• Proceedings of the Korean Society of Computer Information Conference
• /
• 2024.01a
• /
• pp.31-34
• /
• 2024

태양광 에너지는 탄소 중립 이행을 위한 주요 방안으로 많은 주목을 받고 있다. 태양광 발전량은 여러 환경적 요인에 따라 크게 달라질 수 있으므로, 정확한 발전량 예측은 전력 네트워크의 안정성과 효율적인 에너지 관리에 근본적으로 중요하다. 대표적인 인공지능 기술인 신경망(Neural Network)은 불안정한 환경 변수와 복잡한 상호작용을 효과적으로 학습할 수 있어 태양광 발전량 예측에서 우수한 성능을 도출하였다. 하지만, 신경망은 모델의 구조나 초매개변수(Hyperparameter)를 최적화하는 것은 복잡하고 시간이 많이 드는 작업이므로, 에너지 분야에서 실제 산업 적용에 한계가 존재한다. 본 논문은 2단계 신경망 최적화를 통한 태양광 발전량 예측 기법을 제안한다. 먼저, 태양광 발전량 데이터 셋을 훈련 집합과 평가 집합으로 분할한다. 훈련 집합에서, 각기 다른 은닉층의 개수로 구성된 여러 신경망 모델을 구성하고, 모델별로 Optuna를 적용하여 최적의 초매개변숫값을 선정한다. 다음으로, 은닉층별 최적화된 신경망 모델을 이용해 훈련과 평가 집합에서는 각각 5겹 교차검증을 적용한 발전량 추정값과 예측값을 출력한다. 마지막으로, 스태킹 앙상블 방식을 채택해 기본 초매개변숫값으로 설정해도 우수한 성능을 도출하는 랜덤 포레스트를 이용하여 추정값을 학습하고, 평가 집합의 예측값을 입력으로 받아 최종 태양광 발전량을 예측한다. 인천 지역으로 실험한 결과, 제안한 방식은 모델링이 간편할 뿐만 아니라 여러 신경망 모델보다 우수한 예측 성능을 도출하였으며, 이를 바탕으로 국내 에너지 산업에 이바지할 수 있을 것으로 기대한다.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.12 no.6
• /
• pp.2711-2733
• /
• 2018

Minimizing power consumption in bandwidth limited optical traffic grooming networks is presented as a two-objective optimization problem. Since the main objective is to route a connection, the network throughput is maximized first, and then the minimum power consumption solution is found for this maximized throughput. Both transparent IP over WDM (Tp-IPoWDM) and translucent IP over WDM (Tl-IPoWDM) network may be applied to examine such bi-objective algorithms. Simulations show that the bi-objective algorithms are more energy-efficient than the single objective algorithms where only the throughput is optimized. For a Tp-IPoWDM network, both link based ILP (LB-ILP) and path based ILP (PB-ILP) methods are formulated and solved. Simulation results show that PB-ILP can save more power than LB-ILP because PB-ILP has more path selections when lightpath lengths are limited. For a Tl-IPoWDM network, only PB-ILP is formulated and we show that the Tl-IPoWDM network consumes less energy than the Tp-IPoWDM network, especially under a sparse network topology. For both kinds of networks, it is shown that network energy efficiency can be improved by over-provisioning wavelengths, which gives the network more path choices.

• Journal of KIISE
• /
• v.44 no.7
• /
• pp.719-725
• /
• 2017

To reduce management costs and improve performance, the European Telecommunication Standards Institute (ETSI) introduced the concept of network function virtualization (NFV), which can implement network functions (NFs) on cloud/datacenters. Within the NFV architecture, NFs can share physical resources by hosting NFs on physical nodes (commodity servers). For network service providers who support NFV architectures, an efficient resource allocation method finds utility in being able to reduce operating expenses (OPEX) and capital expenses (CAPEX). Thus, in this paper, we analyzed the network service chain embedding problem via an optimization formulation and found a close-optimal solution based on the Markov approximation framework. Our simulation results show that our approach could increases on average CPU utilization by up to 73% and link utilization up to 53%.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.11 no.10
• /
• pp.4865-4886
• /
• 2017

The user experience of practical indoor power line communication (PLC) applications is greatly affected by the system quality-of-service (QoS) criteria. With a general broadcast-and-multi-access (BMA) relay scheme, in this work we investigate the joint source and relay power optimization of the amplify-and-forward (AF) relay systems used under indoor broad-band PLC environments. To achieve both time diversity and spatial diversity from the relay-involved PLC channel, which is time-varying in nature, the source node has been configured to transmit an identical message twice in the first and second signalling phase, respectively. The QoS constrained power allocation problem is not convex, which makes the global optimal solution is computationally intractable. To solve this problem, an alternating optimization (AO) method has been adopted and decomposes this problem into three convex/quasi-convex sub-problems. Simulation results show the fast convergence and short delay of the proposed algorithm under realistic relay-involved PLC channels. Compared with the two-hop and broadcast-and-forward (BF) relay systems, the proposed general relay system meets the same QoS requirement with less network power assumption.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.5 no.11
• /
• pp.1891-1908
• /
• 2011

Cyber-physical systems (CPSs) are an emerging discipline that involves engineered computing and communicating systems interfacing the physical world. The widespread applications of CPSs still face enormous challenges because of the lack of theoretical foundations. In this technical survey, we review state-of-the-art design techniques from various angles. The aim of this work is to provide a better understanding of this emerging multidisciplinary methodology. The features of CPSs are described, and the research progress is analyzed using the following aspects: energy management, network security, data transmission and management, model-based design, control technique, and system resource allocation. We focus on CPS resource optimization, and propose a system performance optimization model with resource constraints. In addition, some classic applications (e.g., integrating intelligent road with unmanned vehicle) are provided to show that the prospects of CPSs are promising. Furthermore, research challenges and suggestions for future work are outlined in brief.

