Analysis on intelligent machine learning enabled with meta-heuristic algorithms for solar irradiance prediction

The solar forecasting is an effective method to enhance the operation of an electrical system for merging a large amount of solar power generation systems and intends to expand a new empirical method to model the prediction uncertainty of the solar irradiance. The proposed model comprises three phases, such as (a) Data Acquisition, (b) Feature Extraction, and (c) Prediction. Initially, benchmark data available from local meteorological organizations are collected that includes the numerical weather forecasting data like temperature, dew point, humidity, visibility, wind speed, and other descriptive information. Once the data is collected, feature extraction is done by first-order and second-order statistical models. First Order Statistics, like mean, median, standard deviation, the maximum value of entire data, and minimum value of entire data, and Second-Order Statistics, like Kurtosis, skewness, correlation, and entropy are extracted as the features. These features are further applied to three machine learning algorithms named Multilayer Perceptron (MLP), Convolutional Neural Network (CNN), and Recurrent Neural Network (RNN). As main novelty of this paper, the number of hidden neurons of all these networks is optimized by a hybrid algorithm merging both the Deer Hunting Optimization Algorithm (DHOA) and Grey Wolf Optimization (GWO), which is named as Grey Updated DHOA (GU-DHOA). The improvement of these networks with the assistance of a hybrid meta-heuristic algorithm will be highly effective for solar irradiance prediction, overcoming the existing machine learning algorithms.