The rapid implementation of wind turbines across the globe corresponds to a set of challenges during power systems operation and planning. This is due to intermittent nature of wind potential. By the end of 2018, the total European Union-installed offshore wind capacity reached 19 GW. Total investments in offshore wind in 2018 were more than 10.3 billion € (WindEurope [
1], 2018). This includes investments in construction of projects, transmission assets, and refinancing. More than 91% of all offshore wind installations were located in shallow or intermediate water depths with a mean water depth equal to 27.1 m. The capacities span from few to hundred megawatts and the installations differ in terms of hub site, distance to shore, water depth and others (Snyder et al. [
2], 2009). The reduction of installation and maintenance costs but also the reliable assessment of energy production of offshore wind parks will signify the next phase of their deployment.
The forecasting models are distinguished into physical and statistical. Physical models take into consideration parameters like the topology of the ground (for on-site parks) and topology of the wind park and temperature. They use the outputs of a Numerical Weather Prediction (NWP) model and provide final forecasts. NWP models are used by meteorologists and usually they provide predictions for the next 48 up to 172 h ahead. For the case of wind power predictions, the wind speed prediction is obtained directly by the NWP and the wind power prediction is obtained by using the power curve of a wind turbine. Physical models are appropriate for long-term predictions. However, it is difficult to scale the forecasts per wind turbine or per wind farm. Also, physical models are complex in terms of inputs requirements and execution time. On the other hand, statistical models are favored in short-term prediction horizons. The wind is treated as a regression of its past values. A relatively large number of historical values are needed to train the models and define their optimal composition. Statistical models refer to time series models, Artificial Neural Networks, Bayesian Networks, Support Vector Machines, and others. Time series models refer to autoregressive models, autoregressive models with moving average, and others. The main advantages are their potential for removing the trend of time series and their availability in software packages. However, there is a difficulty in extracting the optimal structure of the model. Also, time series models require a large number of historical values and are not definitely suitable for highly nonlinear series. Neural networks are suitable for nonlinear series and are a favorable scheme in many forecasting problems. Support vector machines are also a well-known forecasting engine but they demand large durations for their training and their parameters are optimized by a relatively complex process. Bayesian networks are more appropriate for small data sets. Finally, various statistical models can be integrated to form ensemble forecasting models [
4,
5,
6,
7,
8,
9,
10,
11].
In (Sfetsos [
12], 2002), a comparison takes place between a persistent model, autoregressive integrated moving average (ARIMA) and neural network. The latter outperforms the rest. No external variables are considered. The comparison takes place in two different sets that refer to one month each. In (Başaran and Filik [
13], 2017) the authors consider three cases of inputs for the neural network, i.e., using only past wind speed values, using past wind speed values and temperature and, finally, using wind speed, temperature, and pressure. The test refers to five days and two intervals for predictions are taken into account, namely 30 and 90 s ahead. The case with wind speed, temperature, and pressure leads to the lower errors. No comparison with other models is presented. In (More and Deo [
14], 2003) the authors test a feed-forward neural network, a recurrent neural network, and an ARIMA model to forecast daily, weekly, and monthly wind speeds at two coastal locations in India using only past wind speed values. The feed-forward neural network wins the competition. In (Li and Shi [
15], 2010), the authors examine three types of neural networks, namely, the adaptive linear element, backpropagation, and radial basis function. The wind data used are the hourly mean wind speed collected at two observation sites in a United States of America (USA) location. The results show that even for the same wind dataset, no single neural network model outperforms others universally in terms of all evaluation metrics. Moreover, the selection of the type of neural networks for best performance is also depends upon the data sources. In (Zeng and Qiao [
16], 2011) a support vector machine model is presented for wind power forecasting. Instead of predicting wind power directly, the model first predicts the wind speed, which is then used to predict the wind power by using the power–wind speed characteristics of the wind turbine generators. Simulation studies are carried out to validate the proposed model for very short-term and short-term predictions by using the data obtained from the National Renewable Energy Laboratory of USA. The model is compared with feed-forward neural network and radial basis network. The prediction is held using only past wind speed values. In (Zhou et al. [
17], 2011), the authors present a least-squares support vector machine for one-step ahead wind speed forecasting. Three kernels, namely linear, Gaussian, and polynomial kernels, are implemented. The support vector machine‘s parameters considered include the training sample size, order, regularization parameter, and kernel parameters. The support vector machine‘s version are compared with a persistence model and provide better forecasts. The Adaptive Neuro-Fuzzy Inference System (ANFIS) is utilized in (Fazelpour et al. [
18], 2016), and is compared with a feed-forward neural network and radial basis network in hour-ahead forecasting in a location in Tehran, Iran. No exogenous parameters are used. ANFIS results in better forecasts. In (Fortuna et al. [
19], 2016), the clustering tool is used to form wind speed classes. Then, two models, namely the Hidden Markov Model and the Nonlinear Autoregressive are compared for predicting the class of each new wind speed data entry. In general, wind speed series present volatilities and stochasticity. Depending on the data set, an analysis on the wind speed characteristics can take place. For instance, in (Fortuna et al. [
20], 2014), the authors provide a fractal analysis on wind speed observations. Exploitable information can be derived for such analysis for further modeling.