Calculation of solar irradiation prediction intervals combining volatility and kernel density estimates
Introduction
Solar power generation has been steadily increasing worldwide as a response to environmental concerns. Unfortunately, the integration of solar energy into the energy mix of a country brings new challenges. The main problem is due to the variability of the solar energy, which is not available “on demand”. Therefore, reasonably accurate forecasts are required to make this kind of energy economically viable. Depending on the objective, different forecasting horizons are needed, for instance, long-term forecasts are useful for locating potential solar power plants, energy resource planning and scheduling employs mid-term (up to 48 h) solar forecasts, whereas intraday forecasts are required for load following and predispatch [20]. Kraas et al. [21] economically quantified the impact of forecasting errors in the Spanish electricity system for a concentrating solar power plant. In Spain, forecasts of production have to be provided to the transmission system operator (TSO). In case of deviations from the scheduled production, the TSO applies a cost penalty. If the forecast is higher than the real production, the TSO charges falling penalties, and conversely, charges rising penalties for a production dispatch above the forecasted value. In that reference, forecasting improvement may significantly reduce penalty charges by 47.6% compared to the simple persistence forecasts.
In general terms, most of the published literature on solar energy forecasting is based on the application of different techniques in order to provide point forecasts. Among the diverse techniques provided, the selection of the best technique is usually based on choosing the one with the lowest forecast error. Nevertheless, the need of the users of such forecast (or stakeholders) require more information apart from the point forecast. In particular, they require both uncertainty and variability forecasts (chapter 14 [20]), that they can be even more useful than typical point forecasts [10]. In this context, point forecasts and its associated uncertainty are usually given as hourly-average time series, whereas irradiance variability or ramp events are measured in an intra-hour time scale, for example, minutes. Note that this difference between variability and uncertainty might not be unanimous in other research disciplines out of the solar energy literature. For instance, in supply chain applications, variability is commonly employed as a measure of uncertainty. Nonetheless, in this work we follow the distinction made in (chapter 14, [20]).
The aim of this work is twofold. Firstly, to bridge the gap in the solar energy forecasting literature by focusing on uncertainty forecasting, which has remained overlooked in comparison with point forecasting. Secondly, to use such uncertainty forecasts to compute prediction intervals on the basis of a novel methodology that combines them in an optimal manner.
Essentially, uncertainty can be quantified by the standard deviation of the forecast error, also known as volatility in finance terms, and this is assumed to be homoskedastic, independent and normally distributed. However, since the solar irradiance time series involves very complex processes, it is expected that those assumptions may be violated. Therefore, this work explores different models depending on the assumption that might not be fulfilled. For instance, if the forecast error is not normal a potential solution is to use Kernel Density estimates [31]. On the other hand, if the forecast error is neither homoskedastic nor independent, then, volatility estimators as either Generalized Autoregressive Conditional Heteroskedastic (GARCH) [4] or exponential smoothing models [34] can be employed. Since it is also possible that any of the aforementioned assumptions may be fulfilled, the novel methodology proposed intend to compute prediction intervals by combining Kernel Density estimates and volatility models. Furthermore, the combination suggested is optimal in the sense that maximizes the conditional coverage Christoffersen test p-value [9].
The results show that such a combination provides a robust performance by achieving a compromise between prediction interval coverage and average interval width.
In order to illustrate the methodology employed, we are going to focus on one-step-ahead uncertainty forecasts obtained from Global Horizontal Irradiation (GHI) data that is crucial for photovoltaic generators. Note that this study can be extended in a straightforward manner to analyze the Direct Normal Irradiation (DNI) that is more relevant for Concentrated Solar Power applications.
The rest of the paper is organised as follows: Section 2 reviews the literature on prediction intervals and it describes the models employed in this article. Section 3 describes the case study dataset. Section 4 lays out the experimental setup and the discussion of initial results. Section 5 defines a new approach to compute prediction intervals based on combining previous models, and finally, Section 6 presents the concluding remarks.
Section snippets
Prediction intervals
There are two principal streams to compute prediction intervals. In the first place, a theoretical prediction interval can be calculated based on a forecasting model that assumes to reproduce the data correctly and that the forecast error follows a determined distribution [8]. Since the model is assumed to be specified correctly, the forecasts are unbiased and the forecast errors have zero mean and constant variance, which is function of a certain forecasting horizon and the model parameters
Case study data
Solar irradiance data have been collected by the Spanish Institute for Concentration Photovoltaics Systems (ISFOC), located in Ciudad Real in the region of Castilla-La Mancha in Spain (at 38.67 °N, 4.15 °W, 687 m). Minute-by-minute solar irradiance measurements have been recorded using pyranometers and pyrheliometers, which comply with the international standards of Baseline Surface Radiation Network (BSRN) [24].
