Real-Time Demand Response Management for Controlling Load using Deep Reinforcement Learning

YONGJIANG ZHAO,SENFENG CEN,SEUNG JE SEONG,IBRAHIM ALIYU,CHANG GYOON LIM 한국콘텐츠학회 2021 한국콘텐츠학회 ICCC 논문집 Vol.2021 No.12

With the rapid economic growth and the improvement of living standards, electricity has become an indispensable energy source in our lives. Therefore, the stability of the grid power supply and the conservation of electricity are very important. Controlling load as an important part of incentive demand response (DR), it can achieve a rapid response and improve demand side resilience. Controlling load by manually formulating rules, some devices can be selectively turn off during peak power consumption. It reduces the cost of manual operation while reducing the peak value. However, it is difficult to optimize methods based on manual rules. In this paper, Soft Actor-Critic (SAC) is used as a control algorithm to optimize the control strategy. In order to focus on the optimization of the algorithm, the dataset in this paper is based on CityLearn. The results show that through the coordination of SAC to control load, realizes the goal of reducing both the peak load demand of power and the operation costs on the premise of regulating voltage to the safe limit.

Robot Automatic Path Planning by Avoiding Obstacle using Double Deep Q Networks on the Testbed

Yongjiang Zhao,Seung Je Seong,Chenglin Fan,Hur J. G.,Chang Gyoon Lim 한국콘텐츠학회 2022 한국콘텐츠학회 ICCC 논문집 Vol.2022 No.12

Seasonal Power Demand Forecasting by Autoformer Deep Learning to Optimize Power Usage Policies

YONGJIANG ZHAO,Esanov Alibek Rustamovich,Haoyi Zhong,Chang Gyoon Lim 한국콘텐츠학회 2022 한국콘텐츠학회 ICCC 논문집 Vol.2022 No.12

이중 심층 Q 네트워크 기반 장애물 회피 경로 계획

자오 용지앙(Yongjiang Zhao),첸센폰(Senfeng Cen),성승제(Seung-Je Seong),허정규(J. G. Hur),임창균(Chang-Gyoon Lim) 한국전자통신학회 2023 한국전자통신학회 논문지 Vol.18 No.2

심층 강화 학습(Deep Reinforcement Learning)을 사용한 경로 계획에서 장애물을 자동으로 회피하기 위해 로봇을 학습시키는 일은 쉬운 일이 아니다. 많은 연구자가 DRL을 사용하여 주어진 환경에서 로봇 학습을 통해 장애물 회피하여 경로 계획을 수립하려는 가능성을 시도하였다. 그러나 다양한 환경에서 로봇과 장착된 센서의 오는 다양한 요인 때문에 주어진 시나리오에서 로봇이 모든 장애물을 완전히 회피하여 이동하는 것을 실현하는 일은 흔치 않다. 이러한 문제 해결의 가능성과 장애물을 회피 경로 계획 실험을 위해 테스트베드를 만들었고 로봇에 카메라를 장착하였다. 이 로봇의 목표는 가능한 한 빨리 벽과 장애물을 피해 시작점에서 끝점까지 도달하는 것이다. 본 논문에서는 벽과 장애물을 회피하기 위한 DRL의 가능성을 검증하기 위해 이중 심층 Q 네트워크(DDQN)를 제안하였다. 실험에 사용된 로봇은 Jetbot이며 자동화된 경로 계획에서 장애물 회피가 필요한 일부 로봇 작업 시나리오에 적용할 수 있을 것이다. It remains a challenge for robots to learn avoiding obstacles automatically in path planning using deep reinforcement learning (DRL). More and more researchers use DRL to train a robot in a simulated environment and verify the possibility of DRL to achieve automatic obstacle avoidance. Due to the influence factors of different environments robots and sensors, it is rare to realize automatic obstacle avoidance of robots in real scenarios. In order to learn automatic path planning by avoiding obstacles in the actual scene we designed a simple Testbed with the wall and the obstacle and had a camera on the robot. The robot's goal is to get from the start point to the end point without hitting the wall as soon as possible. For the robot to learn to avoid the wall and obstacle we propose to use the double deep Q networks (DDQN) to verify the possibility of DRL in automatic obstacle avoidance. In the experiment the robot used is Jetbot, and it can be applied to some robot task scenarios that require obstacle avoidance in automated path planning.

