Collections hosted in NI4AI and how to find them
January 01, 2100
This blog post lists datasets hosted in NI4AI by collection name, and links to relevant documentation and blog posts.
This collection includes streaming sensors deployed as part of NI4AI. Sensors are deployed in wall outlets, and report single phase voltage magnitude and angle, as well as frequency. Metadata for sensors in the collection give additional information about siting, sensors, and about the grid they are connected to.
This collection includes three sensors deployed as part of the Texas Synchrophasor Network. The data includes one sensor on each interconnection. Anyone using the data is requested to cite the following publication.
Mack Grady and David Costello. “Implementation and Application of an Independent Texas Synchrophasor Network.” SEL Journal of Reliable Power. Vol. 2, No. 2. May, 2011. URL: http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.857.5454&rep=rep1&type=pdf
This collection includes archived data for 6 distribution PMUs deployed as part of a pilot program that ran for 18 months. The pilot was run as part of an ARPA-E project funded under Open 2012 which resulted in the development of the micro-PMU.
Streams: 78 Points: 331,983,600,006
This collection includes 3 months of archived data for 10 distribution PMUs deployed on a single distribution feeder. PMUs are numbered in order of relative distance from the substation, where PMU1 is at the substation.
Streams: 130 Points: 104,342,308,680
This collection gives 1-minute snapshots of three wide-area events across a network of 23 PMUs each with a sampling rate of 30 samples per second. The sensors located on the transmission grid and the data are fully anonymized. In other words, the time stamps have been masked and the geographic locations of the sensors removed.
Streams: 1,311 Points: 2,361,111
This collection includes point on wave measurements for upwards of 250 faults. Each event includes a 2-second snapshot of data, including current and voltage measurements for all three phases. Metadata for each event includes labels obtained through detailed investigations into each event. Labels include the cause, and (where relevant) the type of equipment that failed.
Streams: 1773 Points: 12,033,799
This collection includes point on wave measurements recorded over 24 hours on an underground distribution line. The data reports current and voltage measurements across all three phases.
Streams: 6 Points: 1,990,153,021
This dataset gives 1 hour of PMU data recorded on September 6, 2020 between 5 and 6PM Pacific. One of the ni4ai sensors saw a frequency disturbance during that time, and we reached out to the Texas Synchrophasor Network to see if their sensors had seen it as well. The event occurred in WECC, and forms the basis of an exercise on data quality assessment (link).
Streams: 9 Points: 971,631
The data was collected by three point on wave sensors as part of an experiment run on the grid that serves the EPFL university campus in Switzerland. The grid includes a 560 kWh lithium-ion battery. The data report six discrete events, including baseline measurements, and measurements recorded during five charging and discharging events. Each event lasts 2 seconds with measurements reported every 20us (sampling rate of 50 kHz).
Streams: 12 Points: 7,200,000
This dataset reports geomagnetic field measurements taken at 14 different latitudes on the surface of the earth from 2013 to 2019. Each measurement is reported as a vector quantity with coordinates in the North/South, East/West and vertical dimensions. Data are reported as a 1-second average derived from a 100Hz geomagnetic field sensor. Additional information about the measurement infrastructure and data format are available at the link above.
Streams: 56 Points: 10,826,130,797
This data reports EKG measurements for one patient. EKG (short for electrocardiogram) measurements for an 81 year old female. Metadata fields include additional information about the patient’s medical history.
Streams: 15 Points: 1,728,000
Navigating This Blog
Here's a map to help you quickly find what you're looking for on the NI4AI blog.
January 01, 2100
About the "events" data collection
Anonymized event snapshots to study wide area dynamics
December 01, 2020
NI4AI is short for A National Infrastructure for AI on the Grid. We are an ARPA-E funded initiative designed to eliminate barriers to developing new analytical tools for the grid. We provide a software platform, open access datasets, content, and access to a community of analysts exploring new applications for real-world sensor data.