**4. Power generation forecasts**

Contribution of wind energy has been the largest share out of the renewable energies and expects growth further. For responsible and sustainable growth of wind energy industry, reliability, robustness and stability are important factors. As wind energy integration to the grid is in MW scale, in future it may function as base load plant. So, the decision of economic load dispatch will largely be affected by proper forecasting of wind power. The objective is to improve accuracy in forecasting wind speed and power 1 day ahead so that it becomes reliable, which will be a benefit to the load dispatch centers as well as installation of additional wind turbines onshore and offshore.

Wind forecasting has been taken up in literature by various researchers. The forecasting for power may be very short term (within 2.5 s), short term (10 min to 1 h), long term (15 min to 3 h) or a day ahead (24 h). Forecasting wind speed is an important factor, based on which planning of new wind farm depends. Specifically for offshore wind farms, the safety requirement has less advanced. As wind speed prediction and power prediction takes time for computation and error in forecasting wind power 1 day ahead is more compared to the short term, there is a need for improvement. Research has shown good result from the hybrid method. The researchers are oriented to make wind power predictable. When the wind is

**123**

efficiency also.

generation and time of use.

*Grid-Connected Distributed Wind-Photovoltaic Energy Management: A Review*

predictable, it becomes reliable, which will be a benefit to the load dispatch centers for economic load dispatch as well as the installation of additional wind turbines

Going through the available tools and the accuracy, the methods/prediction models are broadly divided into physical, statistical and artificial intelligence based methods [13]. Out of various statistical methods such as curve fitting, statistical approximation autoregressive integrated moving average (ARIMA), seasonal ARIMA, extrapolation with periodic function, methods of finding probability density

functions (PDF) have been evaluated by the coefficient of determination [14]. Different software models have been developed such as WPMS, WPPT, Prediktor, ARMINES, Previento, Zephyr, AWPPS, Ewind, ANEMOS and adopted in different countries [15]. Some of them are hybrid methods. Prediction of offshore extreme wind is important for the protection of offshore wind system so that such sites can be avoided during planning. Method of independent storms (MIS) stands better as compared to the other three in the study by An et al. [16]. In another work, the extreme wind has been estimated by the combination of swarm optimization with the traditional methods which added improvement [17]. The available software has their limitations up to how many meteorological data required, precision in numerical weather prediction (NWP), different accuracy indices for short and long term prediction etc. Intelligent techniques such as Artificial Neural Network (ANN) [18], Fuzzy, Support Vector Machine (SVM), Wavelet, Hilbert-Huang transform, data mining techniques [19], swarm optimization combining the statistical methods of time series prediction with improvement in nonlinear node functions and training algorithms have given good results as compared to statistical/any method alone [19, 20]. Combination of Fuzzy and ANN take less prediction time thus gives faster result [21]. It has been remarked that grouping wind farms for wind forecasting can give better result [19]. Instead of predicting the wind speed exactly, prediction into lower and upper bounds method (LUBE) [22] in prediction interval with defined confidence level gives better result in performance indices. Wind speed has been estimated by RBF (radial basis function) neural network and wind turbine has been appropriately controlled for maximization of wind power [23]. "Anti-phase correlation" of wind speed and solar radiation has been found after wavelet analysis, implying that wind and solar energy

can complement each other in generating electricity [24].

wind speed for sensor-less control is the need.

