Analysis of Energy System in Sweden Based on Time series Forecasting and Regression Analysis

Introduction

The Kingdom of Sweden is one of the northern countries in Europe, located in cold Scandinavia. The country has the area of 450 295 km² and population of 9.6 million in 2013. Sweden has a relatively low population density (21 inhabitants per km²) and most of its land is covered by forest. It has a strong economic with by solid institutional foundations providing a competitiveness open-market system. It can be said that Sweden has withstood to the global economic crisis, due to its strong economic fundamentals. GDP of Sweden in 2006 was 10% higher than the average of OECD and in 2011, and it ranked fifth among European OECD countries in terms of GDP per capita after Luxembourg, Norway, Switzerland and Denmark (IEA, 2013; IEA, 2008). Nowadays, Sweden have a strong energy system because of heavy investment in their alternative, a system mostly based on renewable energy, no longer afterwards the 1970 when the Oil Companies heavily criticized because of their large profit made during the oil crisis, despite of the fact that countries like Sweden and Finland were victims of the circumstance. So analysing of Swedish energy system can be highly effective for oil-import-based developing countries during smooth transition process to using sustainable and environmentally-friendly energies.

Energy management plays an important role in economic prosperity and security of a country and essentially is needed to properly allocate and exploit the available limited resources. Nowadays, many methods and techniques are being used to demand forecasting, including traditionally the linear and non-linear regression, time series, econometric, ARIMA, neural network, fuzzy logic, and genetic algorithm, and new methods such as support vector regression, ant colony and particle swarm optimization (Suganthi & Samuel, 2012).

This research is to present an overview of the Sweden energy system and subsectors, industrial and residential, and forecast their energy consumption by 2024. The prediction methodology is based on linear multiple regression and AHW algorithm. In the past decades, regression analysis widely had been used to forecast the short term and long term demand and supply of oil (Ediger et al., 2006), gas (Gorucu & Gumrah, 2004) and moreover, electricity (Bianco et al., 2009). Turanoglu et al. (2012) presented a neural network model forecasting oil consumption in Turkey, considering import and export values of oil from 1965 to 2010, GDP, and population as independent variables. Ghiasi et al. 2016 used exponential smoothing approach to predict the trend of independent variables including population growth rate, industrial growth rate, and GDP as inputs for a function forecasting primary energy consumption, derived from linear regression analysis. Tabasi et al. 2016 also presented the same method analyzing energy system of Finland. The both studies indicated on resilience of energy supply as a key factor for understanding the energy systems of countries that are highly dependent on importing fossil fuels, and planning a clean energy system for the future.

The work starts with a description of Sweden energy system, focuses on the energy system analysis and energy policies over the last decade. At the next stage, the main macro factors influencing the energy consumption of the whole energy system and the mentioned subsectors are introduced. By using the regression analysis, the energy consumption is predicted from 2014 till 2024. Conducting such statistical analysis for countries with no resources of fossil fuels like Sweden is very useful to clarify the relationship between effective parameters on the total and sector energy consumption, which can provide an effective road map to the associated governments for a clean and efficient energy system.