Cloud-based Data Exchange Infrastructure for Wide Area Monitoring of Bulk Electric Power Grids

摘要

The GridCloud system is designed to leverage scalable and inexpensive cloud computing resources in support of bulk power grid monitoring, control, and coordinated operations. Basic functionality includes a data collection layer that can capture data from sensors such as synchrophasor measurement units (PMUs), a novel file system into which data are archived for extremely fast and reliable retrieval using temporal indexing, and a cloud manager used to configure the system and repair it after server failures. In the new work reported here, ISO-New England and New York Power Authority shared real-time data through GridCloud, streaming PMU data over secure links, then requesting state estimation. We illustrate this sharing by running a continuous state estimation in the cloud, yielding output that can be visualized on the cloud platform, securely exported to operator control rooms, or used as input for further analysis. State estimation is just one of many options, and the entire platform is designed to be easily extended. To underscore this opportunity, we show how we bridged from GridCloud's data archive to the Spark/Databricks analytics infrastructure, which offers a mix of classic computational tools, modern machine-learning packages, and big data analytics. We illustrate the resulting capability by applying a standard machine learning technology to the real-time data stream captured by the system (specifically, we train a convolutional neural network to predict bus voltages in a wide-area power network, and show how it can be used to estimate network state even when a major weather event is causing a severe disruption that has compromised normal monitoring). The broader point is that GridCloud can support a diversity of tools, including both standard power grid control applications and novel ones that use the extensive programming capabilities, machine-learning tools, and data analytics support available in modern cloud settings.