|
Temporal Change Detection of Battery Anode Slurry Flow Data Using CNN Classifier and Investigation of Significance through Grad-CAM Analysis |
|
Hyejung Oh, Junseop Shin, Hyunjoon Jung, Jaewook Nam, Jong Min Lee(Seoul National University, Korea) |
| LIBs are widely used as a significant energy source. Detecting temporal property changes in electrode production slurry is vital due to their direct impact on electrode performance. This study proposes an algorithm using our own data for binary classification of pipe flow signals, distinguishing day 1 and day 5 signals. Scaling and short-time Fourier transform are applied to data segments, followed by training an autoencoder and classifier with convolutional layers. This technique achieves 69% accuracy and provides a confusion matrix. Gradient class activation maps (Grad-CAM) visualize classifier focus during data classification. |
|
|
Design Condition for Achieving Zero Steady-state Tracking Error in Offset-free MPC with Moving Horizon Estimator |
|
Wonhyeok Choi(Seoul National University, Korea), Sang Hwan Son(PNU, Korea), Jong Min Lee(Seoul National University, Korea) |
| When applying Model Predictive Control (MPC), model-plant mismatch frequently leads to unfavorable offset in controlled variables. This research introduces a design condition to guarantee zero steady-state offset in the presence of model-plant mismatch while incorporating the Moving Horizon Estimator (MHE) within offset-free MPC. Subsequently, MHE-driven offset-free MPC was employed in a numerical illustration involving a CSTR system. It is shown that the suggested methodology, along with the suggested design condition, not only achieves offset-free tracking but also enhances the transient behavior of the process compared to conventional offset-free MPC that employs the Luenberger observer. |
|
|
Enhancing Electric Vehicle Air Conditioning Systems with Model Predictive Control for Optimal Refrigerant Cycle Performance |
|
Jisung Byun, Hyein Jung(Seoul National University, Korea), Jaewoong Kim, Se-Kyu Oh(Hyundai Motor Company, Korea), Jong Min Lee(Seoul National University, Korea) |
| Due to the battery capacity limit, management of electricity consumption of each component in EV is important. A refrigerant cycle of A/C system is affected by physical constraints and multiple variables. Therefore, model predictive control (MPC) can be used for the effective operation and control. However, heat exchanger models in PDE guarantees high accuracy, but its complexity imposes difficulty in real-time control. Thus, Moving boundary method (MBM) is employed to construct a precise yet simplified model. Subsequently, the simplfied model is utilized in MPC effectively regulate the air temperature discharged from the evaporator. |
|
|
Adsorbent Selection for Biogas Upgrading PSA process: A Novel Isotherm Fitting Algorithm for Efficient Connection of Molecular and Process Simulations |
|
Nahyeon An(Institute of Industrial Technology, Korea), Junghwan Kim(Yonsei University, Korea), Seongbin Ga(University of Ulsan, Korea) |
| This paper presents a novel method for the selection of metal-organic frameworks (MOFs) for biogas upgrading within pressure swing adsorption (PSA) processes. This methodology is facilitated by a novel isotherm fitting algorithm designed for molecular and process simulations. MOFs, distinguished by their extraordinary porosity and customizable chemical functionality, are gaining escalating interest. However, finding the most advantageous MOFs from over 10,000 identified MOFs for biogas upgrading media represents a substantial challenge. Conventional simulation methods, notwithstanding their potential, are often hindered by the significant computational burden. To address this problem, we int |
|