Robust modelling framework for short-term forecasting of global horizontal irradiance (original) (raw)
Related papers
The use of photovoltaic solar plants for the generation of electrical energy has been constantly increasing in recent years, and many of these plants are connected to the external electrical network, which makes it necessary to forecast the electrical energy generated by the solar plants to assist in the management of the network operator. This research presents a methodology based on data science to develop the forecast of electrical energy generated from photovoltaic solar plants, using three different techniques for comparison purposes: time series analysis, multiple linear regression, and artificial neural network. Historical data of peak power, solar irradiance, ambient temperature, wind speed, and soiling rate from an experimental NREL photovoltaic solar plant were used. To evaluate the performance of the models, the RMSE, MAE, and MAPE metrics are used, resulting in the ARIMA model of the time series analysis having the best performance with a MAE of 1.38 kWh, RMSE of 1.40 kWh, and MAPE of 6.35%. In the correlation analysis, it was determined that power generation was independent of the soiling rate, so this variable was discarded in the regression models.
Solar Energy, 2010
Due to strong increase of solar power generation, the predictions of incoming solar energy are acquiring more importance. Photovoltaic and solar thermal are the main sources of electricity generation from solar energy. In the case of solar thermal energy plants with storage energy system, its management and operation need reliable predictions of solar irradiance with the same temporal resolution as the temporal capacity of the backup system. These plants can work like a conventional power plant and compete in the energy stock market avoiding intermittence in electricity production. This work presents a comparisons of statistical models based on time series applied to predict half daily values of global solar irradiance with a temporal horizon of 3 days. Half daily values consist of accumulated hourly global solar irradiance from solar raise to solar noon and from noon until dawn for each day. The dataset of ground solar radiation used belongs to stations of Spanish National Weather Service (AEMet). The models tested are autoregressive, neural networks and fuzzy logic models. Due to the fact that half daily solar irradiance time series is non-stationary, it has been necessary to transform it to two new stationary variables (clearness index and lost component) which are used as input of the predictive models. Improvement in terms of RMSD of the models essayed is compared against the model based on persistence. The validation process shows that all models essayed improve persistence. The best approach to forecast half daily values of solar irradiance is neural network models with lost component as input, except Lerida station where models based on clearness index have less uncertainty because this magnitude has a linear behaviour and it is easier to simulate by models.
IEEE Access
Due to the increasing cost of crude oil because of pandemic COVID-19 and global environmental threats, the exploitation of fossil fuels for power generation is discouraged. Further, the demand for electrical power is increasing drastically, and therefore, the exploitation of renewable energy resources, particularly solar photovoltaic-based technology for power generation is invigorated. However, the large-scale penetration of solar photovoltaic is becoming a major challenge in terms of stability, reliability of power when integrated with the grid. Thus, it becomes important to develop a novel approach or strategy which is useful to improve power quality, reliability, and grid stability. Solar photovoltaic power forecasting is a key tool for this new era and becoming the main component for a smart grid environment. Here, in this paper, the ensemble trees approach-based machine learning approach is utilized to forecast the solar photovoltaic power with the help of various meteorological parameters. The high-quality measured data for meteorological parameters for Qassim, Saudi Arabia is used in this research. The performance of the proposed model is evaluated with the help of statistical indices such as Root Mean Square Error (RMSE), Mean Absolute Error (MAE), Mean Square Error (MSE), Mean Absolute Percentage Error (MAPE), Training Time (TT) and found within the desired limits. To validate the obtained results a comparative analysis with other machine learning models is carried out. Moreover, the proposed research may provide the roadmap in achieving the vision 2030 of the government of Saudi Arabia. INDEX TERMS Ensemble trees approach, machine learning, regression analysis, renewable energy resources, solar photovoltaic plants, solar photovoltaic forecasting.
Analysis of Factors for Forecasting Electric Power Generation by Solar Power Plants
POWER ENGINEERING: economics, technique, ecology, 2020
The new model of the wholesale electricity market in Ukraine causes appearance the market for the day ahead. In this market, the generating company undertakes to supply a certain amount of electricity. So, it is necessary to carry on the most accurate forecast of possible electricity generation by solar power plant (SPP). Generation value depends on certain factors. A brief summary of different influence of parameters on the PV cell performance has been provided. The article analyzes and identifies the factors that should be included in the forecast mathematical model of electricity generation by a solar power plant for a certain short-term period. According to analyzed data from SPP located in the Kyiv region, such parameters are the intensity of solar radiation, temperature and humidity, wind speed, and atmospheric pressure. The degree of influence of these factors on the initial function of electric energy generation were estimated by analyzing the scatter plot diagrams of relati...
