# Long-Term Estimation of Wind Power by Probabilistic Forecast Using Genetic Programming

^{1}

^{2}

^{3}

^{4}

^{*}

## Abstract

**:**

## 1. Introduction

- Very short-term forecast: From a few minutes to one hour ahead.
- Short-term forecast: From one hour to several hours ahead.
- Medium-term forecast: From several hours to one week ahead.
- Long-term forecast: From one week to one year or more ahead.

## 2. Materials and Methods

#### 2.1. Wind Speed Probability Distribution and Wind Power

#### 2.2. Data Grouping, Prediction Horizons and Associated Parameters

^{2}calculated from the real data in the chosen time interval. Additionally, we compute the mean wind speed of the Weibull probability distribution $\overline{v}$. The mean wind speed of the fitted Weibull distribution is compared to the average wind speed calculated from the wind speed time series ${v}_{ave}$, as an indicator of whether the fit is good.

#### 2.3. Genetic Programming to Predict Weibull Distributions

_{W}, ${v}_{ave}$, $\overline{v}$ and $\lambda $, all at the time $t-1$. In addition, the generation of random numbers function is included in the terminal set to consider numerical values. Figure 4 shows an example of genotype and encoding used to forecast the Weibull scale parameter. In that specific case, the scale parameter is a function of the solar radiation, the ambient temperature and the atmospheric pressure.

#### 2.4. Estimating Wind Power at Long-Term

#### 2.5. Forecasting Error

## 3. Experiments and Results

#### 3.1. Data Grouping, Prediction Horizons and Associated Parameters

#### 3.1.1. Seasonality

#### 3.1.2. Grouping and Prediction Horizons

#### 30 Days Ahead

#### 25 Days Ahead

#### 20 Days Ahead

#### 15 Days Ahead

#### 10 Days Ahead

#### 3.1.3. Parameters

#### 3.2. Predicting Weibull Distributions

#### 3.3. Estimating Wind Power

#### 3.4. Forecasting Errors

## 4. Discussion

## 5. Conclusions

## Supplementary Materials

## Author Contributions

## Funding

## Acknowledgments

## Conflicts of Interest

## Nomenclature

$v$ | $\mathrm{m}/\mathrm{s}$ | Wind speed |

${f}_{W}\left(v;\lambda ,k\right)$ | − | Weibull probability density function |

${f}_{\mathrm{PDF}}\left(v\right)$ | − | Probability density function |

$\lambda $ | $\mathrm{m}/\mathrm{s}$ | Scale parameter of the Weibull probability density function |

$k$ | − | Shape parameter of the Weibull probability density function |

${T}_{i}$ | − | Percentage of time (of the total time used to construct the Weibull PDF) where an amount of wind power is produced. |

${v}_{l}$ | $\mathrm{m}/\mathrm{s}$ | Lower wind speed in a time interval |

${v}_{u}$ | $\mathrm{m}/\mathrm{s}$ | Upper wind speed in a time interval |

$\overline{v}$ | $\mathrm{m}/\mathrm{s}$ | Mean wind speed from distribution |

${v}_{\mathrm{ave}}$ | $\mathrm{m}/\mathrm{s}$ | Average mean wind speed from data |

$\mathsf{\Gamma}\left(x\right)$ | − | Gamma function |

$I$ | $\mathrm{W}/{\mathrm{m}}^{2}$ | Solar radiation |

$T$ | $\mathrm{C}$ | Ambient temperature |

$P$ | $\mathrm{mbar}$ | Atmospheric pressure |

$H$ | $\%$ | Relative humidity |

${P}_{W}$ | $\mathrm{W}$ | Produced power from wind turbine |

${P}_{\mathrm{average}}$ | $\mathrm{W}$ | Produced power from a wind turbine in an interval of time |

${P}_{\mathrm{turb}}\left(v\right)$ | $\mathrm{W}$ | Power curve of the wind turbine as a function of the wind speed |

${F}_{gp}$ | − | Set of $NF$ functions used by the GP algorithm |

${f}_{gp\_n}$ | − | n function used by the GP algorithm |

${T}_{gp}$ | − | Set of $NT$ terminals used by the GP algorithm |

${a}_{gp\_1}$ | − | n terminal used by the GP algorithm |

${f}_{FIT}$ | − | Fitness function used by the GP algorithm |

$g$ | − | Generation of GP |

$Pop\left(g\right)$ | − | Population at generation $g$ |

$\iota $ | − | Individual of the population |

${R}^{2}$ | − | Coefficient of determination |

${E}_{t}$ | Units of forecasted variable | Forecast or residual error of a variable at a period $t$ |

