Forecasting Of Power Generation Research Articles

LSTM Based Forecasting of PV Power for a Second Order Lever Principle Single Axis Solar Tracker

Nowadays solar power generation has significantly improved all over the world. Therefore, the power estimation of photovoltaic (PV) using weather parameters presents the future management of energy utilization in power system planning. This article presents the power forecast of the Second Order Lever Principle Single Axis Solar Tracker (SOLPSAST) system. A deep neural network is developed using Long Short Term Memory (LSTM) and is validated on sunny, cloudy and partially cloudy days. The performance of the proposed LSTM in comparison with Support Vector Machine (SVM) has improved the Mean Absolute Proportion Error (MAPE) forecasts accuracy to 4.29%, 5.16%, and 4.82% for sunny, cloudy and partially cloudy days, respectively. Also, the estimated Mean Relative Error (MRE) value of the LSTM model for sunny, cloudy and partially cloudy days is 3.19%, 4.10%, and 4.02%, respectively. Finally, the forecasted power generation of the SOLPSAST system’s monthly average and annual generation is found to be 2.45 Wh, and 29.44 kWh, respectively.

Transformers-Based Encoder Model for Forecasting Hourly Power Output of Transparent Photovoltaic Module Systems

Solar power generation is usually affected by different meteorological factors, such as solar radiation, cloud cover, rainfall, and temperature. This variability has shown a negative impact on the large-scale integration of solar energy into energy supply systems. For successful integration of solar energy into the electrical grid, it is necessary to predict the accurate power generation by solar panels. In this work, solar power generation forecasting for two types of solar system (non-transparent and transparent panels) was configured by the smart artificial intelligence (AI) modelling. For deep learning models, the dataset obtained from the target value of electricity generation in kWh and other features, such as weather conditions, solar radiance, and insolation. In PV power generation values from non-transparent and transparent solar panels were collected from 1 January to 31 December 2021 with an hourly interval. To prove the efficiency of the proposed model, several deep learning approaches RNN models, such as LSTM, GRU, and transformers models, were implemented. Transformers model for forecasting power generation expressed the best model for non-transparent and transparent solar panels with lower error rates for MAE 0.05 and 0.04, and RMSE 0.24 and 0.21, respectively. The proposed model showed efficient performance and proved effective in forecasting time-series data.

On the value of expert knowledge in estimation and forecasting of solar photovoltaic power generation

Reliable estimates and forecasts of Photovoltaic (PV) power output form a fundamental basis to support its large-scale integration. This is recognized in literature, where a growing amount of studies deal with the development of PV power estimation and forecasting models. In particular, machine learning techniques received significant attention in the past decade. Yet, the importance of predictor variables are consistently ignored in such developments and as a result those models fail to acknowledge the value of including physics-based models. In this study we quantify the value of predictor variables for PV power estimation and forecasting, assess deficiencies in estimation and forecasting models, and introduce a number of pre-processing steps to improve the overall estimation or forecasting performance. To this end, we use common physical models to create so-called expert variables and test their impact on the performance of single-point and probabilistic models. In addition, we investigate the optimal selection of predictor variables for PV power estimation and forecasting. By means of a sensitivity analysis, the paper shows how the value of expert variables is affected by the tilt angle of the PV system. To allow for a deeper insight into the importance of predictor variables, two case studies in different climate regions are considered in the numerical evaluation.

Impact of overloading of photovoltaic arrays on the evaluation of photovoltaic power generation forecasts

• Increasing inverter loading ratio decreases mean absolute percent error in forecast. • The benefit is most apparent on clear days. • PV module overloading reduces the effect of over-forecasting of PV output. • PV module overloading affects the value of PV output forecasting technology. Photovoltaic power generation (PV output) forecast is based on solar irradiance forecasts; therefore, an increase in overloading of PV arrays may affect errors in the PV output forecast. To study the impact of overloading of PV arrays on evaluation of PV output forecasts, we compared the estimated and forecasted values of PV output under various overloading conditions and investigated the characteristics of forecast errors resulting from overloading. We used data of actual estimated PV output and forecasted PV output for 2016 to calculate the estimated and forecasted values of overloaded P PV output. Our findings indicate that compared with over-forecasted errors, errors due to under-forecasting undergo higher reduction in both time and magnitude with increasing inverter loading ratio. However, contrary to the initial assumption, the effect of PV module overloading was smaller than expected, which could be attributed to the smoothing effect of PV systems in a large area. PV module and array prices have been decreasing in the long term, and the number of overloaded PV systems is expected to increase worldwide. This study provides useful information for the stable operation of power systems including overloaded PV systems and for improving the economics of PV systems.

