In view of the nonlinear, random and uncertain characteristics of FeO content in sintered ore, a FeO content measurement model based on RBF neural network is proposed. The image features of sintering tail and main operating process parameters are integrated to predict the FeO content online, and an \"online intelligent detection system for FeO content in sintered ore\" is developed. The application shows that the model has small calculation amount, high precision, practical and simple algorithm, and achieves ideal results. The FeO content of sintered ore is a comprehensive indicator for evaluating sintering production. Reasonable control of FeO content is conducive to reducing sintering energy consumption and improving the reducibility of sintered ore. Therefore, the detection and control of FeO content in sintered ore is of great significance to sintering production. At present, the main methods for detecting FeO content in sintered ore are: chemical analysis method, waste gas analysis method, sintering machine tail observation method and sintering parameter numerical analysis method. The chemical analysis method requires manual extraction of sintered ore samples for chemical analysis. The results are accurate but cannot be detected in real time. The waste gas analysis method estimates the FeO content by analyzing the content of the components of the sintering tail gas. This method also requires manual sampling and is easily interfered by external factors, resulting in large errors in the results. The observation method at the end of the sintering machine can be divided into the observation method based on pyrotechnics and the method based on machine vision [1]. Both methods estimate the FeO content by analyzing the cross-sectional image of the end of the sintering machine. The numerical analysis method of sintering parameters analyzes the various parameters that affect the sintering process, establishes a mathematical model, and predicts the FeO content. The complex nature of the sintering process and the gray nature of the system determine that it is difficult to describe the FeO content with an accurate mathematical model. The sintering theory and optical CCD imaging theory show that the FeO content of the sintered ore is not only related to the characteristics of the cross-sectional image of the end of the sintering machine, but also has a complex nonlinear mapping relationship with the sintering operation process parameters. In-depth research has been conducted at home and abroad on the prediction of FeO in sintered ore. Zhang Shi et al. [2] used the image feature value of the cross section of the sintering machine tail for FeO regression analysis, but the error was large; Jiang Dajun et al. [3] established a BP network model of FeO content based on the sintering operation process parameters. The model has a high hit rate, but the convergence speed is slow and it is difficult to run online. The radial basis function neural network is a feedforward artificial neural network with good performance. It has a high computing speed and can globally approximate a nonlinear function with arbitrary precision, and it is the global optimal. To this end, this paper uses the RBF neural network to fuse the main process parameters and image features to predict the FeO content online. The \"Online Intelligent Detection System for FeO Content in Sintered Ore\" was developed for the No. 6 sintering machine of the ironmaking plant of a steel group company. The application shows that the model has a fast running speed, a high hit rate, and achieves ideal results.
You Might Like
Recommended ContentMore
Open source project More
Popular Components
Searched by Users
Just Take a LookMore
Trending Downloads
Trending ArticlesMore