PV Power Forecasting with Holt-Winters Method
04 Mar 2020-pp 1-4
TL;DR: In this paper, a short-term PV power forecasting technique based on Holt-Winters method is proposed, which can be applied in energy management and electrical power smoothing applications.
Abstract: Due to global warming issue and growing of energy demand, photovoltaic (PV) power plant is a desirable alternative to electrical power generations. An accurate PV power forecasting is essential to alleviate its negative impacts and to improve its performance. This paper proposed a short-term PV power forecasting technique based on Holt-Winters method. Its property and principle are presented. There are two main parameters, i.e. the number of days (N) and the weight index (α). The values of these two parameters affect the forecasting accuracy, and a set of optimal values for both parameters used for all season is proposed. A real PV generated power profile is collected from the eastern part of Thailand and used in this study. The performance of the proposed forecasting technique is carried out by MATLAB program. The simulation results indicate that the proposed method provides acceptable short-term forecasting results. The proposed technique can be applied in energy management and electrical power smoothing applications.
Citations
More filters
TL;DR: This work was supported by funding from King Abdullah University of Science and Technology (KAUST), Office of Sponsored Research (OSR) under Award No: OSR-2019-CRG7-3800.
Abstract: 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.
59 citations
18 Sep 2022
TL;DR: In this article , the authors presented the development of long shortterm memory (LSTM) and gated recurrent unit (GRU) models to predict photovoltaic energy in an isolated area of Ecuador.
Abstract: In Ecuador, electricity generation is mainly covered by renewable energy sources that feed the National Interconnected System (NIS). Its economical price means that the installation of systems based on non-conventional renewable energy sources does not represent a benefit for the user. However, rural communities isolated from the NIS do not have electricity services. Due to this, isolated systems become an effective option to supply electricity to these communities. In this regard, the Galapagos Islands have unique biodiversity in the world. Since they are not connected to the NIS, their primary energy sources are based on biogas obtained from fossil fuels, with their negative consequences despite the great potential of solar resources. Hence the need to use non-conventional renewable energy sources that covers the energy demand and do not affect the biodiversity there. Photovoltaic energy forecasting is an essential step in the installation of photovoltaic systems. Prediction models based on deep learning (DL) techniques can obtain a high degree of accuracy in energy prediction tasks. For this reason, this work presents the development of long short-term memory (LSTM) and gated recurrent unit (GRU) models to predict photovoltaic energy in an isolated area of Ecuador. The results highlight the performance of both methods through the achieved short-term prediction.
1 citations
17 Oct 2022
TL;DR: In this paper , the authors compare Holt-Winters and Seasonal Variation forecasting methods applied to two rural locations in Ecuador to achieve a short-term forecast of photovoltaic generation.
Abstract: Due to the shortage of electric power in isolated rural areas of Ecuador, implementing a photovoltaic power generation system is an optimal, viable, and sustainable alternative that can reduce the gap in electric power coverage in the nation. However, since the geographical location of a possible PV system implementation directly affects its optimal performance, it is essential to propose generation forecasting algorithms that use actual data from the place under study. For this reason, this paper compares Holt-Winters and Seasonal Variation forecasting methods applied to two rural locations in Ecuador. These methods use historical records of meteorological conditions such as solar irradiation and ambient temperature to achieve a short-term forecast of photovoltaic generation. Percentage errors obtained in the forecast power and the Pearson correlation coefficient are calculated as performance measures. The results show that although both algorithms have similar characteristics when getting the photovoltaic generation prediction, the Holt-Winters method presents improved behavior.
1 citations
TL;DR: In this paper , a comparison of performance losses of two silicon PV technologies installed on the rooftop of the Higher School of Technology in Laâyoune-Morocco is presented.
Abstract: Received Jun 3, 2022 Revised Nov 4, 2022 Accepted Nov 10, 2022 The prediction of performance of photovoltaic technologies is crucial, not only to improve the reliability and durability of these technologies but also to increase the confidence of investors and consumers in them. The accurate calculation of the degradation rate DR (%) in real operating conditions under specific climatic stresses is, therefore, paramount. The present study provides a comparison of performance losses of two silicon PV technologies installed on the rooftop of the Higher School of Technology in Laâyoune-Morocco. The two systems are a polycristalline array (pc-Si: 1.82 kWp) and an amorphous array (a-Si: 1.55 KWp), which are grid connected. In the light of related performance gathered over three-year, the degradation rates of the two systems were estimated using four statistical methods under the open-source software R. The techniques engaged in this paper are: classical seasonal decomposition (CSD), holt-winters (HW), autoregressive integrated moving average (ARIMA), and seasonal and trend decomposition by LOESS (STL). The results obtained using those methods show that DR(%) varies between 0.39% and 0.99% for pc-Si and between 0.29% and 0.64% for a-Si. The analysis of degradation accuracy shows that STL and CSD techniques provide results with high accuracy than ARIMA and HW for the two systems. The present study adds to knowledge on PV degradation under the subtropical desert climate of Laâyoune.
References
More filters
TL;DR: In this article, a comparison of univariate methods for forecasting up to a day-ahead of electricity demand data from ten European countries is performed using intraday electricity demand from 10 European countries as the basis of an empirical comparison.