Intelligent digital twin – machine learning system for real-time wind turbine wind speed and power generation forecasting

Wind power is a key pillar in efforts to decarbonise energy production. However, variability in wind speed and resultant wind turbine power generation poses a challenge for power grid integration. Digital Twin (DT) technology provides intelligent service systems, combining real-time monitoring, predictive capabilities and communication technologies. Current DT research for wind turbine power generation has focused on providing wind speed and power generation predictions reliant on Supervisory Control and Data Acquisition (SCADA) sensors, with predictions often limited to the timeframe of datasets. This research looks to expand on this, utilising a novel framework for an intelligent DT system powered by k-Nearest Neighbour (kNN) regression models to upscale live wind speed forecasts to higher wind turbine hub-height and then forecast power generation. As there is no live link to a wind turbine, the framework is referred to as a “Simulated Digital Twin” (SimTwin). 2019-2020 SCADA and wind speed data are used to evaluate this, demonstrating that the method provides suitable predictions. Furthermore, full deployment of the SimTwin framework is demonstrated using live wind speed forecasts. This may prove useful for operators by reducing reliance on SCADA systems and provides a research and development tool where live data is limited.

Microgrid control for renewable energy sources based on deep reinforcement learning and numerical optimization approaches

Optimal scheduling of battery energy storage system plays crucial part in dis- tributed energy system. As a data driven method, deep reinforcement learning does not require system knowledge of dynamic system, present optimal solution for nonlinear optimization problem. In this research, financial cost of energy con- sumption reduced by scheduling battery energy using deep reinforcement learning method (RL). Reinforcement learning can adapt to equipment parameter changes and noise in the data, while mixed-integer linear programming (MILP) requires high accuracy in forecasting power generation and demand, accurate equipment parameters to achieve good performance, and high computational cost for large- scale industrial applications. Based on this, it can be assumed that deep RL based solution is capable of outperform classic deterministic optimization model MILP. This study compares four state-of-the-art RL algorithms for the battery power plant control problem: PPO, A2C, SAC, TD3. According to the simulation results, TD3 shows the best results, outperforming MILP by 5% in cost savings, and the time to solve the problem is reduced by about a factor of three.

Deterministic and Probabilistic Wind Power Forecasts by Considering Various Atmospheric Models and Feature Engineering Approaches

This study proposed a model for deterministic and probabilistic wind power generation forecasting and its corresponding procedures. The main contents include numerical weather prediction (NWP) systems, data preprocessing techniques, and forecasting models that use artificial intelligence methods. NWP wind speeds generated by the Central Weather Bureau (CWB) of Taiwan based on three atmospheric models, namely deterministic weather research and forecasting (WRFD), radar weather research and forecasting (RWRF), and WRF-based ensemble prediction system (WEPS), were used as model inputs. In terms of data preprocessing, the NWP wind speeds were corrected based on the height of the wind turbines, and principal components analysis (PCA) and empirical mode decomposition (EMD) were also evaluated for their feature extraction performance. Artificial neural network (ANN), long short-term memory (LSTM), gated recurrent unit (GRU), and eXtreme gradient boosting (XGBoost) were the forecast models used to predict wind power generation. The forecast results demonstrated that the use of the XGBoost model in conjunction with both PCA and EMD data preprocessing achieved the most accurate forecasting. The average forecasting errors (i.e., normalized root mean square error (NRMSE) and mean absolute percentage error (MAPE)) were 5.43% and 3.30% for one-hour-ahead and 9.78% and 6.83% for one-day-ahead, respectively. The empirical data collected at a wind farm in Taiwan verified the accuracy of the proposed method. Thus, the importance of model selection, NWP, and data preprocessing is ultimately self-evident.

