|
An Active Noise Control Algorithm Using Variable Step Size Adaptive Filter with Constraint in Sparse System |
|
Chan Park, Jun Hui Lee, Seongrok Moon, PooGyeon Park(POSTECH, Korea) |
| This paper proposes an active noise control (ANC) algorithm that enable to use in a sparse system environment. The filtered-x algorithm was employed through the adaptive filter to address the issue of secondary path distortion in active noise control. For estimating the inverse channels in the sparse system, the l_0-norm was employed, and a newly analyzed mean square deviation (MSD) was applied to calculate the optimal step size that enhance convergence performance. The modified reset algorithm was also applied to ensure that the adaptive filter can track the inverse channel properly even in situations where the system suddenly changes. |
|
|
Real-Time Steel Surface Defect Detection and Classification with Inference Acceleration |
|
Seongrok Moon, Jun Hui Lee, Kyung Soo Kim, Chan Park, PooGyeon Park(POSTECH, Korea) |
| Detecting defects in the steel industry is paramount, where models like YOLOv7 and Faster R-CNN are taking on the challenge. Achieving the right equilibrium between speed and accuracy is pivotal, yet existing models grapple with real-time defect detection and classification. Our mission was to forge a model with a high-detection rate and real-time capabilities. We compared YOLOv7 (one-stage) and Faster R-CNN (two-stage), fine-tuning through TensorRT. The upshot, Our model accomplished a 98.8% detection accuracy all while maintaining a rapid 46 FPS pace using real production data. This distinctly highlights the prowess of our approach in harmonizing detection precision and detection speed. |
|
|
Design of Supervisory Model Predictive Control for HVAC Systems and Two Zones with Consideration of Energy Efficiency and Thermal Comfort |
|
Panithan Srisurapanon, David Banjerdpongchai(Chulalongkorn University, Thailand) |
| This paper presents the design of supervisory model predictive control for HVAC systems with two zones. The supervisory control is designed to find the optimal reference temperature of each zone. Then, the model predictive control is designed to track the optimal reference signals. We compare the results between the decentralized control and the centralized control. The Pareto optimal curve shows that the performance of the centralized control is better than that of the decentralized control. Therefore, the centralized control is more appropriate for two zones because the improved tracking performance and the lower total operating cost. |
|
|
MARG Attitude estimation using global measurement update |
|
Young Soo Suh, Thanh Tuan Pham(University of Ulsan, Korea) |
| The multiplicative quaternion error model is a widely adopted technique for addressing attitude estimation problems. However, one limitation of this approach is the significant time required to mitigate the influence of substantial initial estimation errors. To address this concern, this paper presents a novel measurement update algorithm designed to expedite the removal of large initial estimation errors. The proposed algorithm formulates the measurement update as a linear optimization problem. Through simulations, the results demonstrate the robustness of the proposed algorithm in effectively handling significant initial estimation errors. |
|