Gusti, Kinanti Hanugera (2020) Optimasi Parameter Model Least-Square Support Vector Regression Menggunakan Genetic Algorithm dan Particle Swarm Optimization (Studi Kasus : Beban Listrik Jangka Pendek Area Jawa Timur). Masters thesis, Institut Teknologi Sepuluh Nopember.
Preview |
Text
THESIS_06211850010004_Kinanti Hanugera Gusti.pdf Download (4MB) | Preview |
Abstract
Accuracy is the most important thing in time series analysis for forecasting where the obtained model depends on historical, linear or nonlinear patterns. The linear approach is carried out using ARIMA method. Nonlinear modelling uses SVR and LSSVR which are known to have kernel functions so that they can capture patterns of nonlinearity data, but LSSVR computation tends to be faster because it requires a simpler solution without eliminating the SVR principle. Forecasting is done by using a significant lag based on ARIMA as the input variable. Selection of parameters greatly affects the result of accuracy, so an optimization method is needed. The usual, grid search optimization method does not guarantee reaching the optimum solution and not efficient in some practical applications. So, metaheuristic optimization methods is needed, including GA and PSO that are able to find the entire possible space for the solutions. PSO optimization is easier to
apply but prone to premature convergence which causes it
to be trapped in minimum locale, so it will be fixed by Modification of PSO which always looks back for the optimal position according to the specified criteria. The result in this study is the accuracy of nonlinear approach is better than the linear approach. The addition of an optimization method to the SVR gives a significant change compared to LSSVR. Meanwhile, GA is the best optimization method because produces a lower RMSE value.
| Item Type: | Thesis (Masters) |
|---|---|
| Uncontrolled Keywords: | Electricity Load, GA, Optimization, LSSVR, PSO Beban Listrik, GA, OPtimasi, LSSVR, PSO |
| Subjects: | Q Science > QA Mathematics > QA280 Box-Jenkins forecasting T Technology > TK Electrical engineering. Electronics Nuclear engineering |
| Divisions: | Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49101-(S2) Master Thesis |
| Depositing User: | Kinanti Hanugera Gusti |
| Date Deposited: | 28 Aug 2020 06:23 |
| Last Modified: | 04 Jan 2024 07:13 |
| URI: | http://repository.its.ac.id/id/eprint/81310 |
Actions (login required)
 |
View Item |