This idea involves using advanced weather forecasting and solar modeling techniques to predict the amount of solar irradiance that will be available at a specific location over a given time period. By accurately predicting solar irradiance, it becomes possible to optimize the use of solar energy systems, such as solar panels and solar concentrators, and to plan the development of new infrastructure that relies on solar power. These forecasts are valuable for improving the efficiency and reliability of solar energy production and integration into the electrical grid.
During our engagement with the OpenEO platform, we encountered a technical issue that required resolution. A bug, which had not been immediately solvable by Cassini's technical support team, came to our attention . They asked us to report the bug about getting data for specific period of time. During the debug stage we found that no matter what period user specify to request, it always gives full collection as a result.
In our efforts to work around the bug and continue our solar irradiance analysis, we opted to leverage data from Solcast . We obtained solar irradiance values spanning the period from 2016 to 2023. These data provided us with the information we needed to make predictions related to solar energy production and utilization.
 - https://solcast.com
In our innovative challenge, we tackle the intricate world of solar irradiance prediction to harness the power of the sun effectively. This challenge is divided into four key tasks, each designed to unlock new horizons in sustainable energy usage and infrastructure development.
1. Solar Irradiance Predictions: Forecasting Energy Peaks and Valleys
One of the cornerstones of this challenge is accurate solar irradiance prediction. To build a long-term forecast of the energy generated by solar panels, we must first determine the power and energy of the light that comes from the Sun.
The power of sunlight upon reaching the Earth's atmosphere changes very little over time, so most of the absorption occurs in the atmosphere. We calculated absorption in a dry atmosphere, and then took into account parameters such as cloudiness and aerosols in the atmosphere, as well as the influence of external factors on the efficiency of the panel.
To calculate the light power, we used the well-known formulas (provided in presentation), which assumes that the light power upon entering the atmosphere is equal to the solar constant and is exponentially absorbed according to Beer–Lambert law in the atmosphere with known empirical coefficients. The distance that light will travel through the atmosphere is inversely proportional to the cosine of the zenith angle (theta), which can be obtained from the declination of the Sun (delta), latitude (phi) and hour angle (HRA). The deviation of the Sun, in turn, is related to the angle of inclination of the Earth and must be adjusted for the fact that the calendar year does not correspond to the astronomic one (the winter solstice is not January 1).
Next, we determined the hour angle by taking into account that the sun moves 15 degrees in an hour. Also, we made the correction with the deviation of real astronomical time from the time in a given time zone and an empirical correction known as the equation of time (EoT) for the fluctuation of the Earth’s eccentricity.
Thus, we obtained the value of light power anywhere on the planet at any time. By changing the step to 1 hour, we integrated the power for each day and obtained a graph of the total light energy for each day anywhere on Earth. By summing these values over a week, month or year and taking into account atmospheric phenomena such as cloudiness and aerosol absorption, as well as the efficiency of the panel and its dependence on temperature, we know the average expected energy output for these periods of time at any point of planet.
Our team developed LSTM model that can forecast the amount of solar irradiance at specific locations (specified by coordinates) and time intervals (up to 1 month). This task is crucial for anticipating energy peaks and valleys, helping energy providers, businesses, and individuals optimize energy consumption by aligning it with sunlight availability.
2. Energy Gain Analysis: Solar Panel Efficiency Calculation
The second task focuses on evaluating the efficiency of solar panels. By analyzing and calculating the energy gains from solar panels under varying solar irradiance conditions, we optimize solar energy usage. Improved solar panel efficiency directly translates to increased energy yield, reduced costs, and a more sustainable future.
We developed user-friendly product, which allows everyone to see effectiveness of solar panels in specific region for the whole year.
Try it yourself: https://test.synergist.kiev.ua/cassini/#
3. Benefits for Existing Businesses: Profiting from Accurate Energy Usage Planning
This task explores how businesses can benefit from accurate energy usage planning. By providing businesses with actionable insights based on solar irradiance predictions, we improve energy consumption patterns. This, in turn, can lead to reduced operational costs and a positive impact on the environment.
4. Creation of New Factories: Identification of Regions for Infrastructure Development
The final task looks beyond the horizon to identify regions suitable for the development of new infrastructure, such as solar farms or factories. In future we can combine solar irradiance forecasts with geographical and economic data to assist in the selection of strategic locations for sustainable energy projects. This not only enhances energy generation but also stimulates economic growth and job creation in the chosen regions.
This multifaceted challenge integrates AI technology, sustainability, and economic development, which shapes the future of energy usage and infrastructure. With solar irradiance predictions as the foundation, we advance the cause of renewable energy and build a more sustainable and prosperous world.
Anastasiia Lukianenko – Team lead. Master's degree in Data Science. Quantum Software engineer with Neural Network experience.
Andrii Solovienko – Bachelor in Math (statistics). Full Stack developer with UI/UX and data visualization experience.
Vadym Lozovski – Master's degree in Photonics. Quantum optic engineer with Neural Network experience.
Konstantin Kuzmichev - ETL/DWH/Database Engineer. Master's in CAD/CAM/CAE.
Oleh Ivashtenko – PhD student in Photovoltaics.
Yevhen Tatarynov – Self entrepreneur, PhD in Math (cybernetic), 15+ commercial experience in Software Development and Database Design.
Our diverse and skilled team is well-suited for the challenge for several reasons: