problem of difficulty in popularization.

Description

technical field [0001] The invention relates to the technical fields of coal chemical industry and data mining, in particular to a method for predicting the thermal state of coal blending and coking based on machine learning and a system for predicting the thermal state of coal blending and coking based on machine learning. Background technique [0002] Although coal blending coking technology has been developed in China for many years, coking enterprises in my country still use traditional manual experience and determine the ratio through coke oven experiments, which not only cannot effectively save high-quality coal, but also brings high The time and labor costs will have a great impact on production efficiency and reduce the competitiveness of enterprises. [0003] Although there are existing technologies for coke quality prediction in the industry, due to the large number of indicators required, for most coking enterprises, it may not be possible to provide sufficient ind...

AI Technical Summary

Problems solved by technology

[0004] In view of the above problems, the present invention provides a method and system for predicting the thermal state of coal blending and coking based on machine learning, which predicts coke CSR by cooperating with coal industry indicators for data mining and machine learning model construction, and inputs the coal blending list, and Mining and constructing the index characteristics of a single coal in the coal blending list, combined with the support vector regression model, can not only organically combine with the coking mechanism of coal blending, improve the accuracy and precision of coke quality prediction, but also save high-quality The use of coal can improve the production quality of coke and reduce costs, and the required input data is easy to obtain for most coking enterprises, thus overcoming the difficulty of popularization

Images