Optimizing solar photovoltaic power forecasting via multi-architecture machine learning framework with multiple hyperparameter optimization techniques

In this paper, a stochastic model is proposed for a joint statistical description of solar photovoltaic (PV) power and outdoor temperature. The underlying correlation emerges from solar irradiance that is responsible in part for both the variability in solar PV power and temperature. The proposed model can be used to capture the uncertainty in solar PV power and its correlation with the electric power consumption of thermostatically controlled loads. First, a model for solar PV power that explicitly incorporates the stochasticity due to clouds via a regime-switching process between the three classes of sunny, overcast and partly cloudy is proposed. Then, the relationship between temperature and solar power is postulated using a second-order Volterra model. This joint modeling is leveraged to develop a joint probabilistic forecasting method for solar PV power and temperature. Real-world datasets that include solar PV power and temperature measurements in California are analyzed and the effectiveness of the joint model in providing probabilistic forecasts is verified. The proposed forecasting methodology outperforms several reference methods thus portraying that the underlying correlation between temperature and solar PV power is well defined and only requires a simple lower-complexity sampling space.

Comparison of physical and machine learning models for estimating solar irradiance and photovoltaic power

Solar energy is an abundant, clean, and renewable energy source, crucial for addressing the current global energy crisis. Efficiently harvesting solar power to generate electricity for smart grids is vital. However, the variability of solar radiation presents significant challenges in accurately forecasting solar photovoltaic (PV) power generation. Elements such as cloud cover, atmospheric conditions, and seasonal changes greatly influence the amount of solar energy available for electricity conversion. Accurate estimation of solar power output is therefore critical to evaluate the potential of smart grids. This study explores the use of various machine learning models to predict solar PV power generation in Lubbock, Texas. Performance is measured using Mean Squared Error (MSE) and R² metrics. The findings reveal that the Random Forest Regression (RFR) and Long Short-Term Memory (LSTM) models outperformed the others, achieving MSE values of 2.06% and 2.23%, and R² values of 0.977 and 0.975, respectively. These results indicate that RFR and LSTM are highly effective in capturing the complex patterns and relationships in solar power generation data. The developed machine learning models can assist solar PV investors in optimizing their processes and enhancing their planning for solar energy production

Due to the intermittent behaviour of the sun, accurate prediction of solar photovoltaic (PV) power is crucial for efficient and reliable operation of solar power plants. This paper presents state of the art approach for PV panels power prediction using machine learning (ML) method. Two ML models, namely Random Forest (RF) and Support Vector Machine (SVM) are trained and tested using input data of solar irradiance, ambient temperature, wind speed, humidity, precipitation and PV output power. The case study is presented for the grid tied PV system installed at University Tun Hussein Onn campus Batu Pahat Malaysia. The results indicated regression predictions reasonably fit the actual data, proving good potential of ML for PV power prediction. Besides, the predictive performance of RF and SVM was compared based on three evaluation metrics: coefficient of determination (R2), root mean squared error (RMSE) and mean absolute error (MAE). Both ML models showed comparable predictive power with RF performing slightly better than SVM. The R2 value for RF was 0.850 compared to 0.832 for SVM, indicating that RF was able to explain more of the variability in the data. Additionally, RF had lower values for both RMSE and MAE, indicating that it was better able to predict values of the solar PV power output. The conclusion from this study imparts the importance of ML methods to predict PV power which could be useful for optimizing the efficiency and reliability of solar energy systems.

The road ahead for solar PV power

PurposeThis paper aims to Solar photovoltaic (PV) power can significantly impact the power system because of its intermittent nature. Hence, an accurate solar PV power forecasting model is required for appropriate power system planning.Design/methodology/approachIn this paper, a long short-term memory (LSTM)-based double deep Q-learning (DDQL) neural network (NN) is proposed for forecasting solar PV power indirectly over the long-term horizon. The past solar irradiance, temperature and wind speed are used for forecasting the solar PV power for a place using the proposed forecasting model.FindingsThe LSTM-based DDQL NN reduces over- and underestimation and avoids gradient vanishing. Thus, the proposed model improves the forecasting accuracy of solar PV power using deep learning techniques (DLTs). In addition, the proposed model requires less training time and forecasts solar PV power with improved stability.Originality/valueThe proposed model is trained and validated for several places with different climatic patterns and seasons. The proposed model is also tested for a place with a temperate climatic pattern by constructing an experimental solar PV system. The training, validation and testing results have confirmed the practicality of the proposed solar PV power forecasting model using LSTM-based DDQL NN.

Should China focus on the distributed development of wind and solar photovoltaic power generation? A comparative study

China is a big consumer of energy resources. With the gradual decrease of non-renewable resources such as oil and coal, it is very important to adopt renewable energy for economic development. As a kind of abundant renewable energy, solar power has been widely used. This paper introduces the development status of solar power generation technology, mainly introduces solar photovoltaic power generation technology, briefly describes the principle of solar photovoltaic power generation, and compares and analyzes four kinds of solar photovoltaic power generation technology, among which photovoltaic power generation technology is the most mature solar photovoltaic power utilization technology at present.

