The Study Of Steelmaking Endpoint Forecast Based On Multi-scale And Multi-feature Flame Method

China is knowned as the world’s largest steel-producing and steel consuming country. Not only the steel industry supports the development of the national economy, but also provide raw materials for many fields. Whether the control of the chemical composition of steel best directly influence the quality of steel. the carbon content of steel will determine not only steel’s type, but also affect the strength of steel, hardness, and weldability, So how much of the carbon content of the furnace steel-making process is the key to grasp.According to human experience, there is a corresponding change in the relationship between law and morphological changes in the furnace mouth flame carbon content. So in order to accurately predict the end of steelmaking, making high quality and reasonable carbon content steel, it seems most important to research the carbon content reflected by flame characters.This article used the computer vision to simulate human vision, using analysis algorithm to analysis the flame characteristics without human-beings, and achieving computer to see the fire prediction steelmaking end. This paper analyzes the various ways of other former ob the field of flame feature extraction, summed up the advantages and disadvantages of each method, combined with the difficulty of feature extraction of steelmaking fire.Proposes to determine the steelmaking stage firstly, then detailed analysis the relationship of another characters of later in flame and the carbon content. Which to get the number of fire harris corner to judge the stage of steelmaking. And based YIQ color space to extract contour of flame, get the curvature of edge to fit the carbon content to achieve the end forecast. The main work of this article is as follows:1. Judging steelmaking stage based on Harris corner number. Firstly build a difference of Gaussian space for the fire image and then create an image pyramid. Then, get stable harris corner number based on image pyramid, finally, divide the stage of steelmaking combined the worker’s experience of three stages with corners threshold.2. Extracting contour based on YIQ color space and then get the edge curvature. Firstly after entering the steelmaking late, extracting contour based fire YIQ color space, and then get the curvature based on mathematical sine after get the fire contour,and fit the carbon content of human experience and build data relationship model, to achieve the forecast of end of the steelmaking.In order to verify the feasibility, robustness and efficiency of forecasting steelmaking endpoint based on multi-scale multi-feature flame method proposed in this paper of the study, and experimental design and programming will be designed for each step by the MATLAB, and verification and fitting artificial empirical to establish d data relationship model.and compared this page’s contour extraction algorithms with others In this paper. The final results of this algorithm is well proven efficient, viable and good robustness...