Evaluation Metrics for Wind Power Forecasts: A Comprehensive Review and Statistical Analysis of Errors

Power generation forecasts for wind farms, especially with a short-term horizon, have been extensively researched due to the growing share of wind farms in total power generation. Detailed forecasts are necessary for the optimization of power systems of various sizes. This review and analytical paper is largely focused on a statistical analysis of forecasting errors based on more than one hundred papers on wind generation forecasts. Factors affecting the magnitude of forecasting errors are presented and discussed. Normalized root mean squared error (nRMSE) and normalized mean absolute error (nMAE) have been selected as the main error metrics considered here. A new and unique error dispersion factor (EDF) is proposed, being the ratio of nRMSE to nMAE. The variability of EDF depending on selected factors (size of wind farm, forecasting horizons, and class of forecasting method) has been examined. This is unique and original research, a novelty in studies on errors of power generation forecasts in wind farms. In addition, extensive quantitative and qualitative analyses have been conducted to assess the magnitude of forecasting error depending on selected factors (such as forecasting horizon, wind farm size, and a class of the forecasting method). Based on these analyses and a review of more than one hundred papers, a unique set of recommendations on the preferred content of papers addressing wind farm generation forecasts has been developed. These recommendations would make it possible to conduct very precise benchmarking meta-analyses of forecasting studies described in research papers and to develop valuable general conclusions concerning the analyzed phenomena.

Evaluation of Weather Information for Short-Term Wind Power Forecasting with Various Types of Models

The rising share of renewable energy in the energy mix brings with it new challenges such as power curtailment and lack of reliable large-scale energy grid. The forecasting of wind power generation for provision of flexibility, defined as the ability to absorb and manage fluctuations in the demand and supply by storing energy at times of surplus and releasing it when needed, is important. In this study, short-term forecasting models of wind power generation were developed using the conventional time-series method and hybrid models using support vector regression (SVR) based on rolling origin recalibration. For the application of the methodology, the meteorological database from Korea Meteorological Administration and actual operating data of a wind power turbine (2.3 MW) from 1 January to 31 December 2015 were used. The results showed that the proposed SVR model has higher forecasting accuracy than the existing time-series methods. In addition, the conventional time-series model has high accuracy under proper curation of wind turbine operation data. Therefore, the analysis results reveal that data curation and weather information are as important as the model for wind power forecasting.

Wind Power Generation Scheduling Accuracy in Europe: An Overview of ENTSO-E Countries

Despite the rapid spread of the use of wind energy to generate electricity, harnessing this energy source remains a great challenge due to its stochastic nature. One way of dealing with this is to prepare accurate wind power forecasts. This paper explored the accuracy of day-ahead and intraday scheduling of energy generation of the onshore and offshore wind farms of the member countries of the European Network of Transmission System Operators (ENTSO-E) in the period from 2013 to 2021. The precision of the scheduling activities showed a varying picture: the onshore wind farms of Germany, Spain, France, and Sweden produced more precise forecasts than others, with annual downward and upward regulatory needs between 0.8% and 14.4%, and from 0.8% to 6.5%, of the yearly energy generation, respectively. In certain countries, however, the forecasts were less accurate, with discrepancies exceeding 41% for downward and 132% for upward regulation. As for offshore wind farms, the annual downward and upward regulatory needs ranged between 0.9% and 61.7%, and from 1.3% to 44.1%, respectively, with Germany and Denmark producing the most accurate schedules. The innovative novelty and practical contributions of this study are that it determines and presents information related to the accuracy of the day-ahead and intraday wind power generation forecasting of the ENTSO-E countries, which is of practical relevance to the transmission system operators (TSOs), the main actors in the energy market and the decision-makers, too. This information may also help investors who invest in onshore and offshore wind farms with the economic aspects, and it may also greatly contribute to the market-related development of the management systems of energy storage solutions related to these technologies.