${Y}_{t}$ | Units of forecasted variable | Measured value of wind speed at a time period |

${F}_{t}$ | Units of forecasted variable | Forecasted value of wind speed at a time period |

$N$ | − | Sample number of measured and forecasted values of variable |

$\mathrm{MAE}$ | Units of forecasted variable | Mean Absolute Error |

$\mathrm{MAPE}$ | % | Mean Absolute Percentage Error |

## References

- Guo, P.; Huang, X.; Wang, X. A review of wind power forecasting models. Energy Procedia
**2011**, 12, 770–778. [Google Scholar] - Song, Y.; Liu, F.; Hou, R.; Wang, J. Analysis and application of forecasting models in wind power integration: A review of multi-step-ahead wind speed forecasting models. Renew. Sustain. Energy Rev.
**2016**, 60, 960–981. [Google Scholar] - Grundmeyer, M.; Gervert, M.; Lerner, J. The importance of wind forecasting. Renew. Energy Focus
**2009**, 10, 64–66. [Google Scholar] - Ramirez, P.; Carta, J.A. The use of wind probability distributions derived from the maximum entropy principle in the analysis of wind energy. a case study. Energy Convers. Manag.
**2005**, 47, 2564–2577. [Google Scholar] [CrossRef] - Weibull, W. A statistical distribution function of wide applicability. J. Appl. Mech. Trans. ASME
**1951**, 18, 293–297. [Google Scholar] - Sigl, A.B.; Corortis, R.B.; Klein, J. Probability models for wind velocity magnitude and persistence. Sol. Energy
**1978**, 20, 483–493. [Google Scholar] - Hennessey, J.P., Jr. A comparison of the Weibull and Rayleigh distributions for estimating wind power potential. Wind Eng.
**1978**, 2, 156–164. [Google Scholar] - Yalcin, A.; Justus, C.G.; Hargreaves, W.R. Nationwide assessment of potential output from wind-powered generators. J. Appl. Meteorol.
**1976**, 15, 673–678. [Google Scholar] - Bhattacharya, P. A study on Weibull distribution for estimating the parameters. J. Appl. Quant. Methods
**2010**, 5, 234–241. [Google Scholar] [CrossRef] - Rainer, M.; Giebel, S.M.; Aydin, N.S. Simulation and prediction of wind speeds: A neural network for Weibull. JIRSS
**2013**, 12, 293–319. [Google Scholar] - Alam, M.M.; Ameer Uddin, S.M.; Mondal, S.K.; Azad, A.K.; Rasul, M.G. Analysis of wind energy conversion system using Weibull distribution. Proceedia Eng.
**2014**, 90, 725–732. [Google Scholar] - Arik, I.; Yenilmez, I.; Mert Kantar, Y.; Usta, I. Analysis of the modified Weibull distribution for estimation of wind speed distribution. Proc. Int. Conf. Eng. MIS
**2015**, 49. [Google Scholar] [CrossRef] - Islam, R.; Azad, A.K.; Rasul, M.G.; Shishir, I.R. Analysis of wind energy prospect for power generation by three Weibull distribution methods. Energy Proceedia
**2015**, 75, 722–727. [Google Scholar] - Vermorel, J. Probabilistic Forecasting. 2015. Available online: www.lokad.com/probabilistic-forecasting-definition (accessed on 23 February 2020).
- Little, M.A.; McSharry, P.E.; Taylor, J.W. Generalized linear models for site-specific density forecasting of UK daily rainfall. Mon. Weather Rev.
**2009**, 37, 1029–1045. [Google Scholar] [CrossRef] - Strumbelj, E. On determining probability forecasts from betting odds. Int. J. Forecast.
**2014**, 30, 934–943. [Google Scholar] [CrossRef] - Garrat, A.; Lee, K.; Pesaran, M.P.; Shin, Y. Forecast Uncertainties in Macroeconometric Modelling: An Application to the UK Economy. 2000. Available online: www.le.ac.uk/economics/research/RePEc/lec/leecon/econ00-4.pdf (accessed on 23 February 2020).
- Alders, M.; Keilman, N.; Cruijsen, H. Assumptions for long-term stochastic population forecasts in 18 European countries. Eur. J. Popul.
