Plate shape recognition based on Gaussian function and particle swarm optimization for roller quenching process. (November 2022)

• Record Type: Journal Article
• Authors: Zhang, Wen; Wu, Min; Du, Sheng; Chen, Luefeng
• Is Part Of: Journal of process control. Volume 119(2022)
• Page Start: 115
• Page End: 127
• Publication Date: 2022-11
• Subjects: Roller quenching process -- Plate shape recognition -- Particle swarm optimization -- Information extraction; Process control -- Periodicals; Fabrication -- Contrôle -- Périodiques; Process control; Periodicals; Electronic journals; 660.281
• DOI: 10.1016/j.jprocont.2022.10.001
• Languages: English
• ISSNs: 0959-1524

Abstract

Abstract: Plate shape recognition in the roller quenching process provides effective feedback signals for plate shape control. Its accuracy directly affects the control effect. Most studies have obtained the high recognition accuracy of the plate shape defects by neural networks or parameter identification methods, but cannot provide a sufficiently detailed feedback signal with controllable accuracy for plate shape control, limiting the further improvement of the control effect. For these problems, a plate shape recognition method for roller quenching process is proposed. Firstly, the roller quenching process and the representation of the plate shape are introduced, and the structure of the plate shape recognition method is designed. Secondly, the Gaussian function is used to describe the plate shape defects, and the physical significance of each parameter in the Gaussian function is analyzed. Based on the Gaussian function, a plate shape fitting method with iterative structure and controllable fitting accuracy is designed to extract the detailed information about each defect. Thirdly, according to the actual production experience, the expert rules are designed for plate shape classification and to evaluate the quality. Finally, the method is verified by industrial production data. The results verify the correctness and effectiveness of our method, indicating that the method can provide important guidance for the improvement of plate shape quality and lays an important foundation for plate shape control. Highlights: An plate shape fitting framework is proposed with controllable fitting accuracy. Higher classification accuracy is obtained by integrating the expert rules. The algorithm process is simple and easy to implement in the industrial field.