Computational solar energy – Ensemble learning methods for prediction of solar power generation based on meteorological parameters in Eastern India

Achieving grid parity of solar PV power in China- The role of Tradable Green Certificate

The main objective of this study is to develop ANN-based predictive models for short-term forecasting of solar PV power output and battery state of charge. The 3Ds energy model that integrates Decarbonization, Digitalization, and Decentralization of the energy system to facilitate the shift towards sustainable energy sources is used for this electric vehicle project. The experimental set up includes solar PV panels, a solar inverter, a battery inverter, and a battery bank. Additionally, smart meters were installed to collect real-time performance data from the solar PV-powered electric vehicle (EV) charging station. The weather and real-time system performance data were used to develop short term forecasting models based on Artificial Neural Networks (ANN) and the Levenberg-Marquardt method, to predict the performance of the system (power output and state of the charge) ahead. The R values for the prediction models of solar photovoltaic (PV) specific power and battery state of charge fall within the range of 0.9957–0.9969 and 0.9990 to 0.9996, respectively. Furthermore, the mean squared errors (MSEs) for the artificial neural network (ANN) models pertaining to solar photovoltaic (PV) power output and battery state of charge exhibit a variety of values. Specifically, the MSEs for solar PV power output range from 1.242 x 10^-4 to 1.579 x 10^-4, while the MSEs for battery status of charge range from 0.1889 to 0.3402. Artificial neural network (ANN) models possess considerable promise for practical implementations as they simplify intricate connections among inputs, parameters, and outputs in real-world scenarios. The level of accuracy and short-term predictive models of the solar PV powered EV charging station are very important for achieving a balance between the supply of solar photovoltaic (PV) system and the demand for electric vehicles (EVs). Predictive models will help build complex control mechanisms for controlling and optimizing solar PV-powered charging stations, supervising their operation and maintenance, and simplifying renewable energy pre-purchase. Future works will include the development of an energy management system to improve the efficiency of the solar stations charging the EVs, the establishment of blockchain networks for both the solar PV system and battery bank, and the integration of cybersecurity protocols for the charging station.

For the efficient functioning of grid tied solar photovoltaic array in distribution system, the control algorithm is proposed here utilizing a warped digital filter. The solar photovoltaic (PV) array supplies maximum power to the grid/the load by maximum power point tracking (MPPT) technique namely a perturb and observe (P&O) method. The control is termed as a dual mode, as it performs satisfactorily during availability/unavailability of solar PV array power. When the solar PV array power is incident, the proposed system is capable of transferring active power from solar PV array to the grid/the load while improving power quality using a voltage source converter (VSC) working as a distribution static compensator (DSTATCOM). However, during cloudy days or at night, when solar power is not present, the system performs as DSTATCOM for reactive power compensation and elimination of harmonics. The control algorithm provides smooth transition between these modes of operation of the system. The control implemented using warped digital filters, is adaptive as its coefficients are updated depending upon system requirements. A feed-forward term for solar PV array power is incorporated, for the enhancement of dynamic response. A prototype of the system is developed in the laboratory and its performance is studied for varying loads, varying solar insolation conditions and change of modes between grid tied PV array and DSTATCOM.

Accurate predictions of solar photovoltaic (PV) power generation at different time horizons are essential for reliable operation of energy management systems. The output power of a PV power plant is dependent on non-linear and intermittent environmental factors, such as solar irradiance, wind speed, relative humidity, etc. Intermittency and randomness of solar PV power effect precision of estimation. To address the challenge, this paper presents a Swarm Decomposition Technique (SWD) based hybrid model as a novel approach for very short-term (15 min) solar PV power generation forecast. The original contribution of the study is to investigate use of SWD for solar data forecast. The solar PV power generation data with hourly resolution obtained from the field (grid connected, 857.08 kWp Akgul Solar PV Power Plant in Turkey) are used to develop and validate the forecast model. Specifically, the analysis showed that the hybrid model with SWD technique provides highly accurate predictions in cloudy periods.

Pakistan has an abundant solar power potential which can be effectively utilized for the electricity generation. There are various sites across the country which have sufficient solar irradiation across the year, and thus, suitable for the installation of solar photovoltaic (PV) power projects. This study, therefore, aims to undertake research on the establishment of solar power project site selection in Pakistan. In this context, 14 promising cities of Pakistan are considered as alternatives and studied in terms of economic, environmental, social, location, climate, and orography criteria and further supplemented with 20 sub-criteria. Initially, the analytical hierarchy process (AHP) method has been used to prioritize each of the main criteria and sub-criteria. Later, fuzzy VlseKriterijuska Optimizacija I Komoromisno Resenje (F-VIKOR) method has been employed to prioritize the 14 alternatives. The present investigation reveals that Khuzdar (C2), Badin (C3), and Mastung (C7) are the most suitable cities for the installation of solar PV power projects in Pakistan. Finally, the outcome of the sensitivity analysis revealed that obtained results are reliable and robust for the installation of solar PV power projects in Pakistan. This study shall assist government, energy planners, and policymakers in making cities sustainable by establishing solar power projects in Pakistan.

The increased penetration of solar photovoltaic (PV) energy sources into electric grids has increased the need for accurate modeling and prediction of solar irradiance and power production. Existing modeling and prediction techniques focus on long-term low-resolution prediction over minutes to years. This paper examines the stochastic modeling and short-term high-resolution prediction of solar irradiance and PV power output. We propose a stochastic state-space model to characterize the behaviors of solar irradiance and PV power output. This prediction model is suitable for the development of optimal power controllers for PV sources. A filter-based expectation-maximization and Kalman filtering mechanism is employed to estimate the parameters and states in the state-space model. The mechanism results in a finite dimensional filter which only uses the first and second order statistics. The structure of the scheme contributes to a direct prediction of the solar irradiance and PV power output without any linearization process or simplifying assumptions of the signal's model. This enables the system to accurately predict small as well as large fluctuations of the solar signals. The mechanism is recursive allowing the solar irradiance and PV power to be predicted online from measurements. The mechanism is tested using solar irradiance and PV power measurement data collected locally in our laboratory.