Research On Real-time Prediction Method Of Carbon Content At The End Of Converter Steelmaking Based On Flame Image Feature Extraction

In the converter steelmaking production process,the accurate prediction of the end point carbon temperature is a vital part of the steel industry,and the accurate prediction of the carbon content is of great significance to the improvement of the steel smelting process.This paper aims at the high similarity of the flame images of the furnace end of the converter,and the traditional feature method is difficult to extract the key features of the flame images with similar carbon content.The research is carried out from the extraction of the color and texture features of the flame image of the furnace,in order to improve the flame based on the furnace.The accuracy of carbon content prediction at the end of converter steelmaking based on image feature extraction lays the foundation.The main research contents of this paper are as follows:(1)A method for predicting the carbon content of the converter steelmaking end point based on the flame feature extraction of the convolutional neural network.By adjusting the parameters of the neural network VGG-16,such as adjusting the size of the convolution kernel and the size of the pooling,the zero padding method,the normalization and the parameters of the fully connected layer in the stacked convolutional layer,the amount of parameter calculations is reduced,and improved network training speed.The experimental results show that the VGG-16 network with adjusted parameters is used to extract the abstract features between different levels of the flame image,and to predict the end point carbon content of the 23 types of converter flame images,so that the error range of the end point carbon content of the converter is within 0.02% The prediction accuracy reached 68.2%.(2)An improved complete local binary pattern(ICLBP)color texture feature extraction method is proposed to extract more differentiated flame features at furnace mouth under different carbon contents and predict the endpoint carbon content.Firstly,local phase quantization(LPQ)is used to extract image frequency domain phase information under different color channels,and the fusion feature ICLBP?MP is combined with image spatial domain amplitude information extracted by CLBP to enhance the robustness of CLBP algorithm structure.Then,it is weighted by an improved color information weighting strategy to enhance the color contrast information of the flame image.Finally,the K nearest neighbor regression model is used to predict the carbon content.The experimental results show that the accuracy rate of carbon content prediction is 83.9% within the error range of 0.02%.(3)A prediction algorithm for the carbon content of the converter steelmaking end point based on the flame feature extraction of the completely local binary mode fusion of local and global spatial information is proposed.Firstly,the local segmentation blocks of the flame single-channel and cross-channel images are obtained using completely local binary values,and the color weighting and gradient weighting are performed respectively to enhance the local contrast of the symbol and amplitude blocks of the flame single-channel and cross-channel images;The sign and amplitude block of the local contrast are combined with the local segmentation block at the corresponding position of the flame image grayscale image respectively to fuse the local and global spatial information between the flame single-channel and crosschannel images.The experimental results show that the method in this chapter predicts the end-point carbon content of 23 types of converter flame images,so that the prediction accuracy of the end-point carbon content of the converter within the error range of 0.02% reaches 85.2%,which effectively improves the accuracy of the prediction of the end-point carbon content of the converter.Which solves the problem of low prediction accuracy of the converter end carbon content caused by the high flame image similarity.