Sonasoft’s (SSFT.OTC) new artificial intelligence solution with Delaware Electric Cooperative (DEC) went live at 3 p.m. on Monday, June 29, according to a press release.
The project uses Sonasoft’s AI Bot Engine, called NuGene, as a predictor for peak electricity usage. This gives DEC more insight into the demand for their product, and helps htem slash their operating costs.
“The model combines actual electricity load consumption and weather data to accurately forecast peak and non-peak days. This model has been trained and optimized to accurately predict both the type of peak and the precise timing of the peak. Furthermore, it then makes a precise recommendation for when to trigger an LC event and send out a Beat-the-Peak notice. This ability to make intelligent recommendations sets NuGene apart from other AI platforms,” said Ankur Garg, chief AI officer at Sonasoft.
DEC is an electric cooperative, and purchases and produces their electricity from a variety of sources. The price fluctuates due to a number of criteria. The company attempts to control this using Load Control (LC) events and Beat-the-Peak notices, but they want to avoid and mitigate the risk of what they call Coincident Peak (CP) events, which can be costly.
The artificial intelligence model makes sure DEC never endures a CP event while minimizing the amount of LC events. The model enjoyed its first success the day it went live by offering a local control recommendation to DEC to reduce the energy usage to reduce the load.
“We are pleased that the Sonasoft AI model is already showing its worth. This new approach to predicting peak demand will allow us to cut costs and provide more affordable electricity to our members. I am a long-standing advocate of using AI to improve efficiency in electricity supply and distribution. This is clearly the future, and I am proud that DEC is leading the field here,” said Bill Andrew, president and CEO of DEC.
The AI model was designed to be user-friendly and intuitive. DEC offers a dashboard with charts and graphs displaying both monthly and daily comparisons of actual versus predicted load, and an hourly 14-day load forecast. All of this data combines to help make precise recommendations on required corrective action when circumstances arise.