**2007**, 23, 33–69. [Google Scholar] [CrossRef] [Green Version] - Hong, T.; Pinson, P.; Fan, S.; Zareipour, H.; Troccoli, A.; Hyndman, R.J. Probabilistic energy forecasting: Global Energy Forecasting Competition 2014 and beyond. Int. J. Forecast.
**2016**, 32, 896–913. [Google Scholar] [CrossRef] [Green Version] - Zhou, Z. Application of probabilistic wind power forecasting in electricity markets. Wind Energy
**2013**, 16, 321–338. [Google Scholar] [CrossRef] - Wan, C.; Lin, J.; Wang, J.; Song, Y.; Dong, Z.Y. Direct Quantile Regression for Nonparametric Probabilistic Forecasting of Wind Power Generation. IEEE Trans. Power Syst.
**2016**, 32, 2767–2778. [Google Scholar] [CrossRef] - Xydas, E.; Qadrdan, M.; Marmaras, C.; Cipcigan, L.; Jenkins, N.; Ameli, H. Probabilistic wind power forecasting and its application in the scheduling of gas-fired generators. Appl. Energy
**2017**, 192, 382–394. [Google Scholar] [CrossRef] [Green Version] - Bäck, T. Evolutionary Algorithms in Theory and Practice; Oxford University Press: New York, USA, 1996. [Google Scholar]
- Koza, J.R. Hierarchical Genetic Algorithms Operating on Populations of Computer Programs. In Proceedings of the Eleventh International Joint Conference on Artificial Intelligence, Detroit, MI, USA, 20–25 August 1989; Morgan Kaufmann: MA, USA, 1989; Volume 1, pp. 768–774. [Google Scholar]
- Koza, J.R. Genetic Programming: On the Programming of Computers by Means of Natural Selection; MIT Press: Cambridge, MA, USA, 1992. [Google Scholar]
- Holland, J.H. Adaptation in Natural and Artificial Systems; The University of Michigan Press: Cambridge, MA, USA, 1975. [Google Scholar]
- Koza, J.R. Genetic Programming as a means for programming computers by natural selection. Stat. Comput.
**1994**, 4, 87–112. [Google Scholar] [CrossRef] - Graff, M.; Escalante, H.J.; Ornelas-Tellez, F.; Tellez, E.S. Time series forecasting with genetic programming. Nat. Comput.
**2017**, 16, 165–174. [Google Scholar] [CrossRef] - Danandeh, M.A.; Bagheri, F.; Resatoglu, R. A Genetic Programming Approach to Forecast Daily Electricity Demand. In 13th International Conference on Theory and Application of Fuzzy Systems and Soft Computing—ICAFS-2018. Advances in Intelligent Systems and Computing; Springer: Cham, Switzerland, 2018; p. 896. [Google Scholar]
- Karabulut, K.; Alkan, A.; Yilmaz, A.S. Long Term Energy Consumption Forecasting Using Genetic Programming. Math. Comput. Appl.
**2008**, 13, 71–80. [Google Scholar] [CrossRef] [Green Version] - Lee, D.G.; Lee, B.W.; Chang, S.H. Genetic programming model for long-term forecasting of electric power demand. Electron Power Syst. Res.
**1997**, 40, 17–22. [Google Scholar] [CrossRef] - Graff, M.; Peña, R.; Medina, A. Wind Speed Forecasting using Genetic Programming. IEEE Congr. Evol. Comput.
**2013**, 408–415. [Google Scholar] [CrossRef] - Martínez-Arellano, G.; Nolle, L. Genetic Programming for Wind Power Forecasting and Ramp Detection. In Research and Development in Intelligent Systems XXX; Springer: Heidelberg, Germany, 2013. [Google Scholar]
- Carta, J.A.; Ramírez, P.; Velázquez, S. A review of wind speed probability distributions used in wind energy analysis case studies in the Canary Islands. Renew. Sustain. Energy Rev.
**2009**, 13, 933–955. [Google Scholar] [CrossRef] - Rinne, H. The Weibull Distribution, A Handbook; CRC Press: Boca Raton, FL, USA, 2009. [Google Scholar]
- Unnikrishna, P.S.; Papoulis, A.P. Probability, Random Variables, and Stochastic Processes; McGraw-Hill: Boston, MA, USA, 2002. [Google Scholar]
- Jaramillo, O.A.; Borja, M.A. Bimodal versus Weibull wind speed distributions: An analysis of wind energy potential in la venta, Mexico. Wind Eng.
**2004**, 28, 225–234. [Google Scholar] [CrossRef] - Mathew, S. Wind Energy: Fundamentals, Resource Analysis and Economics; Springer: Berlin/Heidelberg, Germany, 2006. [Google Scholar]
- Johnson, G.L. Wind Energy Systems; Kansas State University: Manhattan, KS, USA, 2006. [Google Scholar]
- Celik, A. Weibull representative compressed wind speed data for energy and performance calculations of wind energy systems. Energy Convers. Manag.
