Elman-Based Forecaster Integrated by Adaboost Algorithm in 15 min and 24 h ahead Power Output Prediction Using PM 2.5 Values, PV Module Temperature, Hours of Sunshine, and Meteorological Data

Abstract

Nowadays, with the depletion of fossil energy and deterioration of environmental quality, solar energy is perceived to be a renewable and clean energy. While developing rapidly all over the world, solar energy is also faced with many challenges resulting from its inherent properties. In order to reduce the impact on the grid and facilitate scheduling, it is a growing problem to build a feasible model to forecast PV power with high precision. Therefore, this paper proposes an Elman-based forecaster integrated by Adaboost algorithm, namely Adaboost + Elman. Before forecasting, input variables containing PM 2.5 values, temperature of the PV module, sunshine hours, and meteorological data are made using correlation, clustering, and discriminate analysis to avoid information redundancy and improve the generalization ability of the model. To verify the developed model's application to shortterm PV forecasting in two different time scales, data of Huangsi in 2016 are used for model construction and verification. An additional 7 models are introduced to make comparison. Experimental results prove that the proposed model is effective and practicable for two different scales of short-term PV power prediction.

Forecasting of PM2.5 Concentration in Beijing Using Hybrid Deep Learning Framework Based on Attention Mechanism

“…According to the length of time to forecast PM 2.5 concentration, PM 2.5 prediction models can be classified into short-term prediction models and long-term prediction models [4]. Short-term forecasting is real-time forecasting, focusing on forecast accuracy and ensuring the safety of human activities in the short term by keeping the forecast period within 12 h [5]. The purpose of long-term forecasting is to forecast PM 2.5 concentration more than two days into the future [6], which can serve as a helpful reference for managers.…”

Applicability of a Three-Stage Hybrid Model by Employing a Two-Stage Signal Decomposition Approach and a Deep Learning Methodology for Runoff Forecasting at Swat River Catchment, Pakistan

“…The standalone models sometimes omit useful information when applied alone, therefore, the hybrid models have been proposed to apply for the forecasting of the hydrological time series [27]. These hybrid models have the ability to deal with the intricate problems more efficiently [28].…”