Design And Research Of FeO Content Prediction System Of Sinter Based On BP Neural Network

At present,the control of sintering production in the sintering production workshops of the metallurgical industry at home and abroad still relies on the accumulated sintering production experience.This also leads to the quality of sintering production,which directly depends on the experience of sintering experts.However,due to the differences in the production experience of different sintering experts,the results of the sintering production guidance will produce varying degrees of fluctuations.The detection of FeO content in sintered ore by chemical method is safer and more reliable,but its biggest shortcoming is aging lag.The FeO content detected in sintered ore is of little significance to the current sintering production guidance.This subject has carried out research on the tail section of the sintering machine through a large number of investigations on the actual sintering production,combined with the experience of on-site sintering experts.Developed a sinter FeO content prediction system based on the BP neural network’s characteristic sintering machine tail section temperature characteristic parameters.The system can realize more accurate FeO content prediction on the sinter ore at the characteristic sintering machine tail section on-line in real time and guide sintering production.In order to realize the above-mentioned sinter FeO content prediction system based on the BP neural network-based characteristic sintering machine tail section temperature characteristic parameters,the main work of this paper is as follows:(1)Use infrared thermal imaging camera to collect real-time temperature of the sintering machine tail,and use the partition trigger capture algorithm to capture the characteristic sintering machine tail section.(2)Perform feature extraction and analysis on the captured temperature information of the characteristic sintering machine tail section,and extract the temperature feature parameters such as the highest temperature of the red fire zone and the area ratio of each 100°C adjacent temperature section between 500°C and1000°C.Using the temperature information of the characteristic tail section,the algorithm realizes the analysis of the underburning,overburning and material uniformity of the characteristic tail section.(3)Collect and sort out the FeO content data of the sintered ore detected by the chemical method at the time corresponding to the section of the tail section of the characteristic sintering machine.(4)Correlate the temperature characteristic parameters of the characteristic sintering machine tail section with the FeO content of the sintered ore measured by the chemical method at the corresponding time,and use the data as the training set and test set of the BP neural network to continuously train and optimize the FeO prediction model.(5)Based on the C/S architecture and using the Qt visual graphical interface framework,the client program of the sintering intelligent analysis system is developed.