Global Horizontal Irradiance (GHI) data have been provided. GHI is the total solar
Experimental setup
The data (8760 observations) have been split down in two parts of the same size approximately. The first part (4392 observations) has been used to estimate the parameters of the ARIMA model (in-sample data) and the GARCH parameters. Once the forecasts have been calculated, we have removed the nights by eliminating observations with an elevation angle lower than 15°. Note that we have not removed the nights in the calculation of the ARIMA-GARCH model in order to use the same model than [28] with
Combination of prediction intervals
Since the non-parametric approach gives a higher hit rate and the parametric approaches, specially the SES approach, tend to be more independent, another alternative is to combine both methods to compute a prediction interval. Actually, Christoffersen [9] pointed out that combining non-parametric error distribution with time-varying variance estimators is likely to present a favorable alternative. In this section we will test such an option.
Given that the non-parametric and SES were the methods
Conclusions
Forecasts of solar irradiation are required to incorporate the electricity generated into the grid. Recently, a large variety of approaches have proliferated so as to provide point forecasts. Nonetheless, the literature about the uncertainty associated to such forecasts is scarce. This work examines the aforementioned uncertainty through prediction intervals computed by means of non-parametric kernel density estimates and parametric approaches based on volatility forecasting models. The results
Acknowledgment
The author is grateful to Alberto Martín and ISFOC for kindly providing the data used in this paper and he is also grateful to Diego Pedregal and three anonymous reviewers for their constructive comments. This work was supported by the European Regional Development Fund and Spanish Government (MINECO/FEDER, UE) under the project with reference DPI2015-64133-R
References (38)
- et al.
Hourly solar radiation over Bahrain
Energy
(1990) - et al.
A novel approach to estimate the clear day global radiation
Renew Energy
(1991) Time-series analysis of climatic variables
Sol Energy
(1995)Generalized autoregressive conditional heteroskedasticity
J Econ
(1986)- et al.
Real-time prediction intervals for intra-hour DNI forecasts
Renew Energy
(2015) - et al.
Review of solar irradiance forecasting methods and a proposition for small-scale insular grids
Renew Sustain Energy Rev
(2013) - et al.
Short-term solar irradiance forecasting using exponential smoothing state space model
Energy
(2013) New methodologies to estimate the hourly global solar radiation; comparisons with existing models
Renew Energy
(2006)- et al.
Economic merits of a state-of-the-art concentrating solar power forecasting system for participation in the spanish electricity market
Sol Energy
(2013) - et al.
Empirical prediction intervals revisited
Int J Forecast
(2014)
Predicting solar radiation at high resolutions: a comparison of time series forecasts
Sol Energy
Univariate and multivariate forecasting of hourly solar radiation with artificial intelligence techniques
Sol Energy
Out-of-sample tests of forecasting accuracy: an analysis and review
Int J Forecast
Volatility forecasting with smooth transition exponential smoothing
Int J Forecast
Short-term solar irradiation forecasting based on dynamic harmonic regression
Energy
Forecasting of global horizontal irradiance by exponential smoothing, using decompositions
Energy
Bollinger on bollinger bands
Investigation of a class of volatility estimators
J Deriv
Time series analysis: forecasting and control
Cited by (33)
Probabilistic-based electricity demand forecasting with hybrid convolutional neural network-extreme learning machine model
2024, Engineering Applications of Artificial IntelligenceInterval forecasting of photovoltaic power generation on green ship under Multi-factors coupling
2023, Sustainable Energy Technologies and AssessmentsInterval forecasting for urban water demand using PSO optimized KDE distribution and LSTM neural networks
2022, Applied Soft ComputingCitation Excerpt :Yu et al. [33]proposed a KDE model with fixed bandwidth to estimate the distribution of wind power predictive errors. Trapero et al. [34] combined volatility forecasting model and the KDE method to improve the performance of solar irradiation PIs. However, a KDE model with optimized bandwidth is more suitable for estimating the PI than a single KDE method since the bandwidth is a key parameter to improve the fitting performance of PDF curves obtained by the KDE model [35].
Impact of the forecast price on economic results for methanol production from olive waste
2021, FuelCitation Excerpt :Forecasting is a vital task in energy applications, particularly, in a renewable energy context due to the lack of solutions for energy storage in power plants. See, for instance, some works related to wind energy [31] and solar energy [32–34]. Regarding price forecasting, literature is mainly focused on electricity time series [35,36] and crude oil [37], whereas, price forecasting of other crucial energy variables as Natural gas is less common [38].