EGH404 Research in Engineering: Forecasting Solar Power Generation

This report addresses the increasing global energy demand and the importance of renewable energy sources, particularly solar power. It highlights the challenges of solar power generation, mainly its variability due to weather conditions. The report focuses on predicting the impact of weather on solar power generation and proposes developing an accurate forecasting model. This model should incorporate various weather factors like solar radiation intensity, air temperature, wind speed, and relative humidity, using a hybrid approach combining statistical techniques, artificial intelligence, and satellite-based methods. Techniques mentioned include multiple linear regression, autoregressive moving averages, sky-imaging systems, support vector machines, and neural networks. The model should predict cloud movement and sunlight occurrence to optimize solar power usage, directing power to batteries or the grid as needed, and should be developed using state-of-the-art technologies to minimize error.

Running head: HOW CAN SOLAR POWER GENERATION BE FORECAST?
HOW CAN SOLAR POWER GENERATION BE FORECAST?
Name
Course
Professor
University
City/state
Date

HOW CAN SOLAR POWER GENERATION BE FORECAST? 2
How Can Solar Power Generation Be Forecast?
Global energy demand is increasing rapidly and so are greenhouse gas emissions since the
largest percentage of global energy is non-renewable.
This has also contributed to the problem of global warming and climate change. Thus the need to
increase generation of renewable energy is inevitable. Currently, renewable energy accounts for
33.3% of the global energy capacity, after increasing by 171 GW in 2018.
Solar is among the top and most promising sources of renewable energy. Production of solar
energy is increasing globally every year. Benefits of solar energy include: promotes
environmental protection, reduces energy bills, applicable everywhere, its production coincides
with the timeframe of peak demand, creates job opportunities, increases fuel independence, and
enhances power grid security.
However, one of the major challenges of solar power generation is variability and seasonality.
Solar power is only generated when there is sunlight. This means that solar power generation is
almost zero or decreases significantly at night or during winter, cloudy and rainy periods.
The main focus of this presentation is on predicting the effect of weather or atmospheric
conditions on the solar power generation. The output of solar panels mainly depends on the
quantity of sunshine it receives. This means that solar power generation is higher in deserts than
in areas that are normally rainy and cloudy, if all other factors are held constant.
One way of improving solar power use is accurate prediction of solar power generation at a
particular time. Solar power forecasting is essential for efficient utilization of the solar power
system, solar power trading and management of power grid. The most effective way of
forecasting solar power generation is by developing a solar power forecasting model. However,

HOW CAN SOLAR POWER GENERATION BE FORECAST? 3
most of the models available do not include every factor of weather conditions that affect solar
power generation.
Majority of these models factor in only solar radiation intensity and excludes other essential
factors such as air temperature, wind speed and relative humidity, which affect photovoltaic
power production.
Therefore, it is very important to develop a model that can accurately predict solar power
generation at any given time or season by factoring in all relevant weather conditions.
The model should be developed using a hybrid solar forecasting method that combines statistical
techniques, artificial intelligence approach and satellite based or physical techniques. Some of
these techniques are: statistical method like multiple linear regression model (MLRM), linear
regression model, autoregressive moving averages (ARMA), non-linear regression analysis, sky-
imaging system, and machine learning techniques like support vector machines and neural
networks.
The model developed should have the capacity to predict movement of clouds and occurrence of
sunlight. This will help in determining when the solar power should be used or when it should be
directed to batteries or the grid.
The model should be developed using state-of-the-art technologies so as to minimize margin
error.