Springer eBooks, 2022
The accurate modeling and forecasting of the power output of photovoltaic (PV) systems are critical to efficiently managing their integration in smart grids, delivery, and storage. This paper intends to provide efficient short-term forecasting of solar power production using Variational AutoEncoder (VAE) model. Adopting the VAE-driven deep learning model is expected to improve forecasting accuracy because of its suitable performance in time-series modeling and flexible nonlinear approximation. Both single-and multi-step-ahead forecasts are investigated in this work. Data from two grid-connected plants (a 243 kW parking lot canopy array in the US and a 9 MW PV system in Algeria) are employed to show the investigated deep learning models' performance. Specifically, the forecasting outputs of the proposed VAE-based forecasting method have been compared with seven deep learning methods, namely recurrent neural network, Long short-term memory (LSTM), Bidirectional LSTM, Convolutional LSTM network, Gated recurrent units, stacked autoencoder, and restricted Boltzmann machine, and two commonly used machine learning methods, namely logistic regression and support vector regression. The results of this investigation demonstrate the satisfying performance of deep learning techniques to forecast solar power and point out that the VAE consistently performed better than the other methods. Also, results confirmed the superior performance of deep learning models compared to the two considered baseline machine learning models.
A Smart Analysis and Visualization of The Power Forecasting in Pakistan
International Journal of Computing and Digital Systems, 2021
Over the last decade, the energy sector has experienced a significant modernization cycle. Its network is undergoing accelerated upgrades. The instability of production, demand and markets is far less stable than ever before. Also, the corporate concept is profoundly questioned. Many decision-making processes in this competitive and complex setting depend on probabilistic predictions to measure unpredictable futures. In recent years, the interest in probabilistic energy forecasting analysis has rapidly begun, even though many articles in the energy forecasting literature focus on points or single-valuation forecasting. In Pakistan, the bulk of early studies require various kinds of econometric modeling. However, the simulation of time series appears to deliver more reliable results, given the projected economic and demographic parameters usually deviate from the achievements. The machine learning technique "ARIMA" and deep learning technique Long Short-Term Memory "LSTM," are used to calculate Pakistan's future primary energy demand from 2019 to 2030. In this paper, it is accessed that the dataset of the electricity sector for forecasting purposes from the hydrocarbon development institute of Pakistan "HDIP."The dataset of HDIP is from 1999 to 2019 with different attributes like Electricity Installed Capacity (Hydel Thermal (WAPDA), Thermal (K-Electric), Thermal (IPPs), Nuclear), Energy Consumption by Sector (Domestic, Commercial), Resource Production (Oil, Gas, Coal, Electricity), and Resource Consumption (Oil, Gas, Coal, Electricity). It is visualized and forecasted the energy demand of each attribute until 2030. Predicting overall primary energy demand using machine learning appears to be more accurate than summing up the individual forecasts.
Anais do 14º Simpósio Brasileiro de Automação Inteligente, 2019
The adoption of photovoltaic systems installed in homes and small/medium commercial and industrial buildings has grown considerably in the last decades in Brazil and worldwide. These systems, which are typically linked to primary (medium-voltage) or secondary (low-voltage) distribution networks, are called grid-connected power photovoltaic (PV) systems. Evaluating the return on investment of these systems requires a robust modelling and tools to prognosticate the amount of generated energy. This work, therefore, is based on the application of Artificial Neural Networks (ANN) to predict the daily-produced energy from a PV system installed at the University of São Paulo (USP). Meteorological and performance data were collected from this PV system for one month. This methodology may allow carrying out the efficiency analysis of the device, to plan the system operation and commercialization of the energy obtained strategically. We present several graphs which estimate the quality of the PV system, along with performance metrics, such as the number of epochs, mean squared errors and variances achieved in the simulations. Finally, the validation studies were performed by comparing the error between the experimentally measured data and the estimated data from ANN. The computational tests show that the purpose is interesting to solve the problem within an acceptable percentage error. Resumo: A adoção de sistemas fotovoltaicos instalados em casas e pequenas e médias instalações comerciais e industriais tem crescido consideravelmente nas últimas décadas no Brasil e também internacionalmente. Esses sistemas, que são tipicamente ligados a redes primária (média tensão) ou secundária (baixa tensão) de redes de distribuição, são também chamados de sistemas fotovoltaicos de energia conectados à rede elétrica (SFCR's). Avaliar o retorno de investimento desses sistemas requerem um modelo robusto e ferramentas para diagnosticar a quantidade de energia gerada. Esse trabalho, portanto, baseado na aplicação de Redes Neurais Artificiais (RNAs) prediz a energia gerada no dia de um SFCR instalado na Universidade de São Paulo. Dados de desempenho e meteorológico foram coletados durante um mês. Essa metodologia pode permitir analisar a eficiência desse dispositivo, planejar a operação e comercialização da energia obtida estrategicamente. Diversos gráficos são apresentados que estimam a qualidade do SFCR, junto com métricas de desempenho, tais como número de épocas, erros quadráticos médios e variâncias obtidas nas simulações. Finalmente, os estudos de validação foram executados comparando os erros entre os dados medidos experimentalmente e os estimados pela RNA. Os testes computacionais mostram que o propósito é interessante para resolver o problema dentro de um erro percentual aceitável.