Towards intelligent building energy management: AI-based framework for power consumption and generation forecasting

Due to global warming and climate changes, buildings including residential and commercial are significant contributors to energy consumption. To this end, net zero energy building (NZEB) has become a progressively popular concept where the annual sum of power generation and consumption is zero. However, occasionally, there exists a mismatching between demand and supply in NZEB due to consumer behaviour and weather conditions disturbing the overall management of the smart grid. To overcome such hurdles, precise prediction of energy usage is a key strategy among others. Therefore, in this study, an efficient hybrid AI-based framework is proposed for accurate forecasting of power consumption and generation that is mainly composed of three steps. Initially, the optimal pre-processing procedure is applied for data refinement. Next, for the spatiotemporal features, a convolutional long short-term memory (ConvLSTM) is used that learns discriminative patterns from the past power knowledge, followed by a bidirectional gated recurrent unit (BDGRU) that extracts on temporal aspects. Eventually, feature descriptors are then passed to multilayer perceptron layers to perform the forecasting. After extensive experiments over the household and photovoltaic energy data, we concluded that our model substantially reduced the errors of 0.012 and 0.045 in terms of mean square error (MSE) on hourly data as compared to the recent state-of-the-art techniques (SOTA).

A hybrid framework for forecasting power generation of multiple renewable energy sources

The accurate power generation forecast of multiple renewable energy sources is significant for the power scheduling of renewable energy systems. However, previous studies focused more on the prediction of a single energy source, ignoring the relationship among different energy sources, and failing to predict accurate power generation for all energy sources simultaneously. This paper proposes a hybrid framework for the power generation forecast of multiple renewable energy sources to overcome deficiencies. A Convolutional Neural Network (CNN) is developed to extract the local correlations among multiple energy sources, the Attention-based Long Short-Term Memory (A-LSTM) network is developed to capture the nonlinear time-series characteristics of weather conditions and individual energy, and the Auto-Regression model is applied to extract the linear time-series characteristics of each energy source. The accuracy and practicality of the proposed method are verified by taking a renewable energy system as an example. The results show that the hybrid framework is more accurate than other advanced models, such as artificial neural networks and decision trees. Mean absolute errors of the proposed method are reduced by 13.4%, 22.9%, and 27.1% for solar PV, solar thermal, and wind power compared with A-LSTM. The sensitivity analysis has been conducted to test the effectiveness of each component of the proposed hybrid framework to prove the significance of energy correlation patterns with higher accuracy and stability compared with the other two patterns.

Solar Power Forecasting Using CNN-LSTM Hybrid Model

Photovoltaic (PV) technology converts solar energy into electrical energy, and the PV industry is an essential renewable energy industry. However, the amount of power generated through PV systems is closely related to unpredictable and uncontrollable environmental factors such as solar radiation, temperature, humidity, cloud cover, and wind speed. Particularly, changes in temperature and solar radiation can substantially affect power generation, causing a sudden surplus or reduction in the power output. Nevertheless, accurately predicting the energy produced by PV power generation systems is crucial. This paper proposes a hybrid model comprising a convolutional neural network (CNN) and long short-term memory (LSTM) for stable power generation forecasting. The CNN classifies weather conditions, while the LSTM learns power generation patterns based on the weather conditions. The proposed model was trained and tested using the PV power output data from a power plant in Busan, Korea. Quantitative and qualitative evaluations were performed to verify the performance of the model. The proposed model achieved a mean absolute percentage error of 4.58 on a sunny day and 7.06 on a cloudy day in the quantitative evaluation. The experimental results suggest that precise power generation forecasting is possible using the proposed model according to instantaneous changes in power generation patterns. Moreover, the proposed model can help optimize PV power plant operations.

