Integration of Genetic Algorithm and Cuckoo Search Optimizer-Based Nested Long Short-Term Memory for Electricity Forecasting

Tirkaamiana, Dean (2022) Integration of Genetic Algorithm and Cuckoo Search Optimizer-Based Nested Long Short-Term Memory for Electricity Forecasting. Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 6010202003-Master_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 September 2025. Download (1MB)

Abstract

Recently, acquiring electricity has become a very important issue for most of sectors such as manufacturing, service, government, or agricultural. Due to electricity generation by non-renewable source, global emission is getting to increase. Non-renewable energy source is the solution, but the development of this energy is not significance especially due to its variability and uncertainty. This issue is enough to force the importance of demand forecasting and the supply should be more precise and to avoid the excessive usage of energy source. There are several methods for electricity forecasting and long-short term memory (LSTM) is one of promising algorithms. One of LSTM variants was developed as nested LSTM (NLSTM) which owns the advantages that LSTM does not have. In addition, attention mechanism (AM) which has been proved can enhance deep learning performance. Thus, the current study intends to combine NLSTM with AM for forecasting. However, there are some parameters which may influence the network performance. Therefore, metaheuristic is employed to find the suitable parameters. Thus, this study integrates genetic algorithm (GA) and cuckoo search optimizer (CSO) called as genetic CSO and applies it to determine the NLSTM-AM parameters in order to reach better network structure. In order to assess the proposed algorithm’s performance, it is compared with some existing algorithms using two datasets in terms of root mean squared error (RMSE). The computational results show that the proposed algorithm can have lower RMSE. Besides, integration of GA and CSO really can outperform GA or CSO individually.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	forecasting, nested long-short term memory, attention mechanism, genetic algorthim, cuckoo search optimizer
Subjects:	T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK1001 Production of electric energy or power
Divisions:	Faculty of Industrial Technology > Industrial Engineering > 26101-(S2) Master Thesis
Depositing User:	Dean Tirkaamiana
Date Deposited:	04 Sep 2023 06:50
Last Modified:	04 Sep 2023 06:50
URI:	http://repository.its.ac.id/id/eprint/96032

Actions (login required)

View Item