Here’s how we approached the problem:  

First, experts examined the battery voltage data received by Eaton's remote monitoring system, which is collected through communication cards installed on Uninterruptible Power Supplies (UPS). They were able to identify unique electrical patterns that showed how the battery's condition worsened over time. This valuable information was then used to train an Artificial Intelligence (AI) model. The AI model proved to be effective in detecting the early signs of battery wear and was able to predict when the battery would no longer be useful, 60 days before it actually reached the end of its life.

To keep customers informed, the same remote monitoring system sends two alerts: the first alert is sent as soon as any sign of battery wear is detected, and the second alert is sent when the battery is about 60 days away from reaching the end of its useful life. This advanced notice gives customers ample time to plan and replace their batteries, preventing any critical interruptions or expensive downtime.

The prediction algorithm is designed to work with all Valve-Regulated Lead-Acid (VRLA) batteries, provided Eaton's advanced battery management (ABM) feature is activated. The algorithm is smart enough to determine the type of battery and whether ABM is active, simply by analyzing the battery voltage data. It is also robust, meaning it can handle unexpected power outages and gaps in data communication. Additionally, the algorithm adjusts to each individual battery by keeping track of how its condition changes over time.