|
Localization of slab identification numbers using deep learning |
|
Sang Jun Lee, Jaepil Ban, Hyeyeon Choi, Sang Woo Kim(POSTECH, Korea) |
| In the steel industries, recognizing product information is an important task for the management of the manufacturing processes. Before recognizing an identification number, localization is conducted to obtain a satisfactory performance. The objective of the paper is to develop a localization algorithm for slab identification numbers in actual steel factory scenes. A deep convolutional neural network was used to improve the performance and to overcome limitations of conventional rule-based algorithms. The proposed algorithm showed a reliable performance in the experiment. |
|
|
Systems Engineering Approach to Designing Smart Condition Monitoring Systems for Smart Manufacturing |
|
Kee-Young Shin(POSCO Technical Research LABs, Korea) |
| Condition monitoring systems conduct important roles in manufacturing process lines by monitoring condition status of machines to prevent accident faults of them. Those systems need to be improved in their functionalities by adapting more intellectual, self-reconfigurable, and self-decisional mechanisms for supporting smart manufacturing operations. In this paper, we propose a systems engineering approach to designing smart condition monitoring systems. |
|
|
Vibration control of a strip in a Continuous Galvanizing Line using Self-Tuning Neuro-PID controller |
|
Junmin Park, Hyungwoong Lee, Poogyeon Park(POSTECH, Korea) |
| This paper proposes a vibration control scheme of a strip in a continuous galvanizing line using neuro-PID
controller which consists of two neural networks and conventional PID controller. One NNs estimates the strip model
using input-output data of the strip and other NNs updates the gains of conventional PID controller using the estimated
strip model and errors between the desired and actual outputs. The proposed scheme gives the adaptation ability for
different strips and operating conditions. |
|
|
Deep learning based modeling for the lateral movement of a strip in hot finishing mill |
|
Wookyong Kwon, Jaemin Baek, Soohee Han, Sangchul Won(POSTECH, Korea) |
| In this paper, the problem of system identification for the lateral motion of a strip in hot finishing mill is investigated. The movement is affected by various asymmetric factors with respect to rolling force. Not only that, the tension between rolling mills determines the direction of strip's moving. Consequently, the movement of a strip is complex dynamics with rolling condition, tension and other phenomena. To identify a neural network type system model in the existence of bouth uncertain parameters and nonlinear signals, deep learning based modelling is employed. |
|
|
Control Methodology for Stabilized Strip Tracking in Tandem Cold Strip Mill System |
|
Jae-Min Baek, Wookyong Kwon(POSTECH, Korea), Jung-Hun Park(POSCO, Korea), Soohee Han(POSTECH, Korea) |
| This paper presents modified sliding-mode control for the tandem cold strip mill during Flying Gauge Change
(FGC) period that is an important component for achieving fast stabilization of strip tracking when FGC period is closed.
It means that the proposed algorithm can attenuate strip oscillations generated by inexactitude of setup values of FGC.
As such, in real system, we expect to reach a high level of merit in cost aspect. The proposed algorithm is compared to
method without proposed algorithm, and the effectiveness is verified through the results of simulation. |
| |