11d • 🔬 Research Decoded
Research paper: Machine Learning for Solar Energy Economy
Few professionals in energy also read research publications. There are over 200 such analysis in sections 6.1 and 6.4 in the Classroom.
Below is a popular paper connecting solar PV economics, machine learning and energy storage. Below there is also a brief summary. Scroll down to download it.
- Title: Optimizing distributed solar energy economics: A machine learning classification approach to storage system management
- Citation: Wenhui, L., Li, H., Laghari, M. A., Guliyeva, S., Amonova, N., & Yuan, H. (2025). Optimizing distributed solar energy economics: A machine learning classification approach to storage system management. Case Studies in Thermal Engineering, 75, 107106. https://doi.org/10.1016/j.csite.2025.107106
- Downloadable resource is attached below
Key Points:
Solar energy is a clean resource, but it faces a major challenge because the sun does not shine consistently. This makes it difficult to match the amount of electricity generated with the amount people need to use. Batteries are used to store extra energy for later, but they are expensive and can wear out if used incorrectly. Traditional methods for managing this storage often fail to look at both the technical needs and the financial costs at the same time.
The researchers in this study developed a new method to solve these problems using a type of artificial intelligence called K-Nearest Neighbors or KNN. This is a classification tool that looks at past examples to decide what to do in the current situation. Instead of only focusing on the technical side of things, this new system includes economic factors. It helps the system decide the best times to charge or discharge the battery to save the most money.
The study uses a concept called entransy dissipation theory alongside the computer model. This is a physics principle that measures the loss of energy quality during transfer processes. By using this theory, the system ensures that energy is not just stored, but is kept at a high quality. This approach balances the need to save energy quantity with the need to maintain its usefulness, which helps in reducing the overall costs of running the system.
To prove that their method works, the team collected real-world data from 12 different distributed solar energy sites in China. They monitored these sites for 18 months. They looked at various factors such as how bright the sun was, the temperature, the battery levels, and the changing price of electricity. This large amount of information allowed them to train their computer model to be very accurate.
The results of the experiment showed that the new KNN method was very successful financially. It reduced the operational costs of the solar systems by nearly 39 percent compared to older methods. The system was so efficient that the money invested in it could be paid back in about 16 months. This shows that using smart software can make renewable energy projects much more profitable for investors.
Beyond saving money, the system also worked better technically. It improved the accuracy of predicting how much solar energy would be produced by over 23 percent. It also increased the efficiency of the battery storage system by roughly 89 percent. This means the batteries were used more effectively, wasting less energy and potentially lasting longer than they would with standard management strategies.
The researchers also calculated how much electricity the computer system itself used to run these complex calculations. They found that the energy consumed by the software was very small compared to the massive savings it generated. For every small amount of power the computer used, it saved a large amount of energy for the whole system, making it a very smart trade-off.
The paper concludes that this technology offers a great opportunity for emerging markets. By using these smart systems, countries can adopt renewable energy faster because the risks are lower and the profits are higher. The authors suggest that policymakers should set standards and encourage the use of such intelligent management systems to help stabilize the power grid and reduce environmental impact.
Real-World implications:
This research is directly applicable to companies and homeowners who use solar panels with battery storage. It demonstrates that the software controlling the battery is just as critical as the battery itself. By adopting this type of machine learning, users can significantly lower their monthly electricity bills and recover their equipment costs much faster. It also helps utility companies by making the power fed into the grid from these solar sites more predictable and stable, which is essential for maintaining a reliable electricity network.
22
3 comments
Research paper: Machine Learning for Solar Energy Economy
powered by
skool.com/software-school-for-energy-7177
A beginner-friendly program for a career in technical roles in the energy sector. Regardless of your location, age, sex, education or experience.
Suggested communities
Powered by