Asset Management

Asset-level generation forecasts? 15-day solar predictions? AI has made it a reality

Asset-level generation forecasts? 15-day solar predictions? AI has made it a reality
(Photo by American Public Power Association on Unsplash)

Amperon, a provider of AI-powered electricity forecasting and analytics services, has announced the launch of its asset-level renewable generation forecasts.

Renewable energy producers, utilities, and other companies that buy and sell electricity can now view solar generation forecasts individually or in aggregate for portfolio management. Each hour’s forecast accuracy is scored and calculated based on the asset’s nameplate capacity.

“As we look to renewables to help us decarbonize our grid, solar energy producers need better, more accurate forecasts that give them a very clear picture of how their generating assets are likely to perform in the days and weeks ahead,” said Amperon CEO and cofounder Sean Kelly. “Without accurate forecasts at the solar-farm level, companies risk having to pay high real-time power prices to cover the difference between their scheduled energy and the actual energy produced. We want to give renewable energy companies more visibility into their future generation, so they can maximize the value of the power they generate, while saving on costs.”

Amperon’s Asset-Level Forecasts provide 15-day solar forecasts, updated hourly, as well as two-day, sub-hourly solar forecasts at 5-minute intervals. The company contends that more accurate solar forecasts reduce the risk of financial losses from over-scheduling or under-scheduling in the real-time market, which occurs when actual generation differs from the generation that was previously forecasted. 

Customers can also track avoided emissions by getting insights into their carbon footprint three days in advance, and calculating how much carbon was offset due to their renewable generation.

“Knowing how much renewable energy is likely to be available to meet demand is critical for ensuring a safe, reliable, and affordable grid,” said Elliott Chorn, EVP of product at Amperon. “If renewables, like wind or solar, produce less than expected, then the grid has to rely on more expensive, highly polluting natural gas-fired “peaker” plants to meet demand. More accurate wind and solar forecasts keep energy prices low, while keeping the grid green.”

Traditional forecasting metrics like Mean Absolute Percentage Error (MAPE), normalized Mean Absolute Error (nMAE), and Root Mean Square Error (RMSE) can fall short when applied to renewable generation, Amperon said, particularly for single-site solar projects. The company argues conventional metrics fail to consider the system’s capacity or the maximum power the system can produce, leading to less accurate assessments. To address this issue, Amperon has introduced a new metric specifically designed for renewable forecasting: capacity normalized mean absolute error (cnMAE).

The cnMAE metric scales the forecast error according to the system’s capacity, offering a relative error measure that aims to ensure that the system’s scale and the observed values during the estimated period do not skew the error score. By calculating cnMAE as MAE divided by capacity, this new approach incorporates system capacity into error measurement, Amperson said.