Overview of a New Power Grid Online Analysis Platform
By Mike Zhou, XueWei Shang, Lin Zhao, DongHao Feng, JianFeng Yan, DongYu Shi, and Ying Chen
A new online analysis platform has been developed to support the next generation online analysis system development. The goal is to achieve response speeds in the order of seconds to help the grid operator perform online analysis in real-time. In a pilot project, two new online analysis application functions have been developed based on the platform to create a new online analysis system to augment the existing online analysis system. The new online analysis system has been deployed and is running in a provincial dispatching center in China. The preliminary testing data indicates that the new online analysis system can achieve sub-second response speed.
After a decade of development, the online analysis has been widely used in the power dispatching control centers in China. In an online analysis system, power grid operation measurement data (RTU) is processed first by Supervisory Control And Data Acquisition (SCADA) system and then by State Estimation (SE) to estimate the state of the grid operation condition and establish a power flow study case before starting the online Dynamic Security Assessment (DSA) analysis. The State Grid of China currently is modeled using a ~40K-bus online analysis network model. An online analysis round-trip, including SCADA, State Estimation and DSA analysis, currently takes about 10 minutes to complete. Therefore, the current online analysis system, when used to handle large-scale power network models, is a near real-time system with response speed in the order of minutes. The online DSA analysis is currently performed periodically at an interval of 15 minutes.
With the development of the ultra-high voltage AC/DC national grid in China, there is a strong need for a new generation of fast online analysis systems with response speed in the order of seconds. Power grid dynamic process, due to the disturbance, develops at a speed of seconds. The fast (second-order) online analysis is expected to help the grid operator perform real-time analysis and assist the decision support process, while the system dynamic process is developing, rather than the after-the-fact analysis. A new fast online analysis system development project, sponsored by the State Grid of China, was started in 2006. The primary goal of the project was set to reduce the online analysis overall round-trip time, from data acquisition to complete the analysis, from the current proximate 10 minutes to less than 60 seconds. The project development work has been completed at the end of 2018. In this Newsletter, we present a high-level overview of the project and some of the preliminary performance testing results. The overview will include the project high-level solution architecture, the Digital Twin concept used in framing the solution architecture, an online analysis platform to support the realization of the solution architecture, and a pilot online analysis system developed based on the online analysis platform.
Solution architecture - A new online analysis solution architecture was proposed in the project. In the proposed architecture, a new parallel data processing path is added to the existing DSA system. In the new path, a data grid is used to host a network analysis model. The model is updated in real-time by subscribing to the grid change events published by the SCADA system. The update to the model is also published as model change events. These model change events are subscribed by a Complex Event Processing (CEP) engine to perform situation awareness analysis. If considerable changes in term of grid security are detected based on the predefined grid operation rules, the CEP engine will drive the data-driven, AI-based fast DSA modules to perform the security assessment.
Digital Twin (DT) - DT has been in the Gartner’s Top 10 Strategic Technology Trends list every year since 2017. The DT concept was proposed originally to address large-scale complex manufacturing issues. In the DT approach, the virtual model could be updated to “mirror”, often in real-time, the physical system. The analysis and simulation could be performed on the virtual model for real-time decision support purpose. In our project, an Online Analysis Digital Twin (OADT) has been implemented. In OADT, the network analysis model, or the virtual model, connects to the power grid, or the physical system, through the SCADA message subscription. The network analysis model, therefore, mirrors the power grid for the online analysis purpose. The mirroring is in real-time with sub-second delay, based on our actual system measurement data. In addition, In-Memory Computing (IMC) based grid analysis algorithms and CEP-based operation monitoring rules could be applied to the virtual model to perform the situation awareness analysis for the real-time decision support.
Online Analysis Platform – Based on the new online analysis solution architecture and the OADT approach , an online analysis platform has been developed in our project. The platform supports both data-driven as well as conventional model-driven online analysis applications. As part of the platform, a set of new supporting technologies have been developed, including: 1) Real-time grid network analysis model hosted in a data grid; 2) IMC based grid network analysis; 3) High-performance parallel computing based grid analysis algorithm implementation; 4) CEP engine; 5) Machine learning based DSA research and development environment.
Online Analysis System – Based on the new online analysis platform, a new pilot online system has been built and deployed in a provincial dispatching center in China. This online system currently has two data-driven analysis applications/functions: 1) Neural Network (NN) based CCT (Critical Clearing Time) prediction; 2) Grid operation monitoring rule evaluation. More online analysis applications/functions are expected to be added to the online analysis system in the near future. The NN-based CCT prediction is based on a set of off-line trained NN models and the online grid operation state snapshot stored in the network analysis model to predict the grid CCT for a set of selected faults in real-time. The grid operation monitoring rule evaluation is based on the online grid operation state snapshot. The rule evaluation is hosted in the CEP engine.
