While the use of a Geospatial Information System (GIS) is nothing new to utilities, leveraging that system with the expanding sets of big data being produced along with the latest machine learning and statistical analysis techniques is new. At present, there is a real need for an industrywide program in GIS analysis that will disseminate the latest technologies and spur the development of commercial packages. Appropriating data cleanup algorithms and adapting them for use in utilities provides a shortcut to more refined and insightful analytical techniques. New insights as to how these algorithms might be used forms the basis for this report.
Geospatial information systems have been experiencing a decade-long breakout from their traditional role as map generators to near-real-time, integrated enterprise analysis tools. Technologies—such as spatial data mining, spatial cognition, geocomputation, and geoinformatics—are being applied more frequently. These technologies are helping to address important societal changes such as population growth, climate change, and the provision of adequate food, water, and energy.
GIS analysis is dependent on GIS data quality—a challenge because each record, feature, and attribute includes several facets that define complete quality. Poor data quality is of the utmost concern because of the potential for bad data to be supplied to other key systems, including the customer information system (CIS), work management system (WMS), and outage management system (OMS). In order to leverage the individual data management strengths of the systems, each must be integrated with the GIS to provide seamless, accurate, and timely data transfer.
To determine the number one issue that utilities have with GIS data and to examine the latest cleanup algorithms available to address that issue.
A comprehensive search of the GIS literature was performed to determine the status of the latest GIS automated data cleanup techniques. A phone survey was conducted to determine utility perceptions of the most urgent data issues. The question asked was:
“If you could correct one GIS data issue, regardless of currently having a method to do it, what would it be?” Issues identified were compared to the technologies and algorithms discussed in this report.
Utility GIS data issues continue to revolve around secondary circuit mapping, the relationship between the meter and the transformer, meter phasing, and conflation. Some techniques are available to address these issues, but very few of them take the form of commercial products. There is a significant opportunity for the development of GIS algorithms that utilize the latest machine learning and statistical analysis. A number of these early algorithms are described here including 1) spatial queries for automated data cleansing, 2) voltage profile correlation analysis using advanced meter infrastructure (AMI) data, 3) terrestrial based image analysis using neural networks, 4) remote sensing, and 5) proximity analysis to determine meter-transformer relationships. This report also focuses on types of GIS errors, accuracy standards, and quality assurance/quality control GIS tools.
Applications, Value, and Use
The intended audience for this report is anyone involved in GIS, GIS analytics, or big data analytics, with emphasis on the demands placed on geospatial data by the integrated grid and associated data analysis.