We applied machine learning to help a company providing grid testing to major utilities to more effectively identify potential equipment failures.
Exacter provides grid testing to major utilities to help them identify and maintain the equipment in the power grid. Their business model involves deploying a fleet of vehicles to drive along the power lines and listen for signals from failing equipment. This model, while effective, can miss some problems since vehicles only travel the grid every few months.
Exacter wanted to install continuous sensors on the power poles and collect data at regular intervals throughout the day (and night). But the company didn’t have the in-house expertise to develop an internet of things (IoT) network, create a testing plan to deploy sensors at optimal locations to collect data, or build a machine learning algorithm to identify potential failures.
The company chose to partner with g2o because we had both the data and analytics expertise they needed.
We helped Exacter develop this new method for supporting utility maintenance. The g2o team developed a sensor network to capture signals from deteriorating equipment. We created an analytic database to store the sensor data as it was collected each day.
Then our data science team developed a machine learning algorithm to find failing equipment and direct the maintenance crews to the trouble spots before failures and outages occur.
Our machine learning model was able to detect defective devices that Exacter’s existing mobile line monitoring devices missed. As a result, g2o proved the value of Exacter’s new maintenance method and how it could improve efficiency for the company.