**2003**, 44, 3057–3072. [Google Scholar] [CrossRef] - Ulgen, K.; Hepbasli, A. Determination of Weibull parameters for wind energy analysis of Izmir, Turkey. Int. J. Energy Res.
**2002**, 26, 495–506. [Google Scholar] [CrossRef] - Langlois, R. Estimation of Weibull parameters. J. Mater. Sci. Lett.
**1991**, 10, 1049–1051. [Google Scholar] [CrossRef] - Azad, A.K.; Rasul, M.G.; Yusaf, T. Statistical Diagnosis of the Best Weibull Methods for Wind Power Assessment for Agricultural Applications. Energies
**2014**, 7, 3056–3085. [Google Scholar] [CrossRef] [Green Version] - Kececioglu, D. Reliability and Life Testing Handbook; Prentice Hill, Inc.: Upper Saddle River, NJ, USA, 1993; Volume I. [Google Scholar]
- Baker, J.E. Reducing Bias and Inefficiency in the Selection Algorithm. In Proceedings of the 2nd International Conference on Genetic Algorithms, NJ, USA, 1987; Grefenstette, Ed.; 1987; pp. 14–21. [Google Scholar]
- Rau, V.G.; Jangamshetti, S.H. Normalized power curves as a tool for identification of optimum wind turbine generator parameters. IEEE Trans. Energy Convers.
**2001**, 16, 283–288. [Google Scholar] [CrossRef] - El-Sharkawi, M.A. Wind Energy, an Introduction; CRC Press: Boca Raton, FL, USA, 2015. [Google Scholar]
- Armstrong, J.S.; Collopy, F. Error measures for generalizing about forecasting methods: Empirical comparisons. Int. J. Forecast.
**1992**, 1, 69–80. [Google Scholar] [CrossRef] [Green Version] - Hernandez-Escobedo, Q.; Manzano-Agugliaro, R.; Zapata-Sierra, A. The wind power of Mexico. Renew. Sustain. Energy Rev.
**2010**, 14, 2830–2840. [Google Scholar] [CrossRef] - Robles, M.; Sanchez-Perez, P.A.; Jaramillo, O.A. Real time Markov chains: Wind states in anemometric data. Renew. Sustain. Energy
**2016**, 8, 024403. [Google Scholar] - Borja, M.A.; Jaramillo, O.A. Wind speed analysis in La Ventosa, Mexico: A bimodal probability distribution case. Renew. Energy
**2004**, 29, 1613–1630. [Google Scholar] - Saldana, R.; Jaramillo, O.A.; Miranda, U. Wind power potential of Baja California Sur, Mexico. Renew. Energy
**2004**, 29, 2087–2100. [Google Scholar] - Jaramillo, O.A.; Cadenas, E.; Rivera, W. Analysis and forecasting of wind velocity in Chetumal, Quintana Roo, using the exponential smoothing method. Renew. Energy
**2010**, 35, 925–930. [Google Scholar] - Tzen, E. Small wind turbines for on grid and off grid applications. IOP Conf. Ser. Earth Environ. Sci.
**2020**. [Google Scholar] [CrossRef] - Olsen, T.L.; Preus, R. Small Wind Site Assessment Guidelines; NREL/TP-5000-63696: September 2015. Available online: https://www.nrel.gov/docs/fy15osti/63696.pdf (accessed on 23 February 2020).
- Ozgener, O. A small wind turbine system (SWTS) application and its performance analysis. Energy Conversion and Management.
**2006**, 47, 1326–1337. [Google Scholar] [CrossRef] - Borunda, M.; De La Cruz, J.; Garduno-Ramirez, R.; Nicholson, A. Technical assessment of small-scale wind power for residential use in Mexico: A Bayesian intelligence approach. PLoS ONE
**2020**, 15, e0230122. [Google Scholar] [CrossRef] [Green Version] - Jain, P.; Wijayatunga, P. Grid Integration of Wind Power: Best Practices for Emerging Wind Markets; Asian Development Bank: Mandaluyong, Philippines, 2016; Volume 43, pp. 2–36. [Google Scholar]
- Danish Energy Agency. Energy Policy Toolkit on System Integration of Wind Power; Experiences from Denmark; Danish Energy Agency: København, Denmark, 2015; ISBN 978-87-93071-93-3. [Google Scholar]
- Idriss, A.I.; Ahmed, R.A.; Omar, A.I.; Said, R.K.; Akinci, T.C. Wind energy potential and micro-turbine performance analysis in Djibouti-city, Djibouti. Eng. Sci. Technol. Int. J.
