DefinePK

Abstract

In the last decade, energy consumption in Indonesia has seen an average increase of 7-8% per year as population and economic growth continue to improve. This condition requires the availability of good energy to support economic activities and social dynamics of the community. Nevertheless, there are various challenges and obstacles to meet the energy needs such as petroleum production, which tends to decline, while the acceleration of new renewable energy development is expected to become the new backbone of national energy is still not maximized. Under these conditions, all efforts to realize energy security must be a priority agenda for Indonesia. The importance of realizing energy security is due to the dynamics of the global energy sector in the coming years not only influenced by supply, demand and price, but also other factors such as geopolitical issues and stability of areas where world energy sources are located. In this research, we will develop a hybrid algorithm application to predict the non renewable energy price in Indonesia. Hybrid algorithm in this study is a combination of genetic algorithm with Nelder Mead and named rvGA-eNM. The development model of computational intelligence conducted in this research is utilizing the advantages of the Nelder Mead algorithm in exploiting the optimal solution through local search and Genetic algorithm capability in conducting optimal solution exploration in the global search area. Data on non-renewable energy prices will be used to measure the performance of proposed hybrid models in the form of historical data of non-renewable energy prices several months earlier. The average prediction error will be the reference in choosing the right model for the non-renewable energy price prediction the next few months. The purpose of this research is to improve the accuracy of non-renewable energy prediction pricing model based on computational intelligence. Non-renewable energy prices are predicted using hybrid algorithm optimization. Predicted non-renewable energy prices during 2005-2014 is shown in figure visualizes the comparison between the actual value and the non-renewable energy price prediction. The values shown show that in most test points, the prediction value approximates the actual (adjacent) values. This explains that the accuracy of the rvGA-eNM model used in the prediction of non-renewable energy prices has high robustness properties.