• Journal of the Korean Institute of Illuminating and Electrical Installation Engineers
• /
• v.25 no.7
• /
• pp.15-25
• /
• 2011

This proposes an online loss minimization algorithm for series PI(SPI) based interior permanent magnet synchronous motor(IPMSM) drive to yield high efficiency and high dynamic performance over wide speed range. The loss minimization algorithm is developed based on the motor model. In order to minimize the controllable electrical losses of the motor and thereby maximize the operating efficiency, the d-axis armature current is controlled optimally according to the operating speed and load conditions. For vector control purpose, a SPI is used as a speed controller which enables the utilization of the reluctance torque to achieve high dynamic performance as well as to operate the motor over a wide speed range. Also, this paper proposes current control of model reference adaptive fuzzy controller(MFC), and estimation of speed using artificial neural network(ANN) controller. The proposed efficiency optimization control, SPI, MFC, ANN in this paper is applied to IPMSM drive system, the validity of this paper is proved by analyzing response characteristics in variety operating conditions.

• Proceedings of the KIEE Conference
• /
• 2005.07d
• /
• pp.2888-2890
• /
• 2005

In this paper, we introduce a new structure of fuzzy-neural networks Fuzzy Set-based Polynomial Neural Networks (FSPNN). The two underlying design mechanisms of such networks involve genetic optimization and information granulation. The resulting constructs are Fuzzy Polynomial Neural Networks (FPNN) with fuzzy set-based polynomial neurons (FSPNs) regarded as their generic processing elements. First, we introduce a comprehensive design methodology (viz. a genetic optimization using Genetic Algorithms) to determine the optimal structure of the FSPNNs. This methodology hinges on the extended Group Method of Data Handling (GMDH) and fuzzy set-based rules. It concerns FSPNN-related parameters such as the number of input variables, the order of the polynomial, the number of membership functions, and a collection of a specific subset of input variables realized through the mechanism of genetic optimization. Second, the fuzzy rules used in the networks exploit the notion of information granules defined over systems variables and formed through the process of information granulation. This granulation is realized with the aid of the hard C- Means clustering (HCM). The performance of the network is quantified through experimentation in which we use a number of modeling benchmarks already experimented with in the realm of fuzzy or neurofuzzy modeling.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.11 no.12
• /
• pp.5712-5728
• /
• 2017

Recently, there are many studies, that considering green wireless cellular networks, have taken the energy consumption of the base station (BS) into consideration. In this work, we first introduce an energy consumption model of multi-mode sharing BS powered by multiple energy sources including renewable energy, local storage and power grid. Then communication load requests of the BS are transformed to energy demand queues, and battery energy level and worst-case delay constraints are considered into the virtual queue to ensure the network QoS when our objective is to minimize the long term electricity cost of BSs. Lyapunov optimization method is applied to work out the optimization objective without knowing the future information of the communication load, real-time electricity market price and renewable energy availability. Finally, linear programming is used, and the corresponding energy efficient scheduling policy is obtained. The performance analysis of our proposed online algorithm based on real-world traces demonstrates that it can greatly reduce one day's electricity cost of individual BS.

• The KSFM Journal of Fluid Machinery
• /
• v.13 no.6
• /
• pp.42-50
• /
• 2010

The present analysis deals with a numerical procedure for optimizing the shape of elliptical dimples in a cooling channel. The three-dimensional Reynolds-averaged Navier-Stokes (RANS) analysis is employed in conjunction with the SST model for predictions of the turbulent flow and the heat transfer. Three non-dimensional geometric design variables, such as the ellipse dimple diameter ratio, ratio of the dimple depth to the average diameter, and ratio of the distance between dimples to the pitch are considered in the optimization. Twenty-one experimental points within design space are selected by Latin Hypercube Sampling. Each objective function values at these points are evaluated by RANS analysis and producing optimal point using surrogate model. The linear combination of heat transfer coefficient and friction loss related terms with a weighting factor is defined as the objective function. The results show that the optimized elliptical dimple shape improves considerably the heat transfer performance than the circular dimple shape.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.11 no.11
• /
• pp.5574-5591
• /
• 2017

Inspired by the idea of Artificial Immune System, many researches of wireless sensor network (WSN) intrusion detection is based on the artificial intelligent system (AIS). However, a large number of generated detectors, black hole, overlap problem of NSA have impeded further used in WSN. In order to improve the anomaly detection performance for WSN, detector generation mechanism need to be improved. Therefore, in this paper, a Differential Evolution Constraint Multi-objective Optimization Problem based Negative Selection Algorithm (DE-CMOP based NSA) is proposed to optimize the distribution and effectiveness of the detector. By combining the constraint handling and multi-objective optimization technique, the algorithm is able to generate the detector set with maximized coverage of non-self space and minimized overlap among detectors. By employing differential evolution, the algorithm can reduce the black hole effectively. The experiment results show that our proposed scheme provides improved NSA algorithm in-terms, the detectors generated by the DE-CMOP based NSA more uniform with less overlap and minimum black hole, thus effectively improves the intrusion detection performance. At the same time, the new algorithm reduces the number of detectors which reduces the complexity of detection phase. Thus, this makes it suitable for intrusion detection in WSN.