**2019**, 23, 65–70. [Google Scholar] [CrossRef] - Barbosa de Alencar, D.; Mattos Affonso, C.; Limao de Oliveira, R.C.; Moya Rodríguez, J.L.; Cabral Leite, J.; Reston Filho, J.C. Different Models for Forecasting Wind Power Generation: Case Study. Energies
**2017**, 10, 1976. [Google Scholar] [CrossRef] [Green Version] - Kanna, B.; Singh, S.N. Long term wind power forecast using adaptive wavelet neural network. In Proceedings of the 2016 IEEE Uttar Pradesh Section International Conference on Electrical, Computer and Electronics Engineering (UPCON), Varanasi, India, 9–11 December 2016. [Google Scholar]
- Barbounis, T.G.; Theocharis, J.B. Locally recurrent neural networks for long-term wind speed and power prediction. Neurocomputing
**2006**, 69, 466–496. [Google Scholar] [CrossRef] - Wang, Y.; Hu, Q.; Meng, D.; Zhu, P. Deterministic and probabilistic wind power forecasting using a variational Bayesian-based adaptive robust multi-kernel regression model. Appl. Energy
**2017**, 208, 1097–1112. [Google Scholar] [CrossRef] - Willis, M.J.; Hiden, H.H.; Marenbach, P.; McKay, B.; Montague, G.A. Genetic Programming: An introduction and survey of applications. In Proceedings of the Second International Conference on Genetic Algoriths in Engineering Systems: Innovations and Applications, Glasgow, UK, 2–4 September 1997. [Google Scholar] [CrossRef] [Green Version]
- Dhunny, A.A.; Lollchund, M.R.; Rughooputh, S.D.D.V. Evaluation of a wind farm project for a smart city in the South-East Coastal Zone of Mauritius. J. Energy S. Afr.
**2016**, 1, 39–50. [Google Scholar] [CrossRef] [Green Version] - Tuller. S.; Brett, A. The characteristics of wind velocity that favor the fitting of a Weibull distribution in wind speed analysis. J. Appl. Meteorol.
**1984**, 23, 124–134. [Google Scholar] [CrossRef] [Green Version] - Akdag, S.A.; Dinler, A. A new method to estimate Weibull parameters for wind energy applications. Energy Convers. Manag.
**2009**, 50, 1761–1766. [Google Scholar] [CrossRef] - Stevens, M.; Smulders, P. The estimation of the parameters of the Weibull wind speed distribution for wind energy utilization purposes. Wind Eng.
**1979**, 3, 132–145. [Google Scholar] - Seguro, J.V.; Lambert, T.W. Modern estimation of the parameters of the Weibull wind speed distribution for wind energy analysis. J. Wind Eng. Ind. Aerodyn.
**2000**, 85, 75–84. [Google Scholar] [CrossRef] - Bilir, L.; Devrim, Y.; Albostan, A. Seasonal and yearly wind speed distribution and wind power density analysis based on Weibull distribution function. Int. J. Hydrog. Energy
**2015**, 40, 15301–15310. [Google Scholar] [CrossRef] - Perea-Moreno, A.-J.; Alcalá, G.; Hernandez-Escobedo, Q. Seasonal wind energy characterization in the Gulf of Mexico. Energies
**2020**, 13, 93. [Google Scholar] [CrossRef] [Green Version] - Carrillo, C.; Cidrás, J.; Díaz-Dorado, E. Obando-Montaño, A.F. An approach to determine the Weibull parameters for wind energy analysis: The case of Galicia (Spain). Energies
**2014**, 7, 2676–2700. [Google Scholar] [CrossRef] [Green Version] - Fyrippis, I.; Axaopoulos, P.J.; Panayiotou, G. Wind energy potential assessment in Naxos Island, Greece. Appl. Energy
**2010**, 87, 577–586. [Google Scholar] [CrossRef] - Darwish, A.S.K.; Sayigh, A.A.M. Wind energy potential in Iraq. Sol. Wind Technol.
**1988**, 5, 215–222. [Google Scholar] [CrossRef] - Katsoulis, B.D.; Metaxas, D.A. The wind energy potential of western Greece. Sol. Energy
**1992**, 46, 463–476. [Google Scholar] [CrossRef] - Akdag, S.A.; Bagiorgas, H.S.; Mihalakakou, G. Use of two-component Weibull mixtures in the analysis of wind speed in the Eastern Mediterranean. Appl. Energy
**2010**, 87, 2566–2573. [Google Scholar] - Jung, C.; Schindler, D. Wind speed distribution selection—A review of recent development and progress. Renew. Sustain. Energy Rev.
**2019**, 114, 109290. [Google Scholar] [CrossRef] - Drobinski, P.; Coulais, C. Is the Weibull distribution really suited for wind statistics modeling and wind power evaluation? arXiv
**2012**, arXiv:1211.333853v1. [Google Scholar]

