Micro-grid Load Forecasting Based on Cloud Computing and Machine Learning Algorithms

Abstract

With the increasingly serious energy and environmental pollution problems,
comprehensive development and rational use of new energy are imperative, and the
construction of microgrids can fully absorb new energy and optimize the energy structure.
Accurate load forecasting can not only provide an important basis for optimal operation of
microgrids and energy management decisions, but also ensure efficient economic
operation of microgrids. Therefore, this paper conducts research on the short-term load
forecast of microgrids, which has important theoretical significance and practical value for
the optimal operation of microgrid systems. This article first analyzes the characteristics of
microgrid load forecasting and its influencing factors. It uses an improved hybrid
frog-leaping algorithm to optimize the combined parameters of the kernel function
extreme learning machine (ISFLA_KELM). At the same time, the Sparkon YARN
platform is introduced to improve the algorithm in parallel To meet the challenges brought
by big data by parallel computing while ensuring the prediction accuracy, and using the
real load data of a micro-grid to verify the prediction accuracy and execution efficiency.
The Spark-based ISFLA_KELM algorithm proposed in this paper achieved a minimum
load forecast MAPE value of 6.125% and a minimum execution time of 236s. The
experimental results show that the load forecasting accuracy of the proposed algorithm is
better than the existing algorithms, and has good parallel performance, which can provide
an effective basis for microgrid load forecasting.