Research On Multi-feature Grading Prediction Ofsintering Flame State

In the steel industry,To obtain high-quality sinter while reducing energy consumption and environmental pollution,it is essential to high-yield,stable and accurate end-point forecasts.The point includes three states: normal burning,overburning and underburning.The states are important process parameters closely related to the end point,and are the basis for the efficient operation of the machine.Using the image of the tail section to judge the end point is a relatively way to predict the sintering end point.The image of the tail section of the machine can most directly and effectively reflect the state of the end point.It is feasible and practical to make full use of the effective information contained in the flame section image to predict the state of the end point.The main work of this subject includes:1)Taking the image of the machine as the object,the series of cross-sectional images are segmented by the K-means clustering method,and the red fire area is extracted by color extraction to improve the target segmentation accuracy and geometric feature accuracy.2)Fuzzy clustering and improved random forest method based on the geometric characteristics of the flame predicts the state of the short-term end point.The results show that the prediction accuracy rate of fuzzy clustering can reach 98.81%,and the prediction accuracy rate of improved random forest can reach 97.78% and 96.67%.Both prediction methods can reduce the oscillation of prediction and improve the stability of prediction.3)To improve the prediction accuracy of the end point,the geometric characteristics of the image of the tail section of the sintering machine and the operating process parameters of the sintering machine are introduced,and the geometric characteristics of image,the information of the average speed of the machine trolley are granulated,the temperature data and the image are granulated.The parameter is the input parameter of the prediction model,and the least squares support vector machine is used to predict the end point in the medium and long term,and the accuracy rate reaches 96.67%,which avoids the interference of the change of external parameters on the prediction of the sintering end point,and improves the prediction of the sintering end point.accuracy and improve forecast stability.Figure 35;Table 13;Reference 57...