**Figure 2.**Weibull probability distribution function for different values of the shape $k$ and scale $\lambda $ parameters. The blue lines correspond to $k=2$ and $\lambda =3,5,7\mathrm{m}/\mathrm{s}$ for the dotted, solid and dash-dotted lines, respectively. The red and yellow solid lines correspond to $k=1,3$, respectively, and $\lambda =5\mathrm{m}/\mathrm{s}$.

**Figure 4.**Example of genotype and its encoding, prefix notation and mathematical model used in the GP.

**Figure 7.**Normalized histogram of wind speeds, in blue, and fitted Weibull distribution, in red, for: (

**a**) 2010; and (

**b**) 2011. The Weibull parameters are $k=2$ and $\lambda =5.99$ and $k=2$ and $\lambda =5.82$, for 2010 and 2011, respectively.

**Figure 8.**Six groups of samples of normalized histograms of wind speed, in blue, and fitted Weibull distribution, in orange, for group periods of 30 days each during WS season. Each graph corresponds to the following period (

**a**) 01/01/2010–01/30/2010; (

**b**) 01/31/2010–03/01/2010; (

**c**) 03/02/2010–03/31/2010; (

**d**) 04/01/2010–04/30/2010; (

**e**) 05/01/2010–05/30/2010; (

**f**) 12/21/2010–01/19/2011.

**Figure 9.**Six groups of samples of normalized histograms of wind speeds, in blue, and fitted Weibull distribution, in orange, for group periods of 25 days each during WS season. Each graph corresponds to the following period (

**a**) 01/01/2010–01/25/2010; (

**b**) 01/26/2010–02/19/2010; (

**c**) 02/20/2010–03/16/2010; (

**d**) 03/17/2010–04/10/2010; (

**e**) 04/11/2010–05/05/2010; (

**f**) 05/06/2010 – 05/30/2010.

**Figure 10.**Six groups of samples of normalized histograms of wind speeds, in blue, and fitted Weibull distribution, in orange, for group periods of 20 days each during WS season. Each graph corresponds to the following period (

**a**) 01/01/2010–01/20/2010; (

**b**) 01/21/2010–02/09/2010; (

**c**) 02/10/2010–03/01/2010; (

**d**) 03/02/2010–03/21/2010; (

**e**) 03/22/2010–04/10/2010; (

**f**) 04/11/2010–04/30/2010.

**Figure 11.**Six groups of samples of normalized histograms of wind speeds, in blue, and fitted Weibull distribution, in orange, for group periods of 15 days each during WS season. Each graph corresponds to the following period (

**a**) 01/01/2010–01/15/2010; (

**b**) 01/16/2010–01/30/2010; (

**c**) 01/31/2010–02/14/2010; (

**d**) 02/15/2010–03/01/2010; (

**e**) 03/02/2010–03/16/2010; (

**f**) 03/17/2010–03/31/2010.

**Figure 12.**Six groups of samples of normalized histograms of wind speeds, in blue, and fitted Weibull distribution, in orange, for group periods of 10 days each during WS season. Each graph corresponds to the following period (

