Metallurgical product quality management and control technology and system based on industrial big data

He Fei

Collaborative Innovation Center of Steel Technology, University of Science and Technology Beijing, Beijing, 100083, China

EXTENDED ABSTRACT: The metallurgical production process is a typical continuous-discrete, multi-process and complex process. At present, the overall equipment level and automation control level of the manufacturing process in large metallurgical enterprises are at the international advanced level. At the same time, a large amount of industrial data of the whole manufacturing process has been accumulated based on a relatively complete information and automation system. How to make full use of the process and quality data from the all levels information system and the whole process to improve the uniformity and stability of product quality is an important scientific research topic. This research focus on the quality control technology of metallurgical products, which includes quality process design, production process monitoring, product quality prediction-rating, quality diagnosis and optimization via quality control closed-loop perspectives: 1) Production process monitoring, study multivariable coupling monitoring under the characteristics of multivariable and non-Gaussian data, and find process abnormalities or trends in time; 2) Product quality prediction, fully combine the mechanism and data driven methods to study the mechanical properties and metallographic prediction methods of high-carbon steel; 3) Product quality rating, aiming at the difficulty of mechanical performance online detection, use the comprehensive uniformity of a large number of process parameters to achieve uniformity evaluation of mechanical properties, and predict the product quality grade, and provide users with the most suitable products; 4) Quality diagnosis, in view of the difficulty in giving the reasons for the quality defects because of the complex nonlinear and strong coupling characteristics between data, analyze the influence of data, study the sensitivity and importance of variables, and then locate the weak stages; 5) Quality optimization and process design, make full use of a large amount of actual production data, and seek more reasonable process control range. Finally, a whole-process quality management and control system is built to form an integrated storage of the whole-process data, and then effectively integrate the process and quality data, which is applied to steel, non-ferrous and other enterprises. The system includes process monitoring, quality prediction and rating, process optimization and so on, which enhance the stability of product quality and the proportion of high-quality products.