CHAPTER 02

LITERATURE REVIEW

The optimization of solar panel systems has emerged as a critical area of

research in response to the increasing global demand for efficient and
sustainable energy solutions. Numerous studies have been conducted to
address the challenges associated with maximizing the efficiency,
reliability, and economic viability of photovoltaic (PV) systems. This
literature review synthesizes insights from diverse research works to provide
a comprehensive understanding of the advancements and challenges in solar
panel optimization.

Saxena et al.[10] leveraged intelligent algorithms like machine learning to

address real-time irradiance fluctuations and shading in solar systems. They
had continuous monitored solar panel performance under changing
conditions and utilizing AI to dynamically adjust the systems for optimized
energy production. Data was collected on various environmental factors, and
algorithms were trained to predict the most effective responses to these
changes. Singh and colleagues [11] used Machine Learning (ML)
techniques, specifically ANN and hybrid AI, to optimize the performance of
solar energy systems. They applied these models to forecast solar power
generation and improve the performance of solar devices through the
optimization of thermo- physical properties of nanofluids. They
demonstrated how AI models could be utilized for advanced solar
forecasting and performance enhancements. Lin et al. [12] studied a
comprehensive review of ground-based sky image-based intra-hour solar
forecasting (GSI-IHSF) introducing a generic framework consisting of
acquisition, preprocessing, cloud forecasting, and solar forecasting modules.
They highlighted the challenges and techniques used in each module and
summarized the current research trends. The paper also discusses potential
advancements and future challenges for improving forecast accuracy in GSI-
IHSF.Juncklaus et al. [13] They highlighted a shift from classical statistical
methods to Machine Learning approaches like Artificial Neural Networks
and Convolutional Neural Networks. The root mean squared error was
identified as the most common validation metric across forecasting
models.Erdener et al. [14] They explore the challenges of limited visibility,
variability, and uncertainty in grid operations due to increasing behind-the-
meter solar penetration. The paper concludes with recommendations for
improving forecasting practices and future implementations to enhance
system-wide performance and utility operations. Yang et al. [15] The study
emphasizes the importance of collaboration between atmospheric science
and power system engineering to enhance solar irradiance and power
forecasting for high solar penetration. The paper concludes that joint efforts
are crucial for improving forecasting accuracy and ensuring reliable power
system operations with high solar power penetration.
Alkhayat et al [16] reviewed recent advancements in deep learning-based
solar and wind energy forecasting, emphasizing data preprocessing,
evaluation methods, and model comparisons. Hybrid models, Recurrent
Neural Networks, and Convolutional Neural Networks dominate, with
probabilistic and multistep forecasting gaining traction. Sharma et al. [17]
reviews short-term intra-hour solar irradiance nowcasting methods using sky
images, emphasizing their importance for reliable power supply.They
examine various forecasting
approaches, including satellite, machine learning, and numerical weather
prediction-based methods. Key sky image features significant for
nowcasting are identified and discussed. Chu et al [18] examined
mathematical tools like regression, deep learning, and genetic algorithms,
alongside data-driven, local-sensing, and hybrid forecasting methods. Future
development suggestions emphasize probabilistic and spatial forecasting
innovations.Mingxuan et al. [19] reviewed MPPT techniquesor solar PV
systems, focusing on methods used under Partial Shading Conditions
(PSCs). They evaluated the use of intelligent control strategies, such as PSO
and MSSA (Memetic Salp Swarm Algorithm), to improve maximum power
point tracking. Their work also compared traditional methods to new AI-
enhanced approaches, offering insights into enhanced global search
capabilities and optimized power extraction. Alfaverh et al.
[20] combined Reinforcement Learning and Fuzzy Logic to control
residential demand response. By analyzing energy consumption patterns,
their model adjusted energy use in real-time to reduce peak-hour demand,
thereby improving overall energy efficiency and reducing costs. Gupta et al.
[21] developed a distributed computing system using Hadoop for smart grid
data management. The system was designed to handle vast amounts of data
generated by smart meters and other grid components. AI algorithms were
used to process this data, leading to a 20x improvement in processing speed,
allowing for more efficient energy management and faster response times in
real-time grid operations. Kumar et al.
[22] review of the different resources and methods used for forecasting solar
irradiance in different time horizons and also give an extensive review of the
sensor networks that are used for determining solar irradiance. The various
error metrics and accessible data sets available for the sensor networks are
also discussed that can be used for validation purposes
Hong et al. [23] discussed importance of reproducible research and points
out six valuable open data sources; makes recommendations about
publishing high-quality research papers; and offers an outlook into the future
of energy forecasting. Li et al. [24] gives a comprehensive review on how
probabilistic solar forecasts are utilized in power systems to address the
challenges. Potential methods to deal with uncertainties in power systems
are summarized, such as probabilistic load flow models, stochastic
optimization, robust optimization, and chance constraints. Ahmed et al [25]
observed that integrating advanced forecasting techniques improves
photovoltaic power grid integration by addressing instability and
inefficiencies. Optimization methods like genetic algorithms and Bootstrap
techniques enhance accuracy and reliability. These approaches provide
actionable insights for stable, efficient, and economically viable solar power
forecasting. Wang et al. [26] categorized AI-based solar power forecasting
models, highlighting their ability to uncover nonlinear features for improved
prediction accuracy. They addressed challenges and suggest future research
directions to enhance reliability and efficiency. Wan and colleagues [27]
developed the Memetic Salp Swarm
Algorithm (MSSA), which combines Salp Swarm Algorithm (SSA) with
Particle Swarm Optimization (PSO) for improved global search capabilities
in Maximum Power Point Tracking (MPPT) for solar PV systems. This
hybrid approach allowed for better convergence and faster tracking of the
optimal power point, particularly under partial shading conditions where
traditional methods often struggle. Bahiraei et al.
[28] utilized ML techniques, such as ANN and fuzzy logic to predict and
optimize the properties of nanofluids used in solar thermal systems. They
applied these techniques to enhance performance of solar collectors by
improving the thermal conductivity and viscosity of nanofluids, which
ultimately helped to optimize energy efficiency. Eseye et al. [29] forecasted
power generation of a PV power system every hour by using Hybrid Model
(Wavelet Transform + Particle Swarm Optimization + Support Vector
Machine) by considering solar radiation intensity, temperature, cloud cover,
humidity, pressure, wind speed
Nikolovski et al. [30] predicted the power output of the PV system (15 min
intervals) and curtailed the load to match the PV generation by using
Adaptive Neuro-Fuzzy Inference System (ANFIS) and Fuzzy Logic. They
considered wind speed, wind power, and temperature for output prediction.
Pahari and Subudhi [31] proposed an Integral Sliding Mode Control (SMC)
method for steady-state performance enhancement under rapidly changing
solar irradiance. This method allowed for improved Maximum Power Point
Tracking (MPPT) performance and optimized energy extraction, especially
in fluctuating environmental conditions. Li et al. [32] combined Fuzzy Logic
Controllers (FLC) with Maximum Power Point Tracking (MPPT) to enhance
the dynamic and steady-state performance of PV systems. Their method
incorporated adaptive adjustments based on fluctuating solar irradiance,
ensuring optimized energy extraction. This work highlighted the advantages
of hybrid control methods for improving the performance of PV systems,
especially under partial shading conditions. Huang et al. [33] integrated the
Genetic Algorithm (GA) and Firefly Algorithm (FA) to enhance MPPT
control under partial shading conditions. Their Findings allowed for faster
convergence and improved maximum power point tracking, ensuring
optimized energy extraction from solar panels even in less-than-ideal
conditions.
Kofinas et al. [34] implemented a hybrid Fuzzy Q-learning model within a
multi-agent system. This system was designed to manage microgrid energy
by dynamically adjusting the load and supply balance. The model was
trained on data from the microgrid to make real-time decisions that improve
the stability and efficiency of energy distribution. Nikolovski et al. [35]
applied the ANFIS model to solar PV forecasting, with an additional layer of
Fuzzy Logic for improved load curtailment. The model was trained to
predict solar energy generation under various environmental conditions,
using historical data from multiple locations. Dou et al.
[36] used K-means clustering to refine data for solar and wind generation
forecasting. By applying this unsupervised
learning technique, they improved the quality of input data, which was
subsequently fed into neural networks to enhance prediction accuracy. The
approach allowed them to make more reliable energy generation forecasts,
critical for effective grid integration and system planning. Kurzrock and
colleagues
[37] focused on geostationary satellite-based short term cloud forecasting,
using AI techniques to predict cloud cover and its impact on solar output.
Their work demonstrated the use of satellite data for improving solar power
predictions, especially for short-term forecasting and system reliability.