• 전자공학회논문지SC
• /
• 제41권3호
• /
• pp.39-46
• /
• 2004

본 논문은 IPMSM 드라이브의 고성능 속도 제어를 위하여 퍼지제어와 신경회로망을 혼합 구성한 적응 FNN 제어기를 제시한다. 적응 FNN 제어기는 기준 모델에 기초한 적응 메카니즘을 적용하여 신경회로망의 고도의 적응제어와 퍼지제어기의 강인성 제어의 장점들을 접목한다. 적응 FNN 제어기의 출력은 FNN 제어기의 출력과 적응 퍼지제어의 출력을 합하여 출력을 얻는다. 적응 FNN 제어기는 다양한 동작조건에서 응답특성을 분석하고 평가한다. 제시한 적응 FNN 제어기의 타당성은 IPMSM 드라이브 시스템에 적용하여 성능 결과로 입증한다.

• Journal of Communications and Networks
• /
• 제9권3호
• /
• pp.319-329
• /
• 2007

This paper presents a price-based quality-of-service (QoS) control framework for two-class network services, in which circuit-switched and packet-switched services are defined as "premium service class" and "best-effort service class," respectively. Given the service model, a customer may decide to use the other class as a perfect or an imperfect substitute when he or she perceives the higher utility of the class. Given the framework, fixed-point problems are solved numerically to investigate how static pricing can be used to control the demand and the QoS of each class. The rationale behind this is as follows: For a network service provider to determine the optimal prices that maximize its total revenue, the interactions between the QoS-dependent demand and the demand-dependent QoS should be thoroughly analyzed. To test the robustness of the proposed model, simulations were performed with gradually increasing customer demands or network workloads. The simulation results show that even with substantial demands or workloads, self-adjustment mechanism of the model works and it is feasible to obtain fixed points in equilibrium. This paper also presents a numerical example of guaranteeing the QoS statistically in the short term-that is, through the implementation of pricing strategies.

• 한국산학기술학회논문지
• /
• 제10권8호
• /
• pp.1993-1997
• /
• 2009

산업 자동화의 고정밀도에 따라 직류서보 전동기는 강인제어가 요구되고 있다. 하지만 PID 제어기를 갖는 전동기 제어 시스템이 부하 외란의 영향을 받게되면 제어 시스템의 강인제어는 어렵게 된다. 이에 대한 보완적인 한 방법으로 본 논문에서는 전동기 제어시스템을 위한 PID-신경망 복합형 제어기법을 제시하였다. 신경망 제어기의 출력은 부하 외란 인가시에 발생되는 오차와 오차 변화율에 의해서 결정된다. 신경망 제어기를 이용한 직류서보 전동기의 강인제어는 시뮬레이션에 의하여 확인하였다.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 2003년도 학술회의 논문집 정보 및 제어부문 B
• /
• pp.882-885
• /
• 2003

In this paper, nonlinear echo cancellation using a correlation LMS (CLMS) algorithm is proposed to cancel the undesired nonlinear echo signals generated in the hybrid system of the telephone network. In the telephone network, the echo signals may result the degradation of the network performance. Furthermore, digital to analog converter (DAC) and analog to digital converter (ADC) may be the source of the nonlinear distortion in the hybrid system. The adaptive filtering technique based on the nonlinear Volterra filter has been the general technique to cancel such a nonlinear echo signals in the telephone network. But in the presence of the double-talk situation, the error signal for tap adaptations will be greatly larger, and the near-end signal can cause any fluctuation of tap coefficients, and they may diverge greatly. To solve a such problem, the correlation LMS (CLMS) algorithm can be applied as the nonlinear adaptive echo cancellation algorithm. The CLMS algorithm utilizes the fact that the far-end signal is not correlated with a near-end signal. Accordingly, the residual error for the tap adaptation is relatively small, when compared to that of the conventional normalized LMS algorithm. To demonstrate the performance of the proposed algorithm, the DAC of hybrid system of the telephone network is considered. The simulation results show that the proposed algorithm can cancel the nonlinear echo signals effectively and show robustness under the double-talk situations.

• Nuclear Engineering and Technology
• /
• 제54권12호
• /
• pp.4751-4758
• /
• 2022

Two-phase flow may almost exist in every branch of the energy industry. For the corresponding engineering design, it is very essential and crucial to monitor flow patterns and their transitions accurately. With the high-speed development and success of deep learning based on convolutional neural network (CNN), the study of flow pattern identification recently almost focused on this methodology. Additionally, the photographing technique has attractive implementation features as well, since it is normally considerably less expensive than other techniques. The development of such a two-phase flow pattern online monitoring system is the objective of this work, which seldom studied before. The ongoing preliminary engineering design (including hardware and software) of the system are introduced. The flow pattern identification method based on CNNs and transfer learning was discussed in detail. Several potential CNN candidates such as ALexNet, VggNet16 and ResNets were introduced and compared with each other based on a flow pattern dataset. According to the results, ResNet50 is the most promising CNN network for the system owing to its high precision, fast classification and strong robustness. This work can be a reference for the online monitoring system design in the energy system.