VALUE CHAIN OF LUXURY INDUSTRY - BRAND AS AN INTERDEPENDENT VALUE CREATION STEP

Pierre Xiao Lu,Wenwen Zhao,Yongjiang Shi 글로벌지식마케팅경영학회 2015 Global Fashion Management Conference Vol.2015 No.06

Increasing attention has been paid to marketing and consumer behavior of luxury industry but research into value creation network and operational mechanisms is very limited. This study focuses on two aspects of the luxury industry: luxury brand and value chain, to inform a comprehensive understanding of the value creation process for high value added brands. In luxury industry, the key elements that create and deliver value are brand, design and research, production, distribution, and retail. A clear brand identity is found as the first step of this value chain, which influences the choices of all other activities. Luxury goods companies will align all the activities in line with brand identity to deliver consistent tangible and intangible values to end users. Furthermore, a luxury value chain is a holistic network with strong coordination among its elements. A combined approach of case study and secondary data collection is pursued. A sample of 9 luxury companies within 6 selected industries is investigated. Data is qualitatively collected via semi-structured interviews, document analysis, and observations as a triangulation approach for the purpose of ensuring the reliability of the research data. Multiple interviews of the general manager, industrial manager, brand/communication manager, creative director, and store manager are conducted in each company to achieve a broader perspective and also make data triangulation procedures possible. This research contributes to the luxury brands management as well as value chain concept. It discusses the value creation network and operational mechanism from a less explored corporate perspective. It unveils a secretive existence of brand in value generation process and further establishes a model to amplify the relationship between each activity in the value chain. Also, it expands the research of value chain into luxury industry. It argues that a supply leading value chain can also command a premium rather than the customer-centric value chain discussed by most researchers recently. It also provides valuable insights for companies who want to have a high-end market position. It shows that the widely adopted luxury strategy invented mainly by French and Italian companies employs fundamentally different rules from those of fast-moving consumer goods in mass market. In short, a luxury strategy is different in nature, not in level.

Analysis of contact characteristics of ball screws under the combined loads considering non-uniform load distribution

YongJiang Chen,Jianghai Zhao,Chuangfan Yuan,Jingkai Sun 대한기계학회 2023 JOURNAL OF MECHANICAL SCIENCE AND TECHNOLOGY Vol.37 No.4

An analysis model of the contact characteristics of the double-nut preloaded ball screw which can consider the combined action of the axial load and the overturning moment is established, the model can consider non-uniform load distribution, and the validity of the model is verified by experiments. Based on this model, the influence of the combined action of axial load and overturning moment on the average value and distribution fluctuation value of the contact angle of the ball screw, the number of unloaded balls, the distribution fluctuation value of the contact load, and the static contact stiffness and other contact parameters are systematically analyzed, and get the following conclusions: The contact parameters of the ball screw are mainly affected by the axial load and the contact state between the ball and the raceway. The gradually increasing axial load basically reduces the influence of the overturning moment on the contact parameters of the ball screw. There is a critical value of the load in the axial direction within the range of the action value of the preload load. Only when the critical value is above the critical value, the contact parameters will change regularly with the changes in the axial stiffness and overturning moment. However, different critical values for the regular variation of contact parameters with axial load and overturning moment vary.

Abnormal State Detection using Memory-augmented Autoencoder technique in Frequency-Time Domain

Haoyi Zhong,Yongjiang Zhao,Chang Gyoon Lim 한국인터넷정보학회 2024 KSII Transactions on Internet and Information Syst Vol.18 No.2

With the advancement of Industry 4.0 and Industrial Internet of Things (IIoT), manufacturing increasingly seeks automation and intelligence. Temperature and vibration monitoring are essential for machinery health. Traditional abnormal state detection methodologies often overlook the intricate frequency characteristics inherent in vibration time series and are susceptible to erroneously reconstructing temperature abnormalities due to the highly similar waveforms. To address these limitations, we introduce synergistic, end-to-end, unsupervised Frequency-Time Domain Memory-Enhanced Autoencoders (FTD-MAE) capable of identifying abnormalities in both temperature and vibration datasets. This model is adept at accommodating time series with variable frequency complexities and mitigates the risk of overgeneralization. Initially, the frequency domain encoder processes the spectrogram generated through Short-Time Fourier Transform (STFT), while the time domain encoder interprets the raw time series. This results in two disparate sets of latent representations. Subsequently, these are subjected to a memory mechanism and a limiting function, which numerically constrain each memory term. These processed terms are then amalgamated to create two unified, novel representations that the decoder leverages to produce reconstructed samples. Furthermore, the model employs Spectral Entropy to dynamically assess the frequency complexity of the time series, which, in turn, calibrates the weightage attributed to the loss functions of the individual branches, thereby generating definitive abnormal scores. Through extensive experiments, FTD-MAE achieved an average ACC and F1 of 0.9826 and 0.9808 on the CMHS and CWRU datasets, respectively. Compared to the best representative model, the ACC increased by 0.2114 and the F1 by 0.1876.