Abstract: This paper uses intraday electricity demand data from ten European countries as the basis of an empirical comparison of univariate methods for prediction up to a day-ahead. A notable feature of the time series is the presence of both an in-traweek and an intraday seasonal cycle. The forecasting methods considered in the study include: ARIMA modeling, periodic AR modeling, an extension for double seasonality of Holt-Winters exponential smoothing, a recently proposed alternative exponential smoothing formulation, and a method based on the principal component analysis (PCA) of the daily demand profiles. Our results show a similar ranking of methods across the 10 load series. The results were disappointing for the new alternative exponential smoothing method and for the periodic AR model. The ARIMA and PCA methods performed well, but the method that consistently performed the best was the double seasonal Holt-Winters exponential smoothing method.
532 citations
"PV Power Forecasting with Holt-Wint..." refers methods in this paper
...Previously, Holt-Winters forecasting method has been applied in many applications, such as short-term load forecasting or mobile network traffic prediction [10]-[11]....
[...]
TL;DR: A weather-based hybrid method for 1-day ahead hourly forecasting of PV power output is presented and achieves better prediction accuracy than the simple SVR and traditional ANN methods.
Abstract: To improve real-time control performance and reduce possible negative impacts of photovoltaic (PV) systems, an accurate forecasting of PV output is required, which is an important function in the operation of an energy management system (EMS) for distributed energy resources. In this paper, a weather-based hybrid method for 1-day ahead hourly forecasting of PV power output is presented. The proposed approach comprises classification, training, and forecasting stages. In the classification stage, the self-organizing map (SOM) and learning vector quantization (LVQ) networks are used to classify the collected historical data of PV power output. The training stage employs the support vector regression (SVR) to train the input/output data sets for temperature, probability of precipitation, and solar irradiance of defined similar hours. In the forecasting stage, the fuzzy inference method is used to select an adequate trained model for accurate forecast, according to the weather information collected from Taiwan Central Weather Bureau (TCWB). The proposed approach is applied to a practical PV power generation system. Numerical results show that the proposed approach achieves better prediction accuracy than the simple SVR and traditional ANN methods.
390 citations
"PV Power Forecasting with Holt-Wint..." refers methods in this paper
...This research utilizes the following evaluation indices presented in [12] to evaluate the forecasting errors:...
[...]
TL;DR: The design of an energy management system capable of forecasting photovoltaic power production and optimizing power flows between PV system, grid, and battery electric vehicles (BEVs) at the workplace is presented.
Abstract: This paper presents the design of an energy management system (EMS) capable of forecasting photovoltaic (PV) power production and optimizing power flows between PV system, grid, and battery electric vehicles (BEVs) at the workplace. The aim is to minimize charging cost while reducing energy demand from the grid by increasing PV self-consumption and consequently increasing sustainability of the BEV fleet. The developed EMS consists of two components: An autoregressive integrated moving average model to predict PV power production and a mixed-integer linear programming framework that optimally allocates power to minimize charging cost. The results show that the developed EMS is able to reduce charging cost significantly, while increasing PV self-consumption and reducing energy consumption from the grid. Furthermore, during a case study analogous to one repeatedly considered in the literature, i.e., dynamic purchase tariff and dynamic feed-in tariff, the EMS reduces charging cost by 118.44 $\%$ and 427.45 $\%$ in case of one and two charging points, respectively, when compared to an uncontrolled charging policy.
154 citations
01 Sep 2007
TL;DR: A technique for evaluation of future radio traffic of circuit switched services in Erlang for near-term outlook using Holt-Winter's exponential smoothing and how to classify traffic data for prediction is presented.
Abstract: This paper presents a technique for evaluation of future radio traffic of circuit switched services in Erlang for near-term outlook using Holt-Winter's exponential smoothing. The proposed traffic prediction technique relies on analysis of traffic data on cells. We propose how to classify traffic data for prediction. The result from a trial in history data of commercial GSM/GPRS network is presented. It shows that predicted traffic is good fitted with original one within our classification. The technique should be used in automated overload warning and express estimation of future traffic for capacity planning.
94 citations
"PV Power Forecasting with Holt-Wint..." refers methods in this paper
...Previously, Holt-Winters forecasting method has been applied in many applications, such as short-term load forecasting or mobile network traffic prediction [10]-[11]....
[...]
TL;DR: A methodology for optimal operation of a smart grid to minimize the interconnection point power flow fluctuation is presented and it is possible to reduce the electric power consumption and the cost of electricity.
Abstract: From the perspective of global warming mitigation and depletion of energy resources, renewable energy such as wind generation (WG) and photovoltaic generation (PV) are getting attention in distribution systems. Additionally, all electric apartment houses or residence such as dc smart houses are increasing. However, due to the fluctuating power from renewable energy sources and loads, supply-demand balancing of power system becomes problematic. The smart grid is a solution to this problem. This paper presents a methodology for optimal operation of a smart grid to minimize the interconnection point power flow fluctuation. To achieve the proposed optimal operation, we use distributed controllable loads such as battery and heat pump. By minimizing the interconnection point power flow fluctuation, it is possible to reduce the electric power consumption and the cost of electricity. This system consists of a photovoltaic generator, heat pump, battery, solar collector, and load. To verify the effectiveness of the proposed system, results are used in simulation presented.
88 citations
"PV Power Forecasting with Holt-Wint..." refers background or methods in this paper
...PV power is forecasted using fuzzy logic in [7] with humidity and amount of cloud as input parameters....
[...]
...Those include artificial neural network (ANN) [6], fuzzy logic [7], stacking-support vector machine (stacking-SVM) [8] and seasonal autoregressive integrated moving average (SARIMA) [9]....
[...]