Today, Alibaba DAMO Academy held a decision-making intelligence product launch event in Beijing, officially introducing the Bajie Meteorological Model. This model, built upon global meteorological models, integrates regional multi-source data to achieve spatial and temporal accuracy up to 1 kilometer × 1 kilometer and 1 hour.
This innovative meteorological forecasting tool significantly enhances the prediction performance of critical meteorological indicators such as temperature, irradiance, and wind speed. It has successfully been implemented in new energy-intensive power systems, notably improving the accuracy rates of new energy power generation and electricity load forecasts, reaching over 96% and 98%, respectively.
The Decision Intelligence Lab at DAMO Academy, leveraging years of technical expertise, has developed a regional high-precision weather forecasting model based on their proprietary global meteorological model. By integrating local station data, meteorological observations, radar images, satellite imagery, and open-source terrain data, the model enhances the granularity and accuracy of its forecasts, capable of delivering hourly updates with a 1-kilometer grid.
The Bajie Meteorological Model employs pre-training and twin MAE masked autoencoder structures to provide better initialization parameters, enabling it to learn robust feature representations hidden within highly fluctuating weather data. As the installation and grid connection of new energy sources continue to increase, the importance of precise meteorological forecasts in the power industry becomes increasingly evident. Weather conditions directly impact the output of solar and wind power, as well as residential electricity demand.
Operational data shows that the Bajie Meteorological Model's prediction accuracy has improved by 40%, 27%, 24%, and 11.8% in regional irradiance, wind speed, cloud cover, and temperature, respectively, compared to mainstream weather forecasts. Additionally, the Bajie Meteorological Model aims to continuously enhance performance in critical meteorological indicators such as cloud cover and precipitation in the future, striving to provide decision support for more scenarios including aviation alerts, agricultural production, and sporting events.
Key Points:
🌤️ The Bajie Meteorological Model introduced by Alibaba DAMO Academy achieves high-precision meteorological predictions at 1 kilometer × 1 kilometer and 1 hour.
⚡ The model significantly improves the accuracy rates of new energy power generation and electricity load forecasts, reaching over 96% and 98%, respectively.
📈 The Bajie Meteorological Model's prediction accuracy has seen significant improvements across multiple domains, providing crucial support for power systems and other industries.