End-to-end Response Speed - The new online system has been running since Dec. 2018 in a provincial dispatching center in China with a network mode size of ~1K-bus. The grid network analysis model consists of a physical (node/breaker) model and a simulation (bus/branch) model. The RTU info, published by the SCADA system, is subscribed by the physical model to update the physical model continuously. Topology analysis is performed to map the physical model changes to update the simulation model to keep the simulation model in-synch with the physical model. The online analysis is performed periodically at an interval of 20 seconds. When performing the online analysis, an IMC approach-based SE algorithm is applied to the simulation model to create a converged power flow study case to provide a snapshot for the two applications. It is observed based the actual measurement data that the total round-trip computation time, including the model updates, SE algorithm execution, and the NN model prediction or the operation rule evaluation, is less than 300 ms.
Recently there is wide interest in applying AI, especially the Deep Machine Learning (ML) approach, to the power grid online analysis and monitoring. The online analysis platform attempts to provide a unified research, development and runtime environment for data-driven AI-based grid online applications/functions to fill the gap between the off-line power grid simulation results/conclusions and creation/execution of the online operation monitoring rules by the grid operators. It is our hope that using this kind advanced AI-based real-time approach will result in more economic and secure grid operation zone and control strategies in the future.
In summary, an end-to-end optimization approach is used to reduce the online analysis system overall round-trip response time, including data acquisition, data processing, and DSA analysis, to the order of seconds. Based on the Digital Twin framework, a new online analysis solution architecture has been proposed. A new online analysis platform and a set of supporting technologies have been developed for the realization of the solution architecture. Based on the platform, a pilot online analysis system has been built and deployed in a provincial dispatching center in China. The preliminary testing data indicates that the pilot online analysis system can achieve sub-second end-to-end round-trip response speed.
Mike Zhou PhD (M’90) Chief Scientist, State Grid Electric Power Research Institute of China. He was a Senior Computer System Architect at TIBCO Software Inc, specializing in large-scale real-time information system integration 2000-2014. In 1996, he introduced the object-oriented programming approach to power system simulation. In 2005, he created the InterPSS project, a free and open power system simulation software development project. He has over thirty years of experience in the design and implementation of object-oriented, service-oriented, distributed large-scale, high-performance real-time computing systems, applications, and IT infrastructure.
XueWei Shang Sr Engineer, Managing Director, Beijing KeDong Company, Nari Corporation. A Member of the national power system management and information exchange Standardization Technical Committee (SAC/TC82). He participated as one of the principal directors/developers in the development of the Chinese power dispatching control center application platform （CC-2000，D5000）. His research interests include power dispatching control center application platform design, platform middleware development, and platform information exchange standardization.
Lin Zhao Sr Engineer, Director of Technology, Beijing KeDong Company, Nari Corporation. He participated as one of the principal managers/developers in the development of the Chinese power dispatching control center application platform (CC-2000, D5000). His research interests include power dispatching control center application platform design, platform human interface development, platform middleware development.
DongHao Feng Engineer, Beijing KeDong Company, Nari Corporation. He obtained his B.S. and M.S. from the South China University of Technology. His current research interests are power system modeling and simulation, in-memory computing and its applications to power system simulation and analysis.
JianFeng Yan PhD, China Electric Power Science Research Institute. He has been engaged in the research and application of power system online Dynamic Security Analysis(DSA) technology. He led the research and development of 19 DSA systems in China Power Grid. He wrote the DSA standard for China State Grid. He has over ten years of experience in the design and implementation of DSA.
DongYu Shi received his B.S. and M.S. degree from Tsinghua University in 2003 and 2006 respectively. Currently, he is working in China Electric Power Research Institute (CEPRI). He participated as one of the principal researchers/developers in the development of the dynamic security assessment (DSA) system, which is widely used in China. His research interests include power system dynamic security assessment, parallel computing, and machine learning.
Ying Chen (M’07) received the B.E. and Ph.D. degrees from Tsinghua University, Beijing, China, in 2001 and 2006, respectively, both in electrical engineering. He is currently an Associate Professor at the Department of Electrical Engineering, Tsinghua University. His research interests are in the areas of power system dynamics and simulation, parallel computing, and cyber-security.
To have the eNewsletter delivered monthly to your inbox, join the IEEE Smart Grid Community.
IEEE Smart Grid Newsletter Editors
To view archived articles, and issues, which deliver rich insight into the forces shaping the future of the smart grid, please visit the IEEE Smart Grid Resource Center.