• 한국인공지능학회지
• /
• 제11권2호
• /
• pp.1-6
• /
• 2023

This paper presents a comprehensive scheme for assessing the importance of multiple microgrids (MGs) network that includes distributed energy resources (DERs), renewable energy systems (RESs), and energy storage system (ESS) facilities. Due to the uncertainty of severe weather, large-scale cascading failures are inevitable in energy networks. making the assessment of the structural vulnerability of the energy network an attractive research theme. This attention has led to the identification of the importance of measuring energy nodes. In multiple MG networks, the energy nodes are regarded as one MG. This paper presents a modified PageRank algorithm to assess the importance of MGs that include multiple DERs and ESS. With the importance rank order list of the multiple MG networks, the core MG (or node) of power production and consumption can be identified. Identifying such an MG is useful in preventing cascading failures by distributing the concentration on the core node, while increasing the effective link connection of the energy flow and energy trade. This scheme can be applied to identify the most profitable MG in the energy trade market so that the deployment operation of the MG connection can be decided to increase the effectiveness of energy usages. By identifying the important MG nodes in the network, it can help improve the resilience and robustness of the power grid system against large-scale cascading failures and other unexpected events. The proposed algorithm can point out which MG node is important in the MGs power grid network and thus, it could prevent the cascading failure by distributing the important MG node's role to other MG nodes.

• Smart Structures and Systems
• /
• 제26권2호
• /
• pp.175-184
• /
• 2020

Conventional Monte Carlo simulation-based methods for seismic risk assessment of water networks often require excessive computational time costs due to the hydraulic analysis. In this study, an Artificial Neural Network-based surrogate model was proposed to efficiently evaluate the flow-based system reliability of water distribution networks. The surrogate model was constructed with appropriate training parameters through trial-and-error procedures. Furthermore, a deep neural network with hidden layers and neurons was composed for the high-dimensional network. For network training, the input of the neural network was defined as the damage states of the k-dimensional network facilities, and the output was defined as the network system performance. To generate training data, random sampling was performed between earthquake magnitudes of 5.0 and 7.5, and hydraulic analyses were conducted to evaluate network performance. For a hydraulic simulation, EPANET-based MATLAB code was developed, and a pressure-driven analysis approach was adopted to represent an unsteady-state network. To demonstrate the constructed surrogate model, the actual water distribution network of A-city, South Korea, was adopted, and the network map was reconstructed from the geographic information system data. The surrogate model was able to predict network performance within a 3% relative error at trained epicenters in drastically reduced time. In addition, the accuracy of the surrogate model was estimated to within 3% relative error (5% for network performance lower than 0.2) at different epicenters to verify the robustness of the epicenter location. Therefore, it is concluded that ANN-based surrogate model can be utilized as an alternative model for efficient seismic risk assessment to within 5% of relative error.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2000년도 추계종합학술대회 논문집(1)
• /
• pp.21-24
• /
• 2000

End-to-end congestion control mechanism have been critical to the robustness and stability of the Internet. Most of today's Internet traffic is TCP, and we expect this to remain so in the future. TCP/IP is the intermediate transport layer candidate for today's applications. TCP uses an adaptive window-based flow control. The congestion avoidance and control algorithms deployed by TCP aims at using the available network bandwidth. This paper compares different congestion control policies, and proposes the new design mechanism for future public networks

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2003년도 하계종합학술대회 논문집 V
• /
• pp.2569-2572
• /
• 2003

In this paper, we Proposed the inverted pendulum control method using single neuron neural network that have weights as PID parameters. The proposed method has three inputs(proportion, integration, differentiation term of the error), and uses weights as P, I, D parameters. In order to verify the effectiveness of the proposed method, we experimented on the rotary inverted pendulum with load effect disturbance. The results showed the effectiveness and robustness of the proposed pendulum controller.

• Journal of the Korean Data and Information Science Society
• /
• 제14권1호
• /
• pp.67-74
• /
• 2003

Median regression analysis has robustness properties which make it an attractive alternative to regression based on the mean. Support vector machine (SVM) is used widely in real-world regression tasks. In this paper, we propose a new SV median regression based on check function. And we illustrate how this proposed SVM performs and compare this with the SVM based on absolute deviation loss function.