**a**) 01/01/2010–01/10/2010; (

**b**) 01/11/2010–01/20/2010; (

**c**) 01/21/2010–01/30/2010; (

**d**) 01/31/2010–02/09/2010; (

**e**) 02/10/2010–02/19/2010; (

**f**) 02/20/2010–03/01/2010.

**Figure 13.**Predicted scale parameters at different horizons. The blue and the gray dashed lines correspond to the values of the measured and predicted values of the scale parameter in the WS season. The yellow and the gray dashed lines correspond to the values of the measured and predicted values of the scale parameter in the SF season. The horizons of the prediction are as follows: (

**a**) 30 days ahead; (

**b**) 25 days ahead; (

**c**) 20 days ahead; (

**d**) 15 days ahead; and (

**e**) 10 days ahead.

**Figure 14.**Estimated and measured wind power for horizons of: (

**a**) 30 days ahead; (

**b**) 25 days ahead; (

**c**) 20 days ahead; (

**d**) 15 days ahead; and (

**e**) 10 days ahead. The blue and yellow lines correspond to the measured wind power in the WS and SF seasons, respectively, and the grey dashed lines correspond to the estimated wind power.

**Figure 15.**Forecast error $E$ for the estimated wind power for horizons of: (

**a**) 30 days ahead; (

**b**) 25 days ahead; (

**c**) 20 days ahead; (

**d**) 15 days ahead; and (

**e**) 10 days ahead. The blue and yellow lines correspond to the WS and SF seasons, respectively.

GP ( ) |
---|

g ← 0 random initialization of each individual $\mathit{\iota}$∈ Pop(g) ∀$\mathit{\iota}$∈ Pop (g) evaluate f _{FIT}($\mathit{\iota}$)do while $\mathit{g}$ < stopping criterion g ← g−1 Pop(g) ← selection (Pop(g−1)) Pop(g) ← crossover (Pop(g)) Pop(g) ← mutation (Pop(g)) ∀ $\mathit{\iota}$ ∈ Pop(g) evaluate f _{FIT}($\mathit{\iota}$)end do |

**Table 2.**Characteristics of the wind resource per year: average power $P$, average wind speed from data ${v}_{ave}$ and mean wind speed from Weibull distribution $\overline{v}$.

Year | $\mathit{P}$ (W/m^{2})
| ${\mathit{v}}_{\mathit{a}\mathit{v}\mathit{e}}(\mathbf{m}/\mathbf{s})$ | $\overline{\mathit{v}}$ (m/s) |
---|---|---|---|

2010 | 175.56 | 5.15 | 5.314 |

2011 | 160.31 | 5.09 | 5.155 |

**Table 3.**Characteristics of the wind resource per season: scale parameter $\lambda $, average power $P$, average wind speed from data ${v}_{ave}$ and mean wind speed from Weibull distribution $\overline{v}$.

Year | Season | Dates | $\mathit{\lambda}\left(\mathbf{m}/\mathbf{s}\right)$ | $\mathit{P}$ (W/^{m})
| ${\mathit{v}}_{\mathit{a}\mathit{v}\mathit{e}}(\mathbf{m}/\mathbf{s})$ | $\overline{\mathit{v}}$ (m/s) |
---|---|---|---|---|---|---|

2010 | Winter | 01/01–03/20 | 6.97 | 276.03 | 6.11 | 6.18 |

2010 | Spring | 03/21–06/20 | 6.28 | 202.03 | 5.58 | 5.56 |

2010 | Summer | 06/21–09/20 | 4.75 | 87.77 | 4.22 | 4.21 |

2010 | Fall | 09/21–12/20 | 5.23 | 116.71 | 4.60 | 4.63 |

2010 | Winter | 12/21–12/31 | 8.35 | 471.43 | 7.09 | 7.40 |

2011 | Winter | 01/01–03/20 | 6.70 | 244.45 | 5.76 | 5.93 |

2011 | Spring | 03/21–06/20 | 6.04 | 179.68 | 5.61 | 5.35 |

2011 | Summer | 06/21–09/20 | 5.04 | 104.10 | 4.53 | 4.46 |

2011 | Fall | 09/21–12/20 | 5.25 | 117.72 | 4.49 | 4.65 |

2011 | Winter | 12/21–12/31 | 6.34 | 207.33 | 5.40 | 5.61 |

10-day Group | λ (m/s) | k (-) | Average Solar Radiation (W/m ^{2}) | Average Ambient Temerature (C) | Average Relative Humidity (%) | Average Atm Pressure (mbar) | Average Wind Power (m/s) | Average Wind Speed (m/s) | $\overline{\mathit{u}}$ (m/s) |
---|---|---|---|---|---|---|---|---|---|