An Integrated Missing-Data Tolerant Model for Probabilistic PV Power Generation Forecasting

Accurate solar photovoltaic (PV) generation forecast is critical to the reliable and economic operation of a modern power system. In practice, due to various faulty issues in the sensor, communication, or database system, the historical and online measurement data may not be always complete, and the missing data could dramatically degrade the forecasting model's accuracy. To solve this problem, this paper proposes an integrated missing-data tolerant model for probabilistic PV power generation forecasting. Taking historical PV generations as input, this model is based on a recursive long short-term memory network (Rec-LSTM), which can provide multi-step ahead forecasting of the probability distribution of PV generation. The unobserved input data will be imputed recursively based on the model output at the previous time step. During the training process, the imputations and forecasting values are iteratively updated by the negative log-likelihood loss function. As a salient advantage, this method can deal with data missing scenarios at both offline and online stages. Numerical experiments are conducted on two one-year datasets from Australia and Singapore, respectively. Probabilistic forecasting for both large-scale and small-scale building-level PV power generation is tested at the time resolution of 15 mins. Testing results show the proposed method can achieve superior probabilistic prediction accuracy as well as strong robustness under various data missing scenarios, compared to other state-of-the-art methods.

Enhancing wind power forecasting from meteorological parameters using machine learning models

Short-term wind power forecasting is crucial for power system stability, dispatching, and cost control. Wind energy has the potential to be a viable source of renewable energy. Wind power generation forecasting is vital for resolving the supply and demand challenges of the smart grid. Moreover, one of the most problematic aspects of wind power is its high fluctuation and intermittent nature, which makes forecasting difficult. The goal of this research is to create machine learning models that can properly estimate wind power production. Significantly, the major contributions of this work are highlighted in the following significant elements. First, a data analysis framework for visualizing the gathered dataset from the SCADA system is presented. Second, for forecasting wind power time-series dataset values, we examine the predicting performance of various machine learning models using various statistical indices. The experimental findings demonstrate that with a minor reconstruction error, the proposed forecast approaches can minimize the complexity of the forecasting. Furthermore, in terms of forecast accuracy, a gradient boosting regression model outperforms other benchmark models. According to the analysis, our methodology might be applied in real-world circumstances to assist the management group in regulating the power provided by wind turbines.

A fuzzy intelligent forecasting system based on combined fuzzification strategy and improved optimization algorithm for renewable energy power generation

• Develop a novel hybrid forecasting system focusing on small sample forecasting. • Forecast the renewable energy power generation of different regions in the world. • Improve the fuzzification ability by combined strategy and hesitant fuzzy sets. • Search the optimal aggregated weights based on improved optimization algorithm. • Introduce the chaotic map into the equilibrium optimization algorithm. Renewable energy power generation has significant potential to electricity supply sector with great sense to greenhouse gas control. Thus, it is vital to develop an effective forecasting model for renewable energy power generation forecasting which can provide a reference and basis for power generation planning and the energy strategic deployment. However, influenced by the complex data characteristic and sample size limitation, the application of some traditional forecasting models is restricted with poor forecasting performance. In this paper, a novel fuzzy time series forecasting based on combined fuzzification strategy and improved optimization algorithm is proposed for renewable energy power generation forecasting. The hesitant fuzzy sets are applied to deal with the combined fuzzification strategy and the improved optimization algorithm is developed to optimize the aggregate weights of the hesitant fuzzy sets. The experimental analysis and discussion all demonstrated the excellent performance of the proposed forecasting system in small sample forecasting for renewable energy power generation forecasting.

A Power Forecasting Method for Ultra-Short-Term Photovoltaic Power Generation Using Transformer Model

The volatility of solar energy, geographic location, and weather factors continues to affect the stability of photovoltaic power generation, reliable and accurate photovoltaic power prediction methods not only effectively reduce the operating cost of the photovoltaic system but also provide reliable data support for the energy scheduling of the light storage microgrid, improve the stability of the photovoltaic system, and provide important help for the optimization operation of the photovoltaic system. Therefore, it is an important study to find reliable photovoltaic power prediction methods. In recent years, researchers have improved the accuracy of photovoltaic power generation forecasting by using deep learning models. Compared with the traditional neural network, the Transformer model can better learn the relationship between weather features and has good stability and applicability. Therefore, in this paper, the transformer model is used for predicting ultra-short-term photovoltaic power generation, and the photovoltaic power generation data and weather data in Hebei are selected. In the experiment, the prediction result of the transformer model was compared to the GRU and DNN models to show that the transformer model has better predictive ability and stability. Experimental results demonstrated that the proposed Transformer model outperforms the GRU model and DNN model by a difference of about 0.04 kW and 0.047 kW in the MSE value, and 22.0% and 29.1% of the MAPE error. In addition, the public DC competition dataset is selected for control experiments to demonstrate the general applicability of the transformer model for PV power prediction in different regions.

