Volume 4, issue 6, 2025
203
ENERGY ACCUMULATION FROM SOLAR PANELS AND FORECASTING MODELS
FOR ENERGY PRODUCTION USING ARTIFICIAL INTELLIGENCE
O.Z. Toirov¹, E.T. Juraev², D.O. Hojiev³
¹ ³Tashkent State Technical University, Tashkent, Uzbekistan
²National Research Institute of Renewable Energy Sources under the Ministry of Energy,
Tashkent, Uzbekistan
Abstract:
This paper explores the process of energy accumulation from solar panels and the
possibilities of analyzing energy production through forecasting models using artificial
intelligence. The efficiency of photovoltaic systems is influenced by various external factors,
such as solar radiation, temperature, dust levels, and seasonal variations. Considering these
factors, the feasibility of predicting energy output using artificial neural networks, regression
models, and machine learning algorithms is examined. Furthermore, the efficiency of modern
energy storage technologies like lithium-ion batteries and supercapacitors is discussed. Research
shows that implementing AI-based forecast models helps to properly manage energy reserves
and reduce grid load.
Keywords:
Solar panels, energy accumulation, artificial intelligence, forecast models, energy
efficiency
Introduction
In recent years, the demand for solar energy has significantly increased. Global warming, the
need to reduce carbon emissions, and international commitments to implement renewable energy
sources have considerably enhanced the role of solar panels. Photovoltaic (PV) panels directly
convert solar radiation into electrical energy. However, these systems are intermittent—meaning
they only generate energy during daylight.
Therefore, it is essential to store the generated energy for later use, especially during the night or
on cloudy days. At the same time, predicting energy production, i.e., forecasting the amount of
energy based on future solar radiation, plays a crucial role in balancing the power grid load.
Modern approaches rely on artificial intelligence (AI) technologies, offering advanced methods
for managing solar energy production and storage. This paper is dedicated to analyzing these
technologies.
Main Content
1. Operation Principle of Solar Panels
Solar panels consist of photovoltaic cells that generate electric current through the photoelectric
effect when exposed to sunlight. The amount of energy produced depends on solar irradiance,
ambient temperature, panel orientation, dust accumulation, and other external factors.
Main components of a solar panel system include:
Photovoltaic cells (usually silicon-based)
Glass protective layer
Inverter (converts DC to AC)
Monitoring system
Due to the intermittency of solar power, energy storage becomes a necessity.
2. Energy Accumulation Systems
Energy accumulation refers to storing produced energy for later use. The most widely used
storage technologies for solar energy include:
Volume 4, issue 6, 2025
204
Lithium-ion
batteries
Known for their high energy density, long service life, and rechargeability. Widely used in small
to medium-scale solar systems. However, they are temperature-sensitive and expensive.
Supercapacitors
Capable of storing and discharging large amounts of energy quickly. Suitable for voltage
stabilization and frequency control but not long-term storage.
Pumped
hydro
storage
systems
Used in large-scale power plants. Excess energy pumps water to a higher elevation, which is
later released to generate electricity through turbines.
If not properly managed, energy storage systems can result in losses or failure. Hence, predictive
models are critical.
3. Forecasting Solar Energy Production Using AI
AI, particularly machine learning methods, is widely used to forecast solar energy production.
These models rely on data such as:
Historical solar irradiance
Temperature and humidity
Cloud cover
Panel tilt angle
Panel performance metrics
Commonly used algorithms include:
Artificial Neural Networks (ANN) – Predict energy generation based on radiation and
other variables
Random Forest – Tree-based ensemble algorithm evaluating multiple variables
Support Vector Machine (SVM) – A high-accuracy classification model
Long Short-Term Memory (LSTM) – Neural networks suitable for time-series data
Studies (Yang et al., 2021) show that ANN models can forecast daily solar energy production
with over 90% accuracy.
4. Model Outputs and Practical Applications
AI-based forecast models enable:
Prevention of battery overcharging
Timely connection to the energy grid
Accurate estimation of stored energy
Efficient system control with minimal energy loss
For example, SmartGrid-integrated solar energy systems with AI support enable real-time
monitoring and optimization. As a result, overall energy efficiency improves by 15–20%.
Conclusion
Efficient storage and utilization of energy from solar panels are critical for the stable operation of
energy systems. With energy storage technologies, solar power can be used even during cloudy
days or nighttime. However, effective operation of these systems requires accurate forecasting of
energy production.
AI-based forecast models allow real-time monitoring and control of production, storage, and grid
load, enhancing both energy efficiency and system reliability.
Future research should focus on developing new AI models, integrating with IoT devices, and
evaluating economic efficiency.
References:
1.
Yang, L., Zhao, X., & Liu, C. (2021). Solar Power Forecasting Using Artificial Neural
Networks with Weather Data. Energy Reports, 7, 3147–3155.
Volume 4, issue 6, 2025
205
2.
Chen, J., Wang, H., & Li, Z. (2020). Energy Storage Systems for Solar Applications: A
Review. Renewable and Sustainable Energy Reviews, 134, 110275.
3.
Kumar, R., & Singh, M. (2019). AI-based Solar Energy Management in Smart Grids.
Journal of Clean Energy Technologies, 7(4), 204–210.
4.
IEA (2022). Global Photovoltaic Forecasting Report. International Energy Agency.
5.
Zhang, Y., & Lee, D. (2022). Smart Energy Systems with Solar Panels and AI-based
Control. IEEE Access, 10, 33621–33630.