Electricity Load Pattern Analysis using Fuzzy C-Means Clustering Algorithm

SENFENG CEN,YONGJIANG ZHAO,YONG MIN KIM,IBRAHIM ALIYU,J. G. Hur,CHANG GYOON LIM 한국콘텐츠학회 2021 한국콘텐츠학회 ICCC 논문집 Vol.2021 No.12

Energy consumption has grown explosively in recent years, and energy shortages occurred occasionally. Demand Response (DR) programs could help energy management entities to balance power generation and consumption. The electricity consumption data were collected with the widely deployed advanced smart meters, which contain valuable information. Consumption data can be used to explore the consumption behavior and help Demand Side Management (DSM). This study proposed clustering algoritsshms to obtain the representative load patterns based on diurnal load profiles. First, we applied discrete wavelet transform (DWT) to extract features from 10-second interval daily electricity consumption data. Then using Principal Component Analysis (PCA) for dimensionality reduction. Lastly, implement Fuzzy C-Means (FCM) clustering algorithm to segment preserved features. Additionally, we discuss the clustering result and load pattern analysis of the dataset with respect to the electricity pattern.

OneNet 클라우드 컴퓨팅 기반 실시간 홈 보안 시스템

김강철(Kang-Chul Kim),Yongjiang Zhao 한국전자통신학회 2021 한국전자통신학회 논문지 Vol.16 No.1

본 논문은 스마트폰으로 집 내의 상태를 제어하는 OneNet 클라우드 플랫폼 기반 실시간 홈 보안 시스템을 설계한다. 제안된 시스템은 로컬과 클라우드 지역으로 구분된다. 로컬 지역은 I/O 디바이스, 라우터와 센서 데이터를 수집, 모니터링하고 클라우드로 데이터를 전송하는 라즈베리파이로 구성되며, 라즈베리파이에 플래스크 웹 서버가 구현된다. 사용자가 집에 있을 경우 플래스크 웹 서버를 통하여 직접 데이터에 접근할 수 있다. 클라우드 지역에서 사용되는 플랫폼은 중국 통신회사의 OnetNet이며, 원격 접속 서비스를 제공한다. 스마트폰 에서 사용자와 홈보안 시스템 사이의 통신을 위하여 하이브리드 앱이 개발되고, 센서 데이터와 비디오스트림을 전송하기 위하여 EDP와 RTSP 프로토콜을 파이썬 언어로 구현한다. 구현된 시스템에서 사용자는 스마트 폰으로 센서 데이터. 비디오스트림과 위험이 발생시에 경고 문자를 받을 수 있고, OneNet 클라우드를 통하여 원격으로 집 내의 상태를 모니터링하고, 제어할 수 있다. This paper builds a real-time home security system based on the OneNet cloud platform to control the status of the house through a smartphone. The system consists of a local part and a cloud part. The local part has I/O devices, router and Raspberry Pi (RPi) that collects and monitors sensor data and sends the data to the cloud, and the Flask web server is implemented on a Rasberry Pi. When a user is at home, the user can access the Flask web server to obtain the data directly. The cloud part is OneNet in China Mobile, which provides remote access service. The hybrid App is designed to provide the interaction between users and the home security system in the smartphone, and the EDP and RTSP protocol is implemented to transmit data and video stream. Experimental results show that users can receive sensor data and warning text message through the smartphone and monitor, and control home status through OneNet cloud.

CNN-LSTM for Smart Grid Energy Consumption Prediction

Ibrahim Aliyu,Usmonov Kamoliddin,Seung Je Seong,Senfeng Cen,Yongjiang Zhao,Chang Gyoon Lim 한국콘텐츠학회 2021 한국콘텐츠학회 ICCC 논문집 Vol.2021 No.12

This paper aims to forecast monthly electricity across different kinds of consumers such as Residential, Industrial, Official and Commercial sectors. To achieve this, a CNN-LSTM prediction framework for energy demand in smart grid is proposed. Efficient consumption prediction is essential for effective Demand Response (DR) which can enable consumers to minimize their energy usage through proper load curtailment, consumption shift over time, or energy generation and storage at certain times to offer flexibility in the grid.