SSA-LSTM: Short-Term Photovoltaic Power Prediction Based on Feature Matching

To reduce the impact of volatility on photovoltaic (PV) power generation forecasting and achieve improved forecasting accuracy, this article provides an in-depth analysis of the characteristics of PV power outputs under typical weather conditions. The trend of PV power generation and the similarity between simultaneous outputs are found, and a hybrid prediction model based on feature matching, singular spectrum analysis (SSA) and a long short-term memory (LSTM) network is proposed. In this paper, correlation analysis is used to verify the trend of PV power generation; the similarity between forecasting days and historical meteorological data is calculated through grey relation analysis; and similar generated PV power levels are searched for phase feature matching. The input time series is decomposed by singular spectrum analysis; the trend component, oscillation component and noise component are extracted; and principal component analysis and reconstruction are carried out on each component. Then, an LSTM network prediction model is established for the reconstructed subsequences, and the external feature input is controlled to compare the obtained prediction results. Finally, the model performance is evaluated through the data of a PV power plant in a certain area. The experimental results prove that the SSA-LSTM model has the best prediction performance.

Medium-term forecasting of power generation by hydropower plants in isolated power systems under climate change

Reliable operation of power systems (PS), including those with a significant share of hydropower plants (HPPs) in the energy balance, largely depends on the accuracy of forecasting power generation. The importance of power generation forecasts increases with the development of renewable power generation, which is stochastic by nature. Those kinds of tasks are complicated by the lack of reliable information on metrological data and estimated energy consumption, which is also stochastic. In the medium-term forecasting (MTF) of power generation by HPPs, the seasonality of changes in flow and inflow of water should be taken into account, which significantly affects the reserves and regulatory capabilities of the power system as a whole. This work discusses the problem of constructing a model for MTF of power generation HPP in isolated power systems (IPS), taking into account such atmospheric parameters as air temperature, wind speed and humidity. To address constant climatic changes, this paper suggests implementing machine learning models. The proposed approach is characterized by a high degree of autonomy and learning automation. The paper provides a comparative study of the machine learning models such as polynomial model with Tikhonov’s regularization (LR), k-nearest neighbors (kNN), multilayer perceptron (MLP), ensembles of decision trees, adaptive boosting of linear models (ABLR), etc. Computational experiments have shown that the machine learning approach yields the results of sufficient quality, which allows to use them for forecasting of power generation HPP in isolated power systems under conditions of climate change. The Adaptive Boosting Linear Regression model is the simplest and most reliable machine learning model that has proven itself well in the tasks with a relatively small amount of training samples.

Feasibility Study on the Influence of Data Partition Strategies on Ensemble Deep Learning: The Case of Forecasting Power Generation in South Korea

Ensemble deep learning methods have demonstrated significant improvements in forecasting the solar panel power generation using historical time-series data. Although many studies have used ensemble deep learning methods with various data partitioning strategies, most have only focused on improving the predictive methods by associating several different models or combining hyperparameters and interactions. In this study, we contend that we can enhance the precision of power generation forecasting by identifying a suitable data partition strategy and establishing the ideal number of partitions and subset sizes. Thus, we propose a feasibility study of the influence of data partition strategies on ensemble deep learning. We selected five time-series data partitioning strategies—window, shuffle, pyramid, vertical, and seasonal—that allow us to identify different characteristics and features in the time-series data. We conducted various experiments on two sources of solar panel datasets collected in Seoul and Gyeongju, South Korea. Additionally, LSTM-based bagging ensemble models were applied to combine the advantages of several single LSTM models. The experimental results reveal that the data partition strategies positively influence the forecasting of power generation. Specifically, the results demonstrate that ensemble models with data partition strategies outperform single LSTM models by approximately 4–11% in terms of the coefficient of determination (R2) score.