Deep Learning Approach For Forecasting And Energy Management Of Renewable Energy System

Abstract

To address the global energy crisis, the need for renewable energy has skyrocketed. newlineRenewable energy options are appealing since they are environmentally friendly and newlineclean in nature. For the last decade, wind energy has been one of the most appealing and newlinepromising renewable energy resources (RESs) for meeting global power demand. Wind newlinespeed and wind power forecasting play an important role in planning the integration of newlinewind power to power grids for efficient energy management tasks such as unit newlinecommitment, capacity planning, load balancing, power quality improvement, and newlinefrequency regulation, among others, due to the rapid growth of wind power penetration newlineinto modern power grids in a micro or smart grid environment. However, wind speed is newlineone of the most difficult weather characteristics due to its influence on other parameters newlinesuch as world rotation, topographical properties of the earth, temperature, and pressure, newlineand so on. As a result, correctly forecasting wind speed and wind power is difficult. newlineFurthermore, precise short-term wind speed and wind power forecasting is required for newlinesystem operators to make power generation schedule and dispatch decisions at newlineconventional power plants, as well as to determine reserve power. As a result, prediction newlinemodels are built to deliver reliable forecasting results in the short term. newlineFor prediction of different wind power, various hybrid networks have been developed. newlineAn efficient novel hybrid time series forecasting model integrating variational mode newlinedecomposition (VMD) and Deep learning mixed Kernel ELM (MKELM) Autoencoder newline(AE) is provided here for exact wind power prediction. It is commonly known that newlineKELM-AE has numerous advantages. The VMD parameters (and#945;, K) are then optimized to newlineproduce an appropriate number of intrinsic modes (IMFs) using a recently developed newlinemeta-heuristic population-based Sine Cosine integrated water cycle algorithm newline(SCWCA). Furthermore, the mixed kernel parameters and related weights are improved newlineusing the same SCWCA to improve the Deep MKELM pred