A data and model-driven predictive diagnosis framework towards hot-rolled coil defect. (14th June 2023)

• Record Type: Journal Article
• Authors: Zhou, Shun; Xiang, Feng; Li, Hongjun; Zhang, Chi; Zhang, Xuerong
• Is Part Of: International journal of service and computing oriented manufacturing. Volume 4:Number 2(2023)
• Page Start: 156
• Page End: 165
• Publication Date: 2023-06-14
• Subjects: deep learning -- digital twin -- hot-rolled coil defect -- predictive diagnosis; Computer integrated manufacturing systems -- Periodicals; Manufacturing industries -- Information technology -- Periodicals; Manufacturing industries -- Computer networks -- Periodicals; Service-oriented architecture (Computer science) -- Periodicals; 670.285
• Languages: English
• ISSNs: 2045-175X

Abstract

The quality defect is one of the important indicators of hot-rolled coil quality. In order to realise real-time prediction of quality defect and timely control, a data and model-driven predictive diagnosis framework towards hot-rolled coil defect is proposed. Firstly, build a digital twin model from four aspects: geometry, physics, behaviour and rule. On this basis, combined with expert knowledge, deep learning and historical data, a predictive diagnostic model for hot-rolled coil defect was constructed. Then, the data-driven defect diagnosis method is used to realise the prediction of defects, and the model-driven result verification method is used to verify the prediction results. Finally, the accuracy of the result is verified by consistency judgement to improve the defect predictive diagnostic model, thereby improving the accuracy of prediction.