Machine Learning for Peak Energy Prediction

Shutting down too often?

Confidently and accurately

predict any peak energy hour in seconds.

Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.

Catching the top 5 peaks is difficult.

On average, this is the reality you face today

20 to 30
times operations curtailed during an average summer peak season
Lost in productivity per hour of curtailment
assuming 200MW under management
Hours of needless curtailment caused by inaccurate forecasts

Capture the top 5 peaks with max 10 calls

Traditional forecasting methods are reasonably accurate on normal days without a lot of demand response. On days where other market actors start curtailing, these forecasts stop being useful.
This is because it is extremely difficult, even for a team of professionals, to navigate all the possible variables and churn through the data that affects the real-time energy market. Scaling to an ever-changing, large set of data is a massive undertaking and one that is unique to the goals you face.

Terrene can capture the top 5 peaks with maximum 10 calls made using machine learning.

Making Predictions Is Simple with Terrene


Connect any data.

Load data from CSV, Excel sheets, SQL database, web scraping, and sensors.
Terrene automatically loads weather forecasts, IESO forecasts, IESO real-time demand data, IESO historical demand data and merges them together.

This data is then used to create a machine learning model to forecast the peak days and hours.

You can use the Terrene provided dataset or enrich it with additional proprietary data.

Predict the peak days and hours.

Terrene can accurately call 3 and 2 hour curtailment windows
Traditional energy forecasters, call 20-30 curtailment days during the summer peak season. They also usually don’t have the accuracy to call 2 or 3-hour windows for curtailment.

Terrene can get the top peaks with significantly fewer calls made. Terrene can also pinpoint the peaks in 2 or 3-hour windows. 

Integrate Terrene into your workflow.

Terrene's API allows you to completely white-label it.
You can integrate predictions generated by Terrene into your current Peak Energy offerings. Terrene's can generate the predictions at your configured times (i.e. every 30 mins) and upload them to your API endpoint.

You can also use Terrene's API to pull data directly from Terrene.

Find Your Insights

Get started with Terrene now
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.

How Machine Learning Works.

A quick primer on machine learning

Humanity has produced more data in the past two years than has existed in our entire history. If that’s not mind-blowing enough less than 0.5% of this data is ever analyzed. 

With all of this information, people like you and our team here at Terrene are realizing we can answer some pretty significant questions and solve serious problems. But scaling, or handling, this continuous growth of data, and finding the needles in the haystack of data is practically impossible. This is where Machine Learning comes into play. 
So the question we’re left with is how does Machine Learning actually work? 
To gain a concrete understanding of how Machine Learning works we’ll want to look back at traditional programming, or what computers have done for the majority of their existence. 

In traditional programming, we give the computer very specific instructions that it is built to understand. Then we give it some input or information, and it is programmed to reply a specific way depending on the information we give it. 

Unlike traditional programming where the computer is given specific instructions to complete a task, Machine Learning is able to recognize patterns on its own and then use that knowledge to make its own predictions.
Machine learning is a data analytics technique. The more data the machines have available the more they’re able to learn and the more accurate they are with discoveries. When we feed those machines data they’re able to churn through datasets faster than people or traditional computer programs ever could. Because they’re able to process so much information at once they’re also able to discover patterns and big pictures we just don’t have the ability or time to catch.
Machine learning uses algorithms (mathematical rules) that specialize in finding patterns in data. Where Machine Learning is best suited is when these patterns are complex and hard for humans to find. This happens most often in large and complex datasets. As it analyses all of that data and as new data comes in it continuously adapts its programming to improve accuracy.
Why Machine Learning is so important?
Now that there is so much data being produced and technology has improved data storage and computational power, Machine Learning can now produce highly accurate models and help businesses make more informed decisions quicker.

Terrene offers world-class Machine Learning modules that use your data to find the answers you're looking for.