A new solar power forecasting model devised by the CDU reports a lower error rate, say the researchers
The university developed this model using the traditional wisdom of the First Nations about seasons combined with AI
FNS-Metrics integrates First Nations seasonal calendars with ecological cues to enhance prediction precision
Conv-Ensemble framework merges Conv1D, transformers, and LSTM to learn complex weather-solar relationships
Researchers from Australia’s Charles Darwin University (CDU) report an increase of 14.6% in solar generation forecast using a new artificial intelligence (AI) model, combined with traditional knowledge of seasons from the local First Nations community.
This ‘world-first’ study by the university proposes a new solar power forecasting method called FNS-Metrics, which uses seasonal knowledge from First Nations calendars. It also presents a Conv-Ensemble model that combines different deep learning tools – Conv1D, transformers, and LSTM networks – to better learn patterns in the data.
The team developed the model using the Tiwi, Gulumoerrgin (Larrakia), Kunwinjku and Ngurrungurrudjba First Nations calendars, and a modern calendar known as Red Centre. Their seasonal insights are based on local ecological cues, such as plant and animal behaviors that are, in turn, closely tied to the changes in sunlight and weather patterns.
To test their results, researchers created a new dataset using solar and weather data from Alice Springs, Australia, along with First Nations seasonal insights. They found that this new model predicts solar power more accurately than older methods, with a 14.6% improvement in accuracy and 26.2% reduction in error.
According to the results, the error rate is less than half of the error rate seen with popular forecasting models used in the industry at present.
“Incorporating First Nations seasonal knowledge into solar power generation predictions can significantly enhance accuracy by aligning forecasts with natural cycles that have been observed and understood for thousands of years,” explained Co-author, CDU PhD student and Bundjalang man Luke Hamlin.
“By integrating this knowledge, predictions can be tailored to reflect more granular shifts in environmental conditions, leading to more precise and culturally informed forecasting for specific regions across Australia,” added Hamlin.
Combining advanced AI and ancient First Nations wisdom could revolutionize prediction technology, argue the researchers.
The complete study, titled Conv-Ensemble for Solar Power Prediction with First Nations Seasonal Information, was published in the IEEE Open Journal of the Computer Society.