The power output of a PV plant is a function of location, time, solar-conversion technology, panel area, panel orientation, and, most important, meteorological and climatological conditions. In principle, the dependence of power output on all of these variables, with the exception of meteorological sky conditions, can be modeled deterministically. In clear-sky conditions, output no longer […]

In this section, we address the performance of different forecasting methods with no exogenous variables. The analysis is based on numerical experiments for a particular problem: the forecasting of the 1 h averaged power output of the 1 MW solar farm in Merced. Hourly-average aggregate data from November 2009 to August 2011 (Figure 15.7) were […]

Once the forecasting models are developed using the approaches described in the previous sections, we can turn to assessing their respective performance for comparison. This forecasting-skill assessment is carried out by a number of qualitative and quantitative tests (see Chapter 8). One of the most frequently used techniques for qualitatively assessing forecast accuracy is a […]

A common strategy for creating the GA’s initial population is to use a uniform random distribution to uniformly cover the search space. If good solutions are known, they are often seeded into the initial population. The best individuals from the population are selected based on their fitness—in this case, forecasting accuracy. One of the most […]

15.3.1. GA/ANN: Scanning the Solution Space In general, the following decisions need to be made when creating an ANN-based forecast model: • ANN architecture: number of layers, number of neurons per layer • Preprocessing scheme: smoothing, spectral decomposition, differencing • Fraction and distribution between training and testing data Additionally, ANNs are well suited to multivariate […]

k-Nearest-Neighbors (kNN) is one of the simplest machine-learning algo­rithms. It is a pattern-recognition method for classifying patterns or features (Duda & Hart, 2000). Classification is based on the similarity of a pattern of current values with respect to training samples in the feature space. For the purpose of time-series forecasting, the kNN model consists of […]

Unlike stationary processes that may fluctuate around a constant mean, nonsta­tionary processes (such as the solar resource) are distinct in one or more respects in various scales because of diurnal, seasonal, meteorological, and climatological variations. As a result, in the analysis of nonstationary time series, time plays a fundamental role—as the independent variable in a […]

One of the simplest methods for predicting the future behavior of a time series is the so-called persistence model. Persistence implies that future values of the time series are calculated on the assumption that conditions remain unchanged between “current” time t and future time t + Th. For a stationary time series—one whose mean and […]