Solar Power Forecasting Methods -A Review
International Journal of Advanced Science and Engineering, 2022
Solar power forecasting is crucial for the purpose of ensuring grid stability and proper grid management. Recent advancements with inside the discipline of solar power forecasting are presented, and the main focus is on the different types of Machine Learning (ML) Techniques used. These ML techniques can solve both the complex and nonlinear data structures. The two types of solar power forecasting are direct and indirect. It entails three models namely: plane of array irradiance, estimating solar irradiance forecast, solar performance. For the purpose of classification of solar power forecasting we take into consideration 3 main parameters such as the Forecast Horizon, Input Parameters and the Forecasting methodology. During the failure of the real-time data acquisition or with inside the case of unavailability of solar power for a new PV plant the concept of Indirect solar power forecasting can be used. According to recent studies models like the hybrid models, deep neural networks take over the conventional methods of the short-term solar forecasting. Data-preparation techniques and various intelligent optimizations enhance the performance accuracy.
Polityka Energetyczna – Energy Policy Journal
The analysis and assessment of the development of solar energy were carried out and it was noted that the production of solar electricity in the world has increased by more than 15% over the last year. In 2020 there are more than 37 countries with a total photovoltaic capacity of more than one GW, and the share of solar energy in total world electricity production was 8.15%. In the regional context, the largest production of electricity by solar energy sources is in Asia (at the expense of India and China) and North America (USA). The study assesses the main factors in the 6 development of solar energy from the standpoint of environmental friendliness and stability of the electricity supply. The problem of the utilization of solar station equipment in the EU and the US is considered. According to the IPCC, IEA, Solar Power Europe, forecasting the development of solar energy in the world is considered. It is proved that the main factor in assessing the economic efficiency of solar energy production is a regional feature due to natural and climatic conditions (intensity of solar radiation). The use of solar generation is auxiliary for the operation of modern electrical networks as long as the efficiency of photovoltaic cells increases by at least 60-65%. Marginal costs of solar energy are minimal in those countries where active state support is provided. The competitiveness of solar energy is relatively low. However, from the standpoint of replacing energy fuel at a cost of USD 10 per 1 Gcal of solar energy saves 10-20 million tons of conventional fuel. Industrial production of solar electricity at modern solar power plants forms a price at the level of USD 250-450 for 1 MWh.
Regression-based method for real-time solar power plant efficiency forecasting
Regression-based method for real-time solar power plant efficiency forecasting, 2024
The importance of this research lies in the growing reliance on solar energy as a key renewable energy source. Solar power plants offer low operational costs, ease of maintenance, and substantial reliability, making them an attractive option for clean energy production. However, the efficiency of these plants can be significantly influenced by external factors such as weather conditions and the physical characteristics of the solar panels. The paper elaborates on various forecasting horizons-ranging from very short-term to long-term-and discusses the suitability of different models like artificial neural networks, time-series forecasting, machine learning, and ensemble methods for these applications. Utilizing data from a solar power plant the study tests several regression models to identify the one with the best forecasting accuracy. The Gradient Boosting Regressor emerged as the most effective model, demonstrating its potential in accurately predicting solar power output. The methodology's success highlights the possibility of integrating solar power plants more efficiently into smart grid systems and optimizing energy management practices. The paper presents a robust method for real-time forecasting of solar power plant efficiency that could significantly benefit energy management and the integration of renewable energy sources into power systems. It opens avenues for further research into improving forecasting techniques and underscores the critical role of accurate prediction models in the advancement of renewable energy technologies.