Open AccessMid‐term electricity load forecasting by a new composite method based on optimal learning MLP algorithm
Author(s) -
Askari Mostafa,
Keynia Farshid
Publication year - 2020
Publication title -
iet generation, transmission and distribution
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.92
H-Index - 110
eISSN - 1751-8695
pISSN - 1751-8687
DOI - 10.1049/iet-gtd.2019.0797
Subject(s) - computer science , particle swarm optimization , artificial neural network , perceptron , scheduling (production processes) , benchmark (surveying) , electric power system , term (time) , multilayer perceptron , electricity , algorithm , data mining , machine learning , mathematical optimization , power (physics) , engineering , physics , quantum mechanics , electrical engineering , mathematics , geodesy , geography
Electricity load forecasting has been developed as an important issue in the deregulated power system in recent years. Many researchers have been working on the prediction of daily peak load for next month as an important type of mid‐term load forecasting (MTLF). Nowadays, MTLF provides useful information for assessing environmental impacts, maintenance scheduling, adequacy assessment, scheduling of fuel supplies and limited energy resources etc. The characteristics of mid‐term load signal, such as its non‐stationary, volatile and non‐linear behaviour, present serious challenges for this forecasting. On the other hand, many input variables and relative parameters can affect the load pattern. In this study, a new composite method based on a multi‐layer perceptron neural network and optimisation techniques has been proposed to solve the MTLF problem. The proposed method has an optimal training algorithm composed of two search algorithms (particle swarm optimisation and improved ant lion optimiser) and a multi‐layer perceptron neural network. The accuracy of the proposed forecast method is extensively evaluated based on several benchmark datasets.