Quality Control system for a hot-rolled metal surface

Mazur, Igor P.; Koinov, Toncho

doi:10.3221/igf-esis.37.38

Cited by 12 publications

(7 citation statements)

References 1 publication

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Cacat permukaan pada lembaran baja menyebabkan penurunan kualitasnya. Dengan pengklasifikasian jenis cacat tersebut memungkinkan untuk dengan cepat mengidentifikasi dan menghilangkan penyebab terjadinya cacat (Mazur & Koinov, 2016). Oleh karena itu, efisiensi dan akurasi klasifikasi cacat merupakan hal yang penting dalam pengendalian kualitas produk logam.…”

Section: Pendahuluanunclassified

Pengendalian Kualitas Produksi Lembaran Baja Melalui Klasifikasi Jenis Cacat Permukaan Menggunakan CNN

Indarti

2023

JRSI

View full text Add to dashboard Cite

Pengendalian kualitas produksi lembaran baja dengan persepsi visual manusia sering kali terjadi kesalahan dan membutuhkan waktu yang lebih lama. Implementasi deep learning dalam pengendalian kualitas produksi lembaran baja dapat memiliki akurasi yang baik dan dilakukan secara real-time. Klasifikasi jenis cacat permukaan pada lembaran baja merupakan hal yang penting dalam pengendalian kualitas produksi lembaran baja secara otomatis. Dengan mengklasifikasikan jenis cacat menggunakan deep learning dapat mengidentifikasi dan menghilangkan penyebab terjadinya cacat saat produksi lembaran baja dengan cepat. Penelitian ini bertujuan untuk mengendalikan kualitas produksi lembaran baja secara otomatis melalui klasifikasi jenis cacat permukaan menggunakan Convolutional Neural Network (CNN). Model CNN yang digunakan pada penelitian ini yaitu CNN dengan transfer learning dari 5 pre-trained model Resnet50, VGG-16, VGG-19, Inception V3, dan Xception. Terdapat 6 jenis cacat permukaan yang diklasifikasikan yaitu crazing, inclusion, pitted, patches, rolled, dan scratch. Penelitian dimulai dengan pengambilan data citra lembaran baja yang memiliki cacat permukaan. Jumlah citra yang digunakan dalam penelitian ini yaitu 1.800 citra terdiri dari 1.152 data pelatihan, 288 data validasi, dan 360 data pengujian. Selanjutnya dilakukan preprocessing yaitu normalisasi, augmentasi, dan one-hot encoding. Setelah preprocessing dilakukan pelatihan dan validasi menggunakan transfer learning dari 5 pre-trained model. Model hasil pelatihan digunakan pada tahap pengujian. Hasil pelatihan dan validasi menunjukkan bahwa Xception memiliki kinerja terbaik karena nilai akurasi pelatihan dan validasi tertinggi, nilai loss pelatihan dan validasi terendah, serta GAP validasi dan loss terendah. Hasil pengujian menunjukkan bahwa transfer learning dari pre-trained model Xception memiliki kinerja terbaik dengan akurasi sebesar 98%.

show abstract

Section: Pendahuluanunclassified

Pengendalian Kualitas Produksi Lembaran Baja Melalui Klasifikasi Jenis Cacat Permukaan Menggunakan CNN

Indarti

2023

JRSI

View full text Add to dashboard Cite

show abstract

“…It follows 18,19,20 from papers that each production builds its quality control system, which takes into account the specifics of its structure and management methods. At the same time, proactive quality management systems 21,22 are widely used, including methods of controlling technological parameters 23,24 , testing quality indicators 25,26 , and forecasting based on mathematical models (e.g., statistical 27,28 or neural network 29,30 ). Thus, in order to control the production process and ensure the predicted quality of thermo-mechanically processed rolled products, it is necessary to develop a proactive quality management system to establish and control the relevant technological factors, which is the objective of this study.…”

Section: Literature Reviewmentioning

confidence: 99%

Implementation of Quality Management System for Production of TMCP Treated 10mn2vnbal Steel Heavy Plates

Kukhar,

Kurpe,

Malii

2023

Miningmetaltech 2023 – The Mining and Metals Sector: Integration of Business, Technology and Education

View full text Add to dashboard Cite

principles using a number of selected quality initiatives and practices 4,5 . The involvement of a quality system in steel-making infrastructures is formal, but not everywhere. The paper 6 shows that ISO 9001 is the world's most popular QMS, certifying over 1 million different companies; moreover, Europe and the Far East make up nearly 90% of all ISO 9001 certificate amounts. These are mainly infrastructure and manufacturing companies whose adherence to QMS principles contributes to global competition. There are positive experiences of TQM implementation in the industrial sectors of developed 7 and developing countries 8 . Special expert evaluations and questionnaires were developed to identify components (factors) affecting the efficiency and sustainability of enterprises. The study 9 proposes a model confirming that the TQM practices have a positive relation with sustainability and organisational performance and showed that leadership, continuous improvement, customer satisfaction, training and education, and customer relations have the greatest influence on organisational effectiveness. To develop the TQM strategy, the authors 10 suggest using a qualitative research technique to collect secondary data from empirical studies and literature reviews. This work 11 is presented as the implementation guide for QMS and follows the four pillars of ISO 9001: quality management principles, process approach, risk-based thinking, and Plan-Do-Check-Act (PDCA, or Deming) cycle.

show abstract

“…The hot rolling process carried out at a high temperature (austenitic region) makes it possible to reduce the flow stress and the strain hardening, resulting in a decrease of the rolling load and stresses. The knowledge of the slab temperature during the rolling process and the rollers heat transfer of the rolling mill can provide information of the microstructure developed in the sheet [6]; the mechanical properties can even be affected due to the temperature changes in the rolling process [7,8].…”

Section: Introductionmentioning

confidence: 99%

3D-FEM Simulation of Hot Rolling Process and Characterization of the Resultant Microstructure of a Light-Weight Mn Steel

et al. 2021

View full text Add to dashboard Cite

Computational simulation has become more important in the design of thermomechanical processing since it allows the optimization of associated parameters such as temperature, stresses, strains and phase transformations. This work presents the results of the three-dimensional Finite Element Method (FEM) simulation of the hot rolling process of a medium Mn steel using DEFORM-3D software. Temperature and effective strain distribution in the surface and center of the sheet were analyzed for different rolling passes; also the change in damage factor was evaluated. According to the hot rolling simulation results, experimental hot rolling parameters were established in order to obtain the desired microstructure avoiding the presence of ferrite precipitation during the process. The microstructural characterization of the hot rolled steel was carried out using optical microscopy (OM), scanning electron microscopy (SEM) and X-ray diffraction (XRD). It was found that the phases present in the steel after hot rolling are austenite and α′-martensite. Additionally, to understand the mechanical behavior, tensile tests were performed and concluded that this new steel can be catalogued in the third automotive generation.

show abstract

Quality Control system for a hot-rolled metal surface

Cited by 12 publications

References 1 publication

Pengendalian Kualitas Produksi Lembaran Baja Melalui Klasifikasi Jenis Cacat Permukaan Menggunakan CNN

Pengendalian Kualitas Produksi Lembaran Baja Melalui Klasifikasi Jenis Cacat Permukaan Menggunakan CNN

Implementation of Quality Management System for Production of TMCP Treated 10mn2vnbal Steel Heavy Plates

3D-FEM Simulation of Hot Rolling Process and Characterization of the Resultant Microstructure of a Light-Weight Mn Steel

Contact Info

Product

Resources

About