1 | 4.08 | 2.00 | 180.44 | 8.06 | 59.28 | 774.86 | 55.14 | 3.66 | 3.61 |

2 | 6.27 | 2.00 | 200.99 | 7.02 | 68.77 | 772.57 | 200.70 | 5.35 | 5.56 |

3 | 8.04 | 2.00 | 262.32 | 11.02 | 39.22 | 770.77 | 413.51 | 7.33 | 7.13 |

4 | 6.65 | 2.00 | 184.73 | 7.80 | 66.92 | 769.76 | 239.22 | 6.17 | 5.90 |

5 | 7.00 | 2.00 | 195.05 | 9.51 | 64.07 | 770.81 | 278.25 | 5.71 | 6.20 |

6 | 8.15 | 2.00 | 304.64 | 9.70 | 39.85 | 770.98 | 429.45 | 7.76 | 7.23 |

7 | 7.68 | 2.00 | 246.50 | 12.57 | 38.99 | 771.59 | 363.64 | 6.81 | 6.81 |

8 | 6.55 | 2.00 | 324.38 | 11.24 | 26.15 | 772.40 | 228.36 | 6.10 | 5.80 |

9 | 6.43 | 2.00 | 347.04 | 13.88 | 26.10 | 772.63 | 216.41 | 5.89 | 5.70 |

10 | 6.71 | 2.00 | 304.84 | 16.77 | 30.41 | 772.94 | 245.53 | 6.34 | 5.95 |

**Table 5.**Specifications of the GP process to forecast the scale parameter of wind speed Weibull distributions.

Parameter | Description |
---|---|

Objective: | To find a mathematical function that accurately forecasts the scale parameter of wind speed Weibull distributions. |

Population size: | 100 individuals (mathematical functions). |

Maximum number of tree nodes: | 50 |

Forecast cases (Prediction horizons): | 10, 15, 20, 25 and 30 days ahead groups. |

Fitness function: | MAPE function of the scale parameter (Equation (10)). |

GP evolutionary operations: | Crossover and Mutation. |

Reproduction probability by mutation: | 10% |

Reproduction probability by crossover: | 90% |

Stopping criterion: | 2000 generations |

Prediction Period | Winter-Spring MAE MAPE (m/s) (%) | Summer-Fall MAE MAPE (m/s) (%) | ||
---|---|---|---|---|

30 days ahead | 0.31 | 17.62 | 0.29 | 24.14 |

25 days ahead | 0.46 | 18.92 | 0.28 | 23.47 |

20 days ahead | 0.55 | 20.97 | 0.32 | 25.05 |

15 days ahead | 0.55 | 21.40 | 0.33 | 29.94 |

10 days ahead | 0.52 | 23.74 | 0.33 | 26.93 |

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## Share and Cite

**MDPI and ACS Style**

Borunda, M.; Rodríguez-Vázquez, K.; Garduno-Ramirez, R.; de la Cruz-Soto, J.; Antunez-Estrada, J.; Jaramillo, O.A.
Long-Term Estimation of Wind Power by Probabilistic Forecast Using Genetic Programming. *Energies* **2020**, *13*, 1885.
https://doi.org/10.3390/en13081885

**AMA Style**

Borunda M, Rodríguez-Vázquez K, Garduno-Ramirez R, de la Cruz-Soto J, Antunez-Estrada J, Jaramillo OA.
Long-Term Estimation of Wind Power by Probabilistic Forecast Using Genetic Programming. *Energies*. 2020; 13(8):1885.
https://doi.org/10.3390/en13081885

**Chicago/Turabian Style**

Borunda, Mónica, Katya Rodríguez-Vázquez, Raul Garduno-Ramirez, Javier de la Cruz-Soto, Javier Antunez-Estrada, and Oscar A. Jaramillo.
2020. "Long-Term Estimation of Wind Power by Probabilistic Forecast Using Genetic Programming" *Energies* 13, no. 8: 1885.
https://doi.org/10.3390/en13081885