|
A generalized multiple-integral inequality based on free matrices: Application to stability analysis of time-varying delay systems |
|
Jun Hui Lee, Chan Park, Seongrok Moon, PooGyeon Park(POSTECH, Korea) |
| This paper proposed a novel generalized multiple-integral inequality based on free matrices (GMIIFM) in a quadratic form of an augmented vector stacked with a state and its derivative, with generalization realized in terms of the two perspectives of bounding and integration orders. The usage of cross information on the structured free matrices guaranteed tight upper bounds, and the combination of the two perspectives can yield various integral inequalities. Moreover, it was observed that the higher the orders of the generalized multiple-integral inequalities and Lyapunov–Krasovskii functions, the less conservative the stability criterion obtained. |
|
|
A standalone energy management system of hybrid battery/supercapacitor storage system using model predictive control: A dissemination version |
|
Ngoc-Duc Nguyen, Young Il Lee(SeoulTech, Korea) |
| - A standalone energy management strategy is developed to reduce the fluctuation of battery power and handle the peak power by supercapacitor (SC) under the EV motor load.
- The hybrid energy storage system composed of a Li-ion battery, DC-DC converter and Li-ion SC has been developed for the range-extended EV.
- The presented approach can operate standalone without communicating with EV’s motor drive system while managing the power flow between the battery and supercapacitor (SC) to reduce the peak-power stress on the battery and to maintain the output voltage inside the pre-defined voltage range of the motor drive. |
|
|
Distributed Graph Matching and Structure of Symmetric Graphs |
|
Hyo-Sung Ahn(GIST, Korea) |
| This presentation is based on our previous publications [1-2]. In this presentation, we introduce distributed computation scheme for graph matching, structure and unfriendliness of symmetric graph, and unfriendliness and uncontrollability in consensus dynamics. We provide distributed optimization for one-to-one vertex correspondences between two undirected and connected graphs under a multi-agent setup. It will be shown that given two isomorphic and asymmetric graphs, there is a unique permutation matrix that maps the vertices in one graph to the vertices in the other.
1. Quoc Van Tran*, Zhiyong Sun, Brian D.O. Anderson, Hyo-Sung Ahn, “Distributed optimization for graph matching,” IEEE Tran |
|
|
Maximum Norm Minimization: A Single-Policy Multi-Objective Reinforcement Learning to Expansion of the Pareto Front |
|
SeonJae Lee, Myoung Hoon Lee, Jun Moon(Hanyang University, Korea) |
| This paper introduces Maximum Norm Minimization (MNM), a type of single-policy Multi-Objective Reinforcement Learning (MORL) approach, designed for addressing multi-objective problems and obtaining Pareto optimal points that form the Pareto front. The effectiveness of MNM in multi-objective problems is demonstrated through experimental results obtained from five complex robotic environments. Moreover, a comparison is conducted between MNM and other advanced MORL techniques to emphasize the superior performance of MNM in terms of the summation of volumes and the mean squared distance of Pareto optimal points on the Pareto front. |
|
|
Tube-based Control Barrier Function for Unknown Input Delay |
|
Ying Shuai Quan, Jing Sung Kim(Hanyang University, Korea), Seung-Hi Lee(Hongik University, Korea), Chung Choo Chung(Hanyang University, Korea) |
| This paper proposes a Control Barrier Function (CBF)-based controller design to achieve safety for systems subjecting to unknown input delay and additive disturbance. Integral quadratic constraints characterizing the input-output behavior of the dynamics caused by the unknown input delay are used to generate a bound of the error between the nominal and the true uncertain system. The bound is incorporated into a tube-based CBF formulation to ensure robust safety. The proposed method guarantees that the constraints are input affine, so the safe controller can be implemented by solving a quadratic programming problem in real-time. A simple example demonstrates the effectiveness of the method. |
|
|
Adaptive Learning Gain Based Control for Nonlinear Systems with External Disturbances: Application to PMSM : A dissemination version |
|
Sesun You, Wonhee KIM(Chung-Ang University, Korea) |
| In this study, we propose an adaptive learning gain (ALG)-based nonlinear control for a strict-feedback nonlinear system with an external disturbance to achieve the desired control performance without any information of the disturbance and gain tuning. An ALG-based nonlinear controller using a backstepping procedure is designed to track the desired trajectory. Furthermore, the time required for tuning reduced because the ALG is automatically and simultaneously updated itself to obtain satisfactory control performance. The expected effect of the proposed method is that an advanced nonlinear controller using the backstepping could be easily used in the industrial field without gain tuning. |
|