Abstract:
With the globally increasing electricity demand, the use of load forecasting for projecting future’s electricity demand becomes imperative for business entities and policymakers. This demand is governed by a set of different variables known as ‘demand determinants’. These demand determinants change with the change in forecasting horizons (long term, medium term, and short term), utility levels, and country’s overall economic profile. In this work, a review of different electricity demand forecasting methodologies is provided in the context of low and middle-income countries with the emphasis on Pakistan. Also, a comparative review of these forecasting methodologies over different time horizons is presented. This review followed a systematic literature review procedure. This included country selection metric development, article selection process and finally the analysis of the selected literature. A comparative review of the forecasting methodologies over different time horizons revealed that the time series modeling approach has been extensively used while forecasting for long and medium terms. For short term forecasts, artificial intelligence-based techniques remain prevalent in the literature. Furthermore, a comparative analysis of the demand determinants in these countries indicates a frequent use of determinants like the population, GDP, weather, and load data over different time horizons. Furthermore, a comparative analysis of all the demand determinants used in Pakistan and developing countries are also given based on different time horizons. Following that, a detailed review of the load forecasting practices in Pakistan is provided as well. In the light of the identified research gaps in the review section, this works includes the development of a benchmark short term load forecasting model for the electric utilities of Pakistan considering the case study of IESCO. The model development was carried out by acquiring weather station data and other variables. These data were run through a data preprocessing step prior to incorporating in the actual model. After that, data redundancy and other checks were made before finally using the data for experimentation. The forecