Abstract
In the relentless pursuit of sustainable energy solutions, this study pioneers an innovative approach to integrating thermoelectric generators (TEGs) and photovoltaic (PV) modules within hybrid systems. Uniquely, it employs neural networks for an exhaustive analysis of a plethora of parameters, including a diverse spectrum of semiconductor materials, cooling film coefficients, TE leg dimensions, ambient temperature, wind speed, and PV emissivity. Leveraging a rich dataset, the neural network is meticulously trained, revealing intricate interdependencies among parameters and their consequential impact on power generation and the efficiencies of TEG, PV, and integrated PV-TE systems. Notably, the hybrid system witnesses a striking 23.1% augmentation in power output, escalating from 0.26 W to 0.32 W, and a 20% ascent in efficiency, from 14.68% to 17.62%. This groundbreaking research illuminates the transformative potential of integrating TEGs and PV modules and the paramountcy of multifaceted parameter optimization. Moreover, it exemplifies the deployment of machine learning as a powerful tool for enhancing hybrid energy systems. This study, thus, stands as a beacon, heralding a new chapter in sustainable energy research and propelling further innovations in hybrid system design and optimization. Through its novel approach, it contributes indispensably to the arsenal of clean energy solutions.
1. Introduction
In an era characterized by burgeoning population densities and escalating concerns over the deleterious consequences of emissions from fossil fuel-dependent systems on both human well-being and the ecological landscape, the imperative for the development and adoption of efficacious clean energy solutions is more pressing than ever [1–3]. A prominent contender within the clean energy pantheon is the solar photovoltaic (PV) system. Solar PV systems have gained preeminence as a quintessential mode of solar energy conversion, primarily attributable to the confluence of benefits they confer [4, 5].
Notwithstanding the aforementioned, the relatively modest performance of solar PV systems, especially in comparison to their fossil fuel counterparts, poses a formidable impediment to their widespread deployment across diverse climatic conditions [6–8]. Consequently, a plethora of scholarly endeavors [9–11] have been relentlessly focused on propelling the frontiers of PV efficiencies to new echelons.
Among the myriad strategies for PV efficiency enhancement, the integration of thermoelectric generators (TEGs) with PV cells emerges as a particularly propitious avenue [12–14]. By harnessing the waste heat generated during the photovoltaic process and transmuting it into supplementary electrical energy, the tandem operation of TEGs and PV cells exhibits a notable augmentation in efficiency [15–17].
However, despite the significant strides in the realm of PV-TEG synergistic research, there remains a discernible lacuna in understanding the optimal thermoelectric materials that can be harnessed to achieve peak power outputs and efficiencies from the integrated PV-TEG systems.
The ensuing sections will provide a detailed examination of the advancements that have been made in this exciting intersection of technologies, as well as identify and explore the extant knowledge gaps, thereby laying the groundwork for further research and development in this critical domain.
Building upon the groundwork of the burgeoning domain of PV-TEG integration, various scholarly efforts have sought to explore, validate, and optimize the inherent potential of these hybrid systems.
Alnajideen and Min [18] made a notable contribution through their empirical analysis. They demonstrated a marked enhancement in the performance of PV-TE systems by employing specular aluminum mirrors. The study revealed a 16.9% overall efficiency in the PV-TE system, which was a noteworthy increment compared to the 15.9% efficiency observed in conventional systems without the mirrors. Furthermore, the TEG was instrumental in contributing to a 6.3% surge in the system’s efficiency as well as elevating the system’s output power from 12.41 mW to 62.04 mW.
In a parallel vein, Bamroongkhan et al. [19] made strides in demonstrating the dual functionality of the PV-TE systems. Through an experimental study, they showcased that a PV-TE system is capable of delivering hot water at 40°C while concurrently generating 5.25 W of electrical power, given a temperature differential of 161.4°C. This was achieved alongside an efficiency of 3.02% from the thermoelectric generator.
Driven by theoretical insights, Beeri et al. [20] undertook an experimental approach to probe the combined efficiency of PV-TE systems under varying optical concentrations. Intriguingly, their findings unveiled a potential combined PV-TE efficiency of 50% under an optical concentration of 300. Moreover, the study elucidated an inverse relationship between concentration levels and PV efficiency, in contrast to the positive correlation observed with TEG efficiency.
Further extending the ambit of research, Ismaila et al. [21] conducted a thermo-economic optimization study focusing on water-cooled concentrated PV-TE technology. They demonstrated the viability of maintaining a maximum allowable PV temperature of 100°C at an optimum overall performance index of 94.7% while sustaining a levelized cost of energy at $0.0392 per kWh.
In an audacious and avant-garde development, Zhao et al. [22] heralded a novel PV-TE system equipped with the capacity to operate during nocturnal hours by leveraging radiative cooling technology. This pioneering approach taps into the cold expanse of the cosmos. It was ascertained that the TEG system, when seamlessly integrated into the PV-TE system, could generate a circuit voltage of 9 mV in the absence of sunlight, signifying the hybrid system’s capacity for power generation even during nonilluminated periods.
In summary, these empirical studies and innovative developments attest to the substantial potential of PV-TE systems and underscore the exigency for continued exploration and optimization in harnessing this hybrid technology for clean energy generation. The following sections delve into the technical nuances and prospective avenues for further advancements in the field.
While commendable strides have been made in the domain of PV-TE research, a conspicuous knowledge lacuna persists in terms of the interplay between various thermoelectric materials and the resultant impact on the overarching performance of PV-TE systems. In particular, the extant body of literature has predominantly focused on the incorporation of conventional bismuth-telluride-based TEG modules, which are inherently attuned to midtemperature ranges. This myopic focus precludes an in-depth understanding of the potential contributions from alternative semiconductors that could be operable across a broader temperature spectrum, encompassing both high and low temperature regimes.
Addressing this critical gap, the present study embarks on a pioneering exploration, adroitly venturing into uncharted territories within the PV-TE design landscape. At the core of this investigation lies an exhaustive analysis of six disparate thermoelectric materials that span high, medium, and low temperature regimes. The objective is to ascertain the material that epitomizes the zenith of power output and efficiency in PV-TE systems under a gamut of solar concentration ratio levels.
This investigation is not solely characterized by its groundbreaking evaluation of materials but is further distinguished by its innovative methodology. Leveraging the wealth of data garnered through this optimization study, the research employs a cutting-edge recurrent neural network architecture for performance optimization. This strategic employment of artificial intelligence serves as a linchpin to expedite and streamline the optimization process, a paradigm shift from conventional numerical optimization techniques which have hitherto been encumbered by protracted timeframes.
In essence, this study stands as a trailblazer, not only by broadening the horizons of material exploration but also by synergizing it with sophisticated artificial intelligence methodologies. The outcomes promise to galvanize the research community and catalyze further advancements in the design and optimization of PV-TE systems for sustainable energy generation.
2. Research Methodology
This section delineates the methodology employed in this research through a series of sequential steps: (1)System Overview. A brief introduction to the photovoltaic-thermoelectric (PV-TE) hybrid system studied, including the materials used(2)Governing Equations. Discussion of the mathematical equations governing the performance of the PV module, the thermoelectric generator (TEG), and the integrated PV-TE system(3)Recurrent Neural Network Architecture. An overview of the recurrent neural network (RNN) architecture employed for predicting optimal system performance(4)Boundary Conditions: Establishment and explanation of the boundary conditions applied to the simulation model(5)Model Verification. A comparative analysis to validate the accuracy of the numerical model by juxtaposing the results of this study with those from previous research(6)Data Generation Process. Detailed account of how the numerical dataset for machine learning was generated, including the range and values of parameters analyzed
Each of these steps is crucial to comprehensively understanding the methodology behind the research and facilitating reproducibility.
2.1. System Description
Figure 1 visually elucidates the hybrid system analyzed in this study. The PV module capitalizes on the incident solar radiation to generate electricity through the photovoltaic effect [23]. Specifically, the PV module harnesses the ultraviolet and visible segments of the solar spectrum. Conversely, the infrared segment, which is detrimental to the PV module as it elevates cell temperature and consequently impairs efficiency, is channeled to the TEG affixed to the PV module’s rear plate.
A thermally conductive paste with a thermal conductivity of 4 W/m-K [24] facilitates the secure attachment of the TEG to the PV module’s backplate. This paste not only ensures effective heat conduction but also enables the TEG to convert the excess heat into supplementary electrical energy. The TEG’s cold junction is subjected to forced convective water cooling, establishing a temperature gradient between its hot and cold junctions, which is vital for continuous power generation.
For computational efficiency while maintaining numerical accuracy, a three-dimensional segment of the thermocouple is selectively isolated and analyzed. Six thermoelectric materials, namely higher manganese silicide (HMS), lead telluride (PbTe), lithium nitride oxide (LiNiO), skutterudite (CoSb), bismuth-telluride (BiTe), and nanomaterials (nano), are employed. The temperature-dependent properties of these materials are rigorously considered to ensure the fidelity of numerical models. Figure 2 and Table 1 illustrate how these material properties evolve with temperature.
2.2. Model Equations
The temperature distribution in the PV system is obtained by solving the three-dimensional energy equations [32] where the thermal properties of the materials including the specific heat capacity (), density (), and thermal conductivity (). The temperature of the various materials is expressed as , is the power generated per unit volume, and is the energy absorption rate per unit volume.
At steady-state conditions, Equation (1) gives
The power output of the PV is evaluated as where is the temperature dependent cell efficiency, which is evaluated using [33] where is the cell reference efficiency (15%) [13], is the temperature coefficient (0.004 K-1) [34], is the cell temperature, and is the reference temperature taken as 298.15 K [35]. is obtained by solving Equation (2) using the finite element method in ANSYS software.
Similarly, the solar energy absorption rate in the th material is described by where is the optical concertation ratio, is the solar energy (measured in irradiance, W/m2), , , and are the reflectivity, absorptivity, and volume of the material, respectively, and is the exposed surface area. These properties can be obtained from the following references [36, 37].
For the TEG system, the coupled thermal-electrical equations are solved to get the temperature and electrical voltage distributions in the module as follows [38]: where and are the Seebeck coefficient and electrical conductivity, respectively. is the current density, is the electric permittivity, and is the scalar electric potential.
is expressed as [39] where is the electric field intensity, obtained from the continuity equation for a dielectric medium, expressed as [40]
Alternatively, if there is no magnetic field, the electric field becomes noncentrifugal (in mathematical terms, ), and the electric field is directly obtained from the scalar electrical potential according to
The heat absorption at the hot and cold junction of the device is given by where the subscripts and denote the - and -type TE semiconductors, respectively. and are the cross-sectional area and height of the TE semiconductors, respectively. is the temperature gradient which is the difference between the hot and cold junction temperatures of the module, respectively, obtained from the thermal-electrical solver in ANSYS software.
is the TEG current which is given as [41] where and are the internal and external resistances of the TEG module, respectively.
The power output and efficiency of the TEG are evaluated using
The hybrid system’s power and efficiency are calculated by summing the individual system’s power and efficiency as follows:
2.3. Recurrent Neural Network
Recurrent neural networks (RNNs) are a class of artificial neural networks designed for sequence prediction problems. Unlike traditional neural networks, RNNs have connections that form directed cycles. This architecture allows them to maintain a memory of previous inputs in their internal state, which makes them suitable for tasks such as time series prediction, natural language processing, and more.
In RNNs, units are connected in cycles to form a directed graph. This enables the network to exhibit dynamic temporal behavior. Unlike feedforward neural networks, RNNs can use their internal memory to process sequences of inputs, making them adept at tasks such as time-series prediction, sequence labeling, and natural language processing.
Each unit in an RNN has a recurrent connection with a delay of one-time step. This allows the network to maintain a state that can represent arbitrary time sequences, giving RNNs a form of memory.
An RNN processes sequences of inputs, with each input affecting the network’s internal state. At each time step , an RNN takes an input vector and the hidden state from the previous time step , and produces an output and a new hidden state
The hidden state is updated according to
Here, and are weight matrices for the hidden state and input vector, respectively. is the bias term, and is an activation function, which in this work is the Sigmoid function.
The output is computed as where is the weight matrix for the output layer, is the output bias, and is the activation function.
The sigmoid function used as the activation function in the RNN, is defined as
It takes a real-valued input and squashes it to a range between 0 and 1. This can be interpreted as transforming the input into a probability or a binary activation.
The sigmoid function is particularly useful in the context of RNNs because it can help in controlling the flow of information through the network, especially when we deal with gates as in long short-term memory (LSTM) or gated recurrent units (GRU) networks. However, in basic RNNs, it’s also common to use the sigmoid activation to introduce nonlinearity into the network.
One drawback of the sigmoid function is that it can cause the vanishing gradient problem, especially in deep networks or long sequences, as its gradient is small when the input is large in magnitude.
The basic components of the RNN as shown in Figure 3 are explained as following: (1)Input Layer. Similar to other neural networks, the input layer in RNNs receives the raw data. In the context of RNNs, data usually comes in sequences, and the network processes one element of the sequence at a time. In this case, the input layer has seven neurons corresponding to the (1) concentration ratio, (2) convective film coefficient, (3) TE leg height, (4) TE leg cross-sectional area, (5) ambient temperature, (6) wind speed, and (7) surface emissivity(2)Hidden Layer. The hidden layer in RNNs is responsible for maintaining memory of past inputs. At each time step, the hidden layer receives not only the current input but also the hidden state from the previous time step. This is crucial as it allows the network to keep track of dependencies in the input sequence. The hidden state at each time step, , denoted by , is evaluated according to Equation (14). In this work, we have 2 hidden layers with 14 neurons per layer(3)Output Layer. The output layer produces the prediction or classification for each time step. The output layer has six neurons corresponding to (1) TEG power, (2) TEG efficiency, (3) PV power, (4) PV efficiency, (5) PV-TE power, and (6) PV-TE efficiency. The output at time step , denoted , is computed using Equation (15)
The hyperparameters of the recurrent neural network (RNN) are integral to its performance and are detailed below. (1)Training Algorithm—Levenberg-Marquardt (TRAINLM). In this research, the Levenberg-Marquardt algorithm is employed as the training function, which is known for its efficiency in training neural networks. Specified using the TRAINLM function in MATLAB, the Levenberg-Marquardt algorithm combines aspects of the Gauss-Newton method and the gradient descent method. By doing so, it seeks to find the optimum weights that minimize the error between predicted and actual output values. This algorithm is particularly suitable for complex models and is known for its faster convergence(2)Adaptive Learning Function—Gradient Descent with Momentum (LEARNGDM). The adaptive learning function employed in this research is gradient descent with momentum, which is specified using the LEARNGDM function in MATLAB. The gradient descent with momentum algorithm enhances the standard gradient descent by considering the past gradients in updating the weights. This momentum term helps to stabilize the updates and reduce oscillations, leading to more efficient and faster convergence in the learning process(3)Transfer Function—Sigmoid Function. The transfer function used in the network is the Sigmoid function. It is a non-linear activation function that maps input values to an output between 0 and 1, making it particularly useful for models where the output is a probability. Mathematically, it is expressed using Equation (16). In the context of an RNN, it is used to capture nonlinear relationships in the data
The selection and tuning of these hyperparameters are critical, as they significantly affect the efficiency and accuracy of the RNN in modeling and predicting the system’s performance. Properly chosen hyperparameters can lead to more robust predictions and an overall better-performing neural network.
The advantages and disadvantages of the RNN architecture are (i)Advantages (1)RNNs can process sequences of variable lengths rather than fixed-size input vectors(2)They can theoretically capture patterns of arbitrary lengths in the input data due to their recurrent connections(ii)Limitations (1)Vanishing Gradient Problem. As the sequence length increases, the gradients during back-propagation can become extremely small, essentially causing the network to stop learning. This makes it difficult for basic RNNs to capture long-term dependencies in the data(2)Exploding Gradient Problem. Conversely, gradients can sometimes grow exponentially as they are back-propagated, causing weights to diverge(3)Computational Complexity. Maintaining the hidden state over time steps can be computationally intensive, especially for long sequences
The extensions and variants of the RNN are (i)Extensions and variants
Because of these limitations, various extensions of basic RNNs have been developed, including long short-term memory networks (LSTMs) and gated recurrent units (GRUs), which are more capable of capturing long-term dependencies in the data without suffering as much from the vanishing and exploding gradient problems.
2.4. Boundary Conditions
In order to carry out a meaningful simulation, it is crucial to establish appropriate boundary conditions for the virtual model. The boundary conditions defined for this simulation are as follows: (1)Solar Irradiation on PV Module. The photovoltaic (PV) module is subject to a constant heat flux, which represents the solar irradiation incident on its surface. This is modeled as a fixed heat flux of 5 kW/m2, under the assumption that the PV module is exposed to direct sunlight without any obstruction or shading(2)TEG Cold Junction Cooling. The cold junction of the thermoelectric generator (TEG) is subject to cooling in order to maintain a temperature gradient across the device, which is essential for its operation. This is modeled by applying a convective film coefficient of 500 W/m2K to the TEG’s cold junction. This represents the heat transfer to the coolant due to forced convection(3)Insulation of External Surfaces. To focus the simulation on the internal dynamics of the hybrid system, it is assumed that all external surfaces are perfectly insulated. This implies that there is no heat loss through these surfaces, and it ensures that the system’s thermal environment is solely governed by the conditions defined within the model(4)Voltage Across Thermoelectric Material. A terminal voltage is applied to the n-type thermoelectric semiconductor within the TEG. This boundary condition is necessary to model the behavior of the TEG in converting thermal energy into electrical energy
These boundary conditions are crucial for accurately modeling and understanding the performance of the PV-TEG hybrid system under the given assumptions. With these conditions in place, the simulation can be executed to study the thermal and electrical characteristics of the system.
2.5. Validation with Experimental Data
To ascertain the reliability and accuracy of the numerical model employed in this analysis, a validation process is conducted. This process entails juxtaposing the results derived from the present study with those gleaned from prior research that utilized conventional bismuth telluride- (BiTe-) based PV-TE systems.
The primary objective of this validation process is to benchmark the outcomes of the present numerical model against established data in the literature. Such comparison is instrumental in corroborating the fidelity of the model and ensuring that it yields results that are in consonance with empirically validated findings.
Figure 4 presents the outcome of the validation process. Upon meticulous examination of the data, there is a noteworthy correspondence between the results of the present study and those reported in earlier literature, with a maximum relative error of 3.5%. Such a minor discrepancy may be attributable to the nuances of real-world experimental conditions that are not necessarily accounted for in numerical simulations, including measurement inaccuracies and human factors. The congruence between the data sets not only attests to the accuracy and reliability of the numerical model but also underscores its appropriateness for the analysis undertaken in this study.
(a)
(b)
(c)
(d)
(e)
(f)
2.6. Data Generation
The dataset used for training the machine learning model is generated through an extensive series of simulations conducted on the virtual model of the photovoltaic-thermoelectric (PV-TE) system using ANSYS software. A comprehensive analysis is performed to investigate the effects of seven key parameters on the power output and efficiencies of the PV module, the thermoelectric generator (TEG), and the integrated PV-TE system. The parameters and their respective ranges and constant values are summarized in Table 2.
For each of the seven parameters, a total of 1000 simulations are executed, with the system’s performance being documented for each output parameter. The parameters and their respective ranges and constant values are as follows:
As a result of this extensive simulation process, a numerical dataset comprising 7000 samples is generated. Each sample in the dataset contains 7 input features (corresponding to the seven parameters) and 6 output features (representing various aspects of the system’s performance). This dataset serves as the foundation for training and validating the machine learning model. 70% of the dataset is used for training the neural network, 15% is used for testing the neural network’s results, and the remaining 15% is used to validate the neural network.
3. Results and Discussions
In the pursuit of optimizing PV-TE systems, the selection of appropriate thermoelectric semiconductors plays a critical role. The choice of semiconductors can significantly impact the performance and efficiency of the system, thereby affecting its overall energy generation capabilities. In this paper, we discuss the results obtained from the extensive analysis of different semiconductor materials, including HMS, PbTe, LiNiO, CoSb, BiTe, and nano. The evaluation is based on key performance indicators such as PV temperature, TEG power, and TEG change in temperature, TEG efficiency, and PV-TE power generation. These indicators provide valuable insights into the effectiveness and suitability of each semiconductor material for PV-TE systems. The subsequent discussion delves into the detailed analysis of the results, examining the performance trends of the semiconductor materials under varying sunshine intensities. The outcomes shed light on the strengths and limitations of each material and provide essential guidance for material selection and system optimization. By unraveling the intricacies of these results, we aim to contribute to the ongoing efforts in enhancing the efficiency and sustainability of PV-TE systems.
In the following sections, we present a comprehensive discussion of the results obtained from the evaluation of the thermoelectric semiconductor materials, performance optimization of the best system, finding the best semiconductor, performance of the neural network, highlighting their performance in terms of PV temperature, TEG power, TEG change in temperature, TEG efficiency, PV-TE power generation, the convective cooling film coefficient, TE leg height, TE leg cross-sectional area, the ambient temperature, the wind speed, the PV emissivity, and also the best performance metrics of the neural network. The insights gained from this analysis pave the way for further advancements in solar photovoltaic-thermoelectric systems and drive us closer to realizing a cleaner and more sustainable energy future.
3.1. Finding the Best Semiconductor
Figure 5 demonstrates how the PV temperature of each semiconductor changes with increasing sunshine intensity. It is observed that as the sunshine intensity increases, the PV temperature for all the thermoelectric semiconductors also increases. However, there are notable variations among the different materials. Starting with HMS, it can be observed from the plot that its PV temperature increases with the rising sunshine intensity, but it achieves the lowest PV temperature among all the semiconductors, reaching a value of 336.78 K. This indicates that the HMS semiconductor is not as efficient in converting solar energy into electricity and thermal energy compared to the other semiconductor materials. On the other hand, the nanosemiconductor exhibits the highest increase in PV temperature among all the semiconductors considered in the study. At a sunshine intensity of 25 kW/m2, the nanosemiconductor experiences a substantial temperature increase of 631.80 K. This suggests that the nanomaterial is more effective in converting solar energy into electricity and thermal energy, leading to a higher PV temperature compared to the other semiconductors.
The plot also further indicates that all the semiconductor materials experienced an increasing TEG change in temperature as the sunshine intensity increased. This implies that higher sunshine intensities lead to larger temperature differentials across the TEG devices, which is an important factor in generating electrical power. Among the materials studied, nanosemiconductors achieved the maximum TEG change in temperature. At a sunshine intensity of 25 kW/m2, nanosemiconductors exhibited a significant TEG change in temperature of 266.93 K. This suggests that nanosemiconductors are highly effective in harnessing and converting the temperature difference between the hot and cold sides of the TEG device, resulting in a substantial temperature gradient. However, HMS semiconductor material recorded the lowest TEG change in temperature among the studied materials. Specifically, at a sunshine intensity of 5 kW/m2, HMS demonstrated the least TEG change in temperature of 18.54 K. This indicates that HMS has a relatively lower capacity to generate significant temperature differences under the given conditions. These findings suggest that nanosemiconductors have a higher potential for efficient energy conversion in TEG devices, while HMS may have limitations in generating substantial temperature gradients.
Furthermore, the results indicate that as the sunshine intensity increases, there is a general trend of slight increases in TEG power for all the materials. However, there are notable differences in the extent of this increase among the various semiconductors. Among the materials studied, LiNiO exhibits the highest increase in TEG power. At a sunshine intensity of 25 kW/m2, LiNiO demonstrates an increase of up to 0.84 W in TEG power. This indicates that LiNiO has a relatively high efficiency in converting the temperature difference between the hot and cold sides of the thermoelectric generator into electrical power when exposed to higher sunshine intensities. Conversely, HMS shows the lowest TEG power among the materials considered in the study. Specifically, at a sunshine intensity of 5 kW/m2, HMS records the lowest TEG power of 0.0015 W. This suggests that HMS has a relatively lower efficiency in converting thermal energy into electrical power under the given conditions.
The plots also suggest that all the thermoelectric semiconductors experienced a slight increase in TEG efficiency as the sunshine intensity increased. This implies that higher levels of sunlight have a positive impact on the efficiency of the TEG devices, indicating that more energy is being effectively converted into usable electrical power. Among the semiconductors studied, HMS exhibited the lowest TEG efficiency. With a sunshine intensity of 5 kW/m2, HMS achieved a TEG efficiency of 0.19%. This indicates that the conversion of thermal energy into electrical power in the HMS material is relatively inefficient under the given conditions. Conversely, nanosemiconductors demonstrated the highest TEG efficiency. At a sunshine intensity of 25 kW/m2, nanosemiconductors reached an impressive TEG efficiency of up to 9.31%. This suggests that the nanomaterial is highly efficient in converting thermal energy into electrical power, resulting in a significantly higher TEG efficiency compared to the other semiconductors. The remaining semiconductors, BiTe, LiNiO, CoSb, and PbTe, all exhibited TEG efficiencies ranging between 0.69% and 5.53%. While they showed relatively lower efficiencies compared to nanosemiconductors, they still demonstrated an improvement in TEG efficiency with increasing sunshine intensity. The impact of these findings is significant for the field of solar photovoltaic-thermoelectric systems. Higher TEG efficiency implies better utilization of solar energy, resulting in increased power generation and improved overall system performance. The variation in TEG efficiencies among different semiconductor materials highlights the importance of material selection in optimizing the energy conversion process.
According to the figure, BiTe and nanosemiconductors exhibit parallel maximum PV-TE power generation as the sunshine intensity increases. At the peak of their performance, BiTe achieves a maximum PV-TE power of 0.28 W, while nanosemiconductor reaches a slightly higher value of 0.29 W. This suggests that these materials are relatively efficient at converting solar energy into both electrical power (PV) and thermal power (TE). In contrast, LiNiO semiconductor stands out with a significant increase in PV-TE power as the sunshine intensity rises, particularly reaching a high value of 0.82 W at 25 kW/m2. This indicates that LiNiO is highly effective in generating power from the combined photovoltaic and thermoelectric mechanisms under intense sunlight conditions. On the other hand, CoSb, HMS, and PbTe semiconductors demonstrate a slight decrease in PV-TE power as the sunshine intensity increases. Gradually, their power outputs decrease to 0.13 W (CoSb), 0.09 W (HMS), and 0.09 W (PbTe) at 25 kW/m2. These results suggest that these particular materials may experience limitations in power generation as the solar irradiance becomes more intense. The impact and effect of these findings are significant for the development and optimization of solar photovoltaic-thermoelectric systems. The identification of LiNiO as a material with high PV-TE power generation potential can guide researchers in selecting suitable semiconductor materials for efficient energy conversion. Furthermore, understanding the limitations observed in CoSb, HMS, and PbTe semiconductors provides insights into areas that require further improvement.
It also suggests that all the semiconductor materials experienced a drastic decrease in PV-TE efficiency as the sunshine intensity increased. This implies that higher levels of sunlight have a negative impact on the efficiency of the PV-TE systems, indicating a reduced ability to convert solar energy into usable electrical and thermal power. Among the materials studied, LiNiO semiconductor obtained the highest PV-TE efficiency. At a sunshine intensity of 5 kW/m2, LiNiO demonstrated a maximum PV-TE efficiency of 12.29%. This indicates that LiNiO is relatively efficient in converting solar energy into combined electrical and thermal power under moderate sunshine intensities. nanosemiconductor material exhibited the least PV-TE efficiency. At a sunshine intensity of 25 kW/m2, nanosemiconductor recorded the lowest PV-TE efficiency of 0.77%. This suggests that nanomaterial is less effective in converting solar energy into usable power, both electrically and thermally, particularly under high-intensity sunlight conditions. It is important to note that the materials experienced higher PV-TE efficiency at lower sunshine intensities. This indicates that the performance of the materials is influenced by the intensity of the sunlight they are exposed to. Lower intensities seem to provide more favourable conditions for efficient energy conversion in PV-TE systems. The impact and effect of these findings highlight the importance of optimizing solar photovoltaic-thermoelectric systems based on the specific characteristics of the semiconductor materials used.
3.2. Performance Optimization of the Best System
From Figure 6, the influence of the convective cooling film coefficient on the system’s power generation has been explored. This analysis is aimed at assessing the impact of varying convective cooling film coefficients on the TEG power (), PV power, and PV-TE power and to identify the optimal operating conditions for maximum power generation. The results indicate that the system achieved its maximum TEG power of up to 0.0097 W at a convective cooling film coefficient of 400 W/m2K. However, as the convective cooling film coefficient increased beyond this point, the TEG power gradually decreased to 0.0074 W at a convective cooling film coefficient of 3600 W/m2K. This suggests that the TEG power generation is sensitive to changes in the convective cooling conditions, with higher cooling coefficients potentially leading to reduced power output. On the other hand, the PV power exhibited a peak of 0.2658 W at a convective cooling film coefficient of 3600 W/m2K and subsequently decreased to 0.2410 W at a convective cooling film coefficient of 400 W/m2K. Similarly, the PV-TE power reached its maximum at 0.2733 W when the convective cooling film coefficient was 3600 W/m2K and its minimum at 0.2508 W at a convective cooling film coefficient of 400 W/m2K. These trends indicate that both PV power and PV-TE power display a favourable response to increasing convective cooling film coefficients, resulting in improved power generation. The observed behaviour of the system highlights the importance of finding an optimal balance between the convective cooling film coefficient and power generation. The findings suggest that the system’s performance in terms of PV-TE power generation is optimal, as it demonstrates an increasing trend with increasing convective cooling film coefficients. The impact and effect of this performance optimization are significant for the design and operation of PV-TE systems. Also, the results indicate that the TEG system achieved its peak efficiency of 2.85% at a convective cooling film coefficient of 3600 W/m2K, while its lowest efficiency of 2.79% was obtained at a convective cooling film coefficient of 400 W/m2K. This suggests that the TEG efficiency is sensitive to changes in the convective cooling conditions, with higher cooling coefficients leading to slightly improved efficiency. Similarly, the PV efficiency exhibited its lowest efficiency of 10.56% at a convective cooling film coefficient of 400 W/m2K, while its optimal efficiency of 11.001% was achieved at a convective cooling film coefficient of 3600 W/m2K. Furthermore, the PV-TE efficiency reached its optimal efficiency of 13.85% at 3600 W/m2K convective cooling film coefficient and its lowest efficiency of 13.35% at 400 W/m2K convective cooling film coefficient. This highlights the significant impact of the convective cooling film coefficient on the overall efficiency of the system. The observed trends indicate that higher convective cooling film coefficients lead to improved efficiencies for all three components: TEG, PV, and PV-TE. It further suggests that efficient heat dissipation through enhanced convective cooling can positively influence the performance of both the thermoelectric and photovoltaic subsystems, resulting in higher overall system efficiency. The impact and effect of this performance optimization are crucial for the design and operation of PV-TE systems. By adjusting the convective cooling film coefficient, it is possible to enhance the efficiency of both the TEG and PV subsystems, leading to improved overall energy conversion efficiency.
The results reveal interesting trends related to TE leg height. First, there is a decrease in both PV power and PV-TE power as TE leg height increases, which is contrary to the behaviour observed for TEG power. The system achieved its optimal TEG power of 0.0145 W at a TE leg height of 9 mm, while the lowest TEG power of 0.0024 W was obtained at a TE leg height of 1 mm. This suggests that the TEG power generation is influenced by the height of the TE leg, with taller legs leading to increased power output. Conversely, the PV component exhibited its optimal power of 0.2752 W at a TE leg height of 1 mm, whereas the lowest PV power of 0.2390 W was observed at a TE leg height of 9 mm. Similarly, the PV-TE power reached its maximum of 0.2777 W at a TE leg height of 1 mm, while the lowest power of 0.2536 W was obtained at a TE leg height of 9 mm. These results indicate that both PV power and PV-TE power decrease as the TE leg height increases. Regarding efficiency, the TEG demonstrated its optimal efficiency of 1.25% at a TE leg height of 9 mm, while the lowest efficiency of 0.16% was observed at a TE leg height of 1 mm. In contrast, for the PV and PV-TE components, the efficiencies decrease with increasing TE leg height. The PV achieved its highest efficiency of 14.99% at a TE leg height of 1 mm, whereas the lowest efficiency of 13.01% was recorded at a TE leg height of 9 mm. Similarly, the PV-TE component exhibited its optimal efficiency of 15.15% at a TE leg height of 1 mm, while the lowest efficiency of 14.27% was observed at a TE leg height of 9 mm. The optimization of performance in the hybrid system is impacted by the choice of TE leg height. The results indicate that taller TE legs favour increased TEG power generation and efficiency but lead to decreased PV power and PV-TE power. Conversely, shorter TE legs result in higher PV power and PV-TE power, along with improved PV and PV-TE efficiencies, but at the expense of reduced TEG power and efficiency. The implications of these findings are significant for the design and operation of hybrid systems integrating TE and PV components. Engineers and researchers can use this knowledge to fine-tune the TE leg height based on their priorities and specific system requirements. By carefully selecting the appropriate TE leg height, it is possible to optimize either TEG power generation and efficiency or PV power generation and efficiency, depending on the desired outcome.
The results further demonstrate a clear relationship between the TE leg cross-sectional area and power generation. Both PV power and PV-TE power increased as the TE leg cross-sectional area increased. The PV power reached its maximum value of 0.26 W when the TE leg cross-sectional area was 32.5 mm2, while the minimum power of 0.11 W was observed at a TE leg cross-sectional area of 2.5 mm2. Similarly, the optimal PV-TE power of 0.26 W was obtained at 32.5 mm2 of TE leg cross-sectional area, with the lowest power of 0.17 W observed at 2.5 mm2 of TE leg cross-sectional area. Conversely, the TEG power exhibited an opposite trend, with the minimum power of 0.0076 W at 32.5 mm2 of TE leg cross-sectional area and the optimal TEG power of 0.05 W at 2.5 mm2 of TE leg cross-sectional area. Efficiency, a critical performance metric, also demonstrated variation with the TE leg cross-sectional area. The optimal PV efficiency was recorded at 32.5 mm2 of TE leg cross-sectional area, achieving 14.18%, while the lowest efficiency of 6.30% was observed at 2.5 mm2 of TE leg cross-sectional area. Similarly, the PV-TE efficiency exhibited its maximum value of 14.77% at 32.5 mm2 of TE leg cross-sectional area, with the minimum efficiency of 12.45% observed at 2.5 mm2 of TE leg cross-sectional area. In contrast, the TEG efficiency reached its maximum of 6.15% at 2.5 mm2 of TE leg cross-sectional area, while the lowest efficiency of 0.58% was observed at 32.5 mm2 of TE leg cross-sectional area. Increasing the TE leg cross-sectional area resulted in higher PV power and PV-TE power generations, leading to improved PV and PV-TE efficiencies. However, this increase in TE leg cross-sectional area had a detrimental effect on TEG power generation and efficiency, with lower power outputs and reduced TEG efficiency.
Interestingly, the TEG power exhibited an increasing trend with rising ambient temperature, in contrast to the PV and PV-TE powers, which experienced a significant decrease. The optimal TEG power of 0.012 W was achieved at an ambient temperature of 313 K, while the minimum TEG power of 0.005 W was observed at 273 K. On the other hand, the PV power reached its peak of 0.28 W at 273 K, while its lowest value of 0.22 W was obtained at 313 K. Similarly, the PV-TE power showed its optimal power output of 0.29 W at 273 K, with the least power of 0.24 W occurring at 313 K. In terms of efficiency, the TEG efficiency exhibited an increasing trend with increasing ambient temperature, in contrast to the PV and PV-TE efficiencies, which decreased as the ambient temperature rose. The optimal TEG efficiency of 0.87% was achieved at 313 K, while the minimum TEG efficiency of 0.51% was observed at 273 K. The PV efficiency reached its maximum value of 15.74% at 273 K, while its lowest efficiency of 12.49% was obtained at 313 K. Similarly, the optimal PV-TE efficiency was 16.25% at 273 K, while the least efficiency of 13.37% was observed at 313 K ambient temperature. The results highlight the contrasting behaviour of TEG power and efficiency compared to PV and PV-TE powers and efficiencies with varying ambient temperatures. The increasing TEG power and efficiency with rising ambient temperature suggest that higher temperatures provide a favourable environment for TEG performance. On the other hand, the decreasing trend observed in PV and PV-TE powers and efficiencies emphasizes the negative impact of elevated ambient temperatures on their performance. These findings have significant implications for system optimization and real-world applications. Engineers and researchers can utilize this information to design hybrid systems that are better suited for specific operating conditions. For applications where TEG performance is crucial, operating at higher ambient temperatures can lead to increased power generation and efficiency. Conversely, for applications that prioritize PV and PV-TE performances, efforts should be made to mitigate the negative effects of high ambient temperatures to maintain optimal power output and efficiency. The performance optimization of the hybrid system with varying ambient temperatures reveals distinct behaviours in TEG, PV, and PV-TE power generations and efficiencies. The observed trends emphasize the importance of temperature management and system design considerations to achieve optimal performance.
The TEG power exhibited a decreasing trend as wind speed increased, while the opposite pattern was observed for the PV and PV-TE powers. The optimal TEG power of 0.008 W was achieved at a wind speed of 0.5 m/s, whereas the minimum TEG power of 0.004 W was recorded at an 8.5 m/s wind speed. On the other hand, the PV power reached its maximum of 0.27 W at 8.5 m/s wind speed, with the lowest PV power of 0.25 W obtained at 0.5 m/s wind speed. Similarly, the PV-TE power exhibited its optimal power output of 0.27 W at 8.5 m/s wind speed, while the least power of 0.26 W was observed at 2.5 m/s wind speed. In terms of efficiency, the TEG efficiency displayed its maximum value of 0.71% at 0.5 m/s wind speed, while the least efficiency of 0.48% was achieved at 8.5 m/s wind speed. The PV efficiency reached its peak value of 14.81% at 8.5 m/s wind speed, whereas the lowest efficiency of 13.91% was obtained at 2.5 m/s wind speed. Similarly, the PV-TE efficiency exhibited its optimal efficiency of 15.30% at 8.5 m/s wind speed, with the least efficiency of 14.62% observed at 2.5 m/s wind speed. These findings highlight the contrasting effects of wind speed on TEG, PV, and PV-TE performances within the hybrid system. While TEG power and efficiency decrease with increasing wind speed, PV and PV-TE powers and efficiencies demonstrate an increasing trend. This suggests that higher wind speeds provide favourable conditions for PV-based power generation, while TEG performance is hindered. By harnessing higher wind speeds, the PV and PV-TE components can achieve increased power output and efficiency, enhancing the overall performance of the hybrid system. On the other hand, for applications where TEG performance is prioritized, wind speed management may be crucial to mitigate the decrease in power generation and efficiency.
Furthermore, the TEG power exhibited a decreasing trend as PV emissivity increased, indicating a negative correlation between the two parameters. The maximum TEG power of 0.009 W was achieved at a PV emissivity of 0.1, while the minimum TEG power of 0.008 W was observed at a PV emissivity of 0.9. In contrast, the PV power displayed its optimal power output of 0.257 W at a PV emissivity of 0.9, with the lowest PV power of 0.253 W obtained at a PV emissivity of 0.5. Similarly, the PV-TE power reached its maximum power of 0.265 W at a PV emissivity of 0.9, while the lowest power of 0.262 W was recorded at a PV emissivity of 0.1. Furthermore, the TEG efficiency demonstrated its optimal efficiency of 0.74% at a PV emissivity of 0.1, while the least efficiency of 0.68% was observed at a PV emissivity of 0.9. Similarly, the PV efficiency reached its peak value of 14.01% at a PV emissivity of 0.9, whereas the lowest efficiency of 13.77% was obtained at a PV emissivity of 0.1. Additionally, the PV-TE efficiency exhibited its highest efficiency of 14.70% at a PV emissivity of 0.9, with an efficiency of 14.52% recorded at a PV emissivity of 0.1. These findings indicate the influence of PV emissivity on the performance of the hybrid system. As PV emissivity increases, TEG power and efficiency tend to decrease, suggesting that higher emissivity values may hinder TEG performance. Conversely, PV power, PV-TE power, and the corresponding efficiencies show an increasing trend with higher PV emissivity, indicating a positive impact on PV-based power generation and overall system performance.
3.3. Performance of the Neural Network
The performance metrics of the neural network from Figure 7 for the hybrid system were evaluated using various parameters, including validation performance, mean squared error (MSE), gradient value, regression score, and regression plot. The best validation performance was achieved at epoch 0, with a value of . This indicates that the neural network model performed exceptionally well during the validation phase, producing highly accurate predictions. It demonstrates the effectiveness of the model in capturing and generalizing the underlying patterns and relationships within the hybrid system data. The mean squared error (MSE) was used as a measure of the model’s prediction accuracy. It can be said that the neural network achieved a low MSE, given the excellent validation performance observed. A low MSE signifies that the model’s predicted values closely align with the actual values, indicating a high level of precision in its predictions. The gradient value at epoch 99 was determined to be . The gradient represents the rate of change of the model’s performance with respect to its parameters. A small gradient value suggests that the model has converged close to an optimal solution and is exhibiting stable performance. This indicates that the neural network has effectively learned the patterns and features of the hybrid system data, enabling accurate predictions. The regression score, reported as 0.99 for training, validation, and testing, signifies the strength of the linear relationship between the predicted and actual values. A score of 0.99 indicates a high correlation and a strong linear relationship, suggesting that the neural network model has successfully captured the underlying patterns and trends within the hybrid system data. The regression plot, which does not show any outliers, further confirms the accuracy and effectiveness of the neural network model. An absence of outliers suggests that the model’s predictions closely match the actual values across the entire range of the data. This indicates that the model’s performance is consistent and reliable, providing valuable insights and predictions for the hybrid system. Overall, the performance metrics of the neural network for the hybrid system demonstrate its capability to accurately predict and model the behaviour of the system. The high validation performance, low mean squared error, small gradient value, and strong regression score collectively indicate the effectiveness of the model in capturing the complex relationships within the hybrid system data. This, in turn, enables the system to make precise predictions and valuable insights, contributing to improved efficiency, optimization, and decision-making for the hybrid system.
3.4. Optimization Benefits
Throughout our comprehensive discussion, we have explored the performance optimization of various components and parameters in the hybrid system. By analyzing the effects of different factors on power generation, efficiency, and overall system performance, we can assess the benefits of these optimizations. When examining the impact of semiconductors, we observed that the PV-TE efficiency was maximized at lower sunshine intensity, with LiNiO semiconductor exhibiting the highest efficiency. In contrast, nanosemiconductor showed the least efficiency. Interestingly, both materials experienced higher PV-TE efficiency at lower sunshine intensity. The convective cooling film coefficient played a crucial role in system performance. The TEG power decreased with increasing coefficient, while the PV and PV-TE powers reached their peak at a specific coefficient. The TEG system demonstrated its highest efficiency at this optimal coefficient.
TE leg height also influenced the system’s performance. TEG power decreased, and PV and PV-TE powers followed suit with increasing TE leg height. On the contrary, TEG efficiency reached its maximum at a higher leg height, while PV and PV-TE efficiencies were optimal at a lower leg height. Examining the TE leg cross-sectional area, we found that increasing the area resulted in higher PV and PV-TE power generations. However, TEG power decreased with increasing area. Similarly, PV and PV-TE efficiencies increased, while TEG efficiency decreased. The system’s performance was also affected by ambient temperature. TEG power increased, and PV and PV-TE powers decreased with rising temperature. TEG efficiency increased, while PV and PV-TE efficiencies decreased. Higher ambient temperatures led to optimal TEG power and efficiency, while lower temperatures resulted in optimal PV and PV-TE powers and efficiencies. Wind speed had a contrasting effect on different aspects of the system. TEG power decreased as wind speed increased, whereas PV and PV-TE powers increased. TEG efficiency decreased, while PV and PV-TE efficiencies increased with increasing wind speed. Lower wind speeds led to optimal TEG power and efficiency, while higher wind speeds resulted in optimal PV and PV-TE powers and efficiencies. Finally, PV emissivity impacted TEG power and efficiency. TEG power decreased as emissivity increased, while PV and PV-TE powers remained relatively constant. TEG efficiency decreased, while PV and PV-TE efficiencies remained relatively constant with increasing emissivity.
These optimization benefits collectively contribute to enhancing the overall performance, power generation, and efficiency of the hybrid system. By fine-tuning and optimizing the different components and parameters, we can design more efficient and reliable systems for various applications. These optimizations have significant implications for renewable energy generation, power harvesting, and the development of energy-efficient technologies. Ultimately, they enable us to maximize energy generation, utilize resources more effectively, and contribute to a more sustainable and cost-effective operation of hybrid energy systems.
Table 3 provides a comparison between the initial and optimized values of various parameters in the hybrid system. The convective cooling film coefficient, which determines the heat transfer rate, was initially set at 500 W/m2K but was optimized to 3600 W/m2K. This increase indicates the improved ability of the system to dissipate heat, leading to enhanced performance. The ambient temperature, which affects the overall system temperature, was initially at 295.15 K but was optimized to 273 K. The lower optimal temperature suggests a more favourable operating condition for the system, resulting in increased power generation. The wind speed, an external factor influencing the system’s performance, had an initial value of 1 m/s. However, through optimization, it reached an optimal value of 8.5 m/s. The higher wind speed likely leads to better cooling and improved power output from the system. The PV emissivity, representing the efficiency of the photovoltaic modules in converting sunlight into electricity, had an initial value of 0.8. Through optimization, it reached an optimal value of 0.9. This increase suggests an improved ability of the PV modules to absorb and convert solar energy, resulting in higher power generation. Additionally, the optimization process involved adjusting the TE leg height and TE semiconductor cross-sectional area. The TE leg height was optimized to 1 mm from an initial value of 5 mm, while the TE semiconductor cross-sectional area was optimized to 32.5 mm2 from an initial value of 25 mm2. These adjustments likely contributed to improving energy conversion efficiency and power generation. The concentration ratio had an optimal value of 5, which remained the same as its initial value. This suggests that further optimization did not significantly impact this particular parameter. The optimization of these parameters in the hybrid system resulted in improved performance, including enhanced heat transfer, lower operating temperature, increased wind speed, improved PV efficiency, and optimized TE leg height and cross-sectional area. These adjustments collectively contribute to the overall improvement in power generation and efficiency of the hybrid system.
Table 4 showcases the performance gains achieved through the optimization process of the hybrid system. The initial power output of the system was recorded as 0.26 W, which was improved to 0.32 W after optimization. This enhancement represents a significant 23.1% increase in power generation. The higher power output indicates an improved capability of the system to convert energy from TEG and PV modules. In addition to the power gains, the optimization process also had a positive impact on the efficiency of the system. The optimized efficiency reached 17.62%, while the unoptimized efficiency was measured at 14.68%. This signifies a 20% enhancement in efficiency, which is a notable improvement in the system’s ability to convert available energy into usable electrical power. The impact of these performance gains is substantial for the hybrid system. The increase in power output implies that more electricity can be generated, which is crucial for meeting energy demands. It enhances the overall effectiveness of the system in generating sustainable power from both the TEG and PV modules. The improved efficiency also indicates that a higher proportion of the available energy is being effectively converted into usable electrical power, resulting in a more efficient and sustainable energy conversion process. These performance gains have practical implications for the system’s application in various fields, such as renewable energy generation, off-grid power systems, and energy harvesting.
4. Conclusions and Future Studies
Conclusively, this comprehensive investigation into optimal thermoelectric semiconductors and the performance optimization of TEG-PV hybrid systems using neural networks has yielded valuable insights and significant advancements in sustainable energy solutions. Through the analysis of various influential factors, including semiconductor materials, cooling film coefficient, leg height and cross-sectional area, ambient temperature, wind speed, and PV emissivity, this study has successfully optimized the power generation and efficiency of the hybrid system. The unoptimized power output of the system was measured at 0.26 W, whereas the optimized power output achieved a remarkable enhancement, reaching 0.32 W. This represents a substantial percentage increase of 23.1% in power generation. Similarly, the unoptimized efficiency was measured at 14.68%, which was significantly improved to 17.62% through the optimization process. This enhancement in efficiency represents a notable percentage increase of 20%. The findings emphasize the importance of careful parameter optimization in achieving enhanced performance in TEG-PV hybrid systems. The optimization process considered factors such as semiconductor materials, cooling film coefficient, leg dimensions, ambient temperature, wind speed, and PV emissivity, highlighting their individual and combined impact on power generation and efficiency.
Furthermore, this research has shed light on the close effects between different parameters in the hybrid system, reinforcing the significance of a holistic approach in achieving optimal performance. By understanding and harnessing the interplay between various factors, further advancements can be made in the design and operation of TEG-PV hybrid systems. The results of this study have practical implications for the development and implementation of sustainable energy solutions. The optimized power output and efficiency enhancements contribute to the viability and effectiveness of TEG-PV hybrid systems in real-world applications. Through continued research and development, TEG-PV hybrid systems can play a crucial role in meeting the world’s energy demands while reducing environmental impact.
4.1. Future Studies
The presented study on optimizing TEG-PV hybrid systems opens up several exciting avenues for future research. These thrusts are aimed at further enhancing the performance and viability of these systems in sustainable energy solutions.
Firstly, exploring advanced materials holds great potential for improving the efficiency and stability of thermoelectric semiconductors. Investigating emerging materials like perovskites or quantum dots can lead to significant advancements in power generation and efficiency.
Secondly, optimizing system design configurations and exploring innovative approaches, such as multistage systems, can maximize the synergistic effects between TEG and PV modules. This would require studying the impact of scaling, component placement, and thermal management strategies on overall system performance.
Thirdly, developing more accurate models and simulations will enable better predictions of TEG-PV hybrid system performance under varying operating conditions. Incorporating factors like partial shading or dynamic weather conditions will lead to more realistic estimations and aid in system design optimization.
Additionally, exploring intelligent control and management strategies will be crucial. Advanced algorithms for power flow management, energy storage integration, and real-time system control can enhance system stability and reliability.
Economic and environmental analyses are also vital to evaluating the cost-effectiveness and sustainability of TEG-PV hybrid systems. Assessing their feasibility for large-scale deployment and identifying barriers to commercialization will aid in promoting their adoption.
Lastly, field testing and validation of prototype systems are necessary to validate performance improvements and gather real-world data. By pursuing these research thrusts, we can advance our understanding of TEG-PV hybrid systems, making them more efficient, economically viable, and environmentally friendly.
Data Availability
The data used in this manuscript is available upon reasonable request from the corresponding author.
Additional Points
Highlights. (1) Improved efficiency: 20% enhancement from 14.68% to 17.62%. (2) Multifactor optimization: factors analyzed include semiconductor materials, cooling coefficient, temperature, wind speed, and PV emissivity. (3) Synergistic effects: consideration of the interplay between parameters for optimal performance. (4) Machine learning-driven optimization: utilization of neural networks for comprehensive data analysis
Conflicts of Interest
The authors declare that they have no conflicts of interest.
Acknowledgments
The authors would like to acknowledge the Deanship of Scientific Research at Najran University for funding this work under the research collaboration funding program grant code (NU/DRP/SERC/12/12), the Deanship of Scientific Research at King Faisal University for funding this work under the research collaboration funding program grant number GRANT: 3717, and the Office of Research and Sponsored Programs (ORSP) at Abu Dhabi University for funding this paper through a research fund.