Smart grid performs also with penetration of PV and has to consolidate its performance figures in the presence of variability. Many researchers report on the novel hybrid intelligent algorithm for PV forecasting taking its fluctuating behavior. In this regard, wavelet transforms (WT), stochastic learning, remote sensing method and fuzzy ARTMAP (FA) network. Forecasting accurately improves system

As numerical prediction depends on weather data, which is provided by sensors, reduction of dependence on sensors for wind speed, rather estimation method of

Different capacity of battery, wind, PV is considered to check which proportion of each component is economical for a specific location Dhahran in Saudi Arabia taking historical weather data during the demand of different months in a year for the wind-PV hybrid power system (WPVHPS) [25]. The addition of wind generation is more economical than PV. The addition of more battery can reduce the diesel

Prediction is vital for energy management [26]. The energy management functions in a wind-battery system are to (1) charge the battery from wind (2) supply the load from wind power (3) trade the wind/battery power to the grid (4) buy power from grid and store in the battery or supply it to the load; and (5) supply the local load from battery. The day ahead electricity rate and wind energy are

*DOI: http://dx.doi.org/10.5772/intechopen.88923*

onshore and offshore.

### *Grid-Connected Distributed Wind-Photovoltaic Energy Management: A Review DOI: http://dx.doi.org/10.5772/intechopen.88923*

predictable, it becomes reliable, which will be a benefit to the load dispatch centers for economic load dispatch as well as the installation of additional wind turbines onshore and offshore.

Going through the available tools and the accuracy, the methods/prediction models are broadly divided into physical, statistical and artificial intelligence based methods [13]. Out of various statistical methods such as curve fitting, statistical approximation autoregressive integrated moving average (ARIMA), seasonal ARIMA, extrapolation with periodic function, methods of finding probability density functions (PDF) have been evaluated by the coefficient of determination [14]. Different software models have been developed such as WPMS, WPPT, Prediktor, ARMINES, Previento, Zephyr, AWPPS, Ewind, ANEMOS and adopted in different countries [15]. Some of them are hybrid methods. Prediction of offshore extreme wind is important for the protection of offshore wind system so that such sites can be avoided during planning. Method of independent storms (MIS) stands better as compared to the other three in the study by An et al. [16]. In another work, the extreme wind has been estimated by the combination of swarm optimization with the traditional methods which added improvement [17]. The available software has their limitations up to how many meteorological data required, precision in numerical weather prediction (NWP), different accuracy indices for short and long term prediction etc. Intelligent techniques such as Artificial Neural Network (ANN) [18], Fuzzy, Support Vector Machine (SVM), Wavelet, Hilbert-Huang transform, data mining techniques [19], swarm optimization combining the statistical methods of time series prediction with improvement in nonlinear node functions and training algorithms have given good results as compared to statistical/any method alone [19, 20]. Combination of Fuzzy and ANN take less prediction time thus gives faster result [21]. It has been remarked that grouping wind farms for wind forecasting can give better result [19]. Instead of predicting the wind speed exactly, prediction into lower and upper bounds method (LUBE) [22] in prediction interval with defined confidence level gives better result in performance indices. Wind speed has been estimated by RBF (radial basis function) neural network and wind turbine has been appropriately controlled for maximization of wind power [23]. "Anti-phase correlation" of wind speed and solar radiation has been found after wavelet analysis, implying that wind and solar energy can complement each other in generating electricity [24].

Smart grid performs also with penetration of PV and has to consolidate its performance figures in the presence of variability. Many researchers report on the novel hybrid intelligent algorithm for PV forecasting taking its fluctuating behavior. In this regard, wavelet transforms (WT), stochastic learning, remote sensing method and fuzzy ARTMAP (FA) network. Forecasting accurately improves system efficiency also.

As numerical prediction depends on weather data, which is provided by sensors, reduction of dependence on sensors for wind speed, rather estimation method of wind speed for sensor-less control is the need.

Different capacity of battery, wind, PV is considered to check which proportion of each component is economical for a specific location Dhahran in Saudi Arabia taking historical weather data during the demand of different months in a year for the wind-PV hybrid power system (WPVHPS) [25]. The addition of wind generation is more economical than PV. The addition of more battery can reduce the diesel generation and time of use.

Prediction is vital for energy management [26]. The energy management functions in a wind-battery system are to (1) charge the battery from wind (2) supply the load from wind power (3) trade the wind/battery power to the grid (4) buy power from grid and store in the battery or supply it to the load; and (5) supply the local load from battery. The day ahead electricity rate and wind energy are

*Wind Solar Hybrid Renewable Energy System*

kind of diurnal and seasonal trends.

**4. Power generation forecasts**

bines onshore and offshore.

more). These terminologies carry their own implication.

Distribution can be short-term (minutes, less than 1 h) or long-term (1 h or

Power from sun, wind, and ocean additionally exhibit predictable seasonal patterns recognized as a distinguishing variability characteristic. Pattern forecast for this trend of wind and sun is complicated, and it is a subject matter taken up in many papers. In a precise study, Tande et al. [11] have viewed reanalysis data set for illustrating of wind variability characteristics. With information of a temporal resolution of 6 h and a spatial resolution of 2.5° in each latitude and longitude, a two-dimensional linear interpolation of neighboring locations is utilized to get wind speeds at the chosen sites. Both offshore and onshore information can be dealt with in this way for explaining the variability. It is apparent that entry of offshore

In the study performed by Wiemken et al. [12] record from 1995 extracted from 100 monitored PV systems (rooftop plants 1–5 kW) with a 5 min time resolution ensembled for 243 kW (grid connected) is used. A model is developed taking onshore wind and PV energy generation for the period 2001–2011 across 27 nations in Europe. The data is taken from NASA for hourly values of wind speed and solar irradiance documented at a spatial resolution of 0.5° E/W and 0.66° N/S. The generation from wind and PV translated from the climatic record were later on combined to structure regional or nation-specific datasets. The model first considered PV and wind sources to contribute half of the energy supply of total requirement. Further PV share in the wind/PV proportions of 0, 20, 40 and 60% are investigated.

Contribution of wind energy has been the largest share out of the renewable energies and expects growth further. For responsible and sustainable growth of wind energy industry, reliability, robustness and stability are important factors. As wind energy integration to the grid is in MW scale, in future it may function as base load plant. So, the decision of economic load dispatch will largely be affected by proper forecasting of wind power. The objective is to improve accuracy in forecasting wind speed and power 1 day ahead so that it becomes reliable, which will be a benefit to the load dispatch centers as well as installation of additional wind tur-

Wind forecasting has been taken up in literature by various researchers. The forecasting for power may be very short term (within 2.5 s), short term (10 min to 1 h), long term (15 min to 3 h) or a day ahead (24 h). Forecasting wind speed is an important factor, based on which planning of new wind farm depends. Specifically for offshore wind farms, the safety requirement has less advanced. As wind speed prediction and power prediction takes time for computation and error in forecasting wind power 1 day ahead is more compared to the short term, there is a need for improvement. Research has shown good result from the hybrid method. The researchers are oriented to make wind power predictable. When the wind is

wind generation and its variability will noticeably affect the grid.

Many such related papers refer to "step changes" as a variability characteristic. These are the alteration in the available resource that takes place in small time steps of minutes to some hours. Another variable characteristic is autocorrelation [10] which figures out the statistical relation among values of the same parameter in a series. The relationship of wind speed information between different locations and the corresponding relationship of solar irradiance for different locations are under study by several projects. This spatial correlation is perceived as one of the instrument to gauge variability characteristics. Wind and solar sources may also show one

**122**

forecasted through Wavelet-ARMA of time series breaking it into smooth subseries. The state of charge (SOC) of battery is predicted in a longer time horizon. In another case study of Turkey [27], based on 15 years of data of global solar radiation distribution, no relationship between the distribution of annual time lapse and solar energy and solar radiation intensity are established.

The solar and wind energy potential are surveyed for five sites in Corsica [28]. From this study, two sites with the desirable trait are chosen and the sizing and the economics for an isolated hybrid PV/wind systems are compared. The trend is dependent on site-specific resource analysis. The sites with more wind potential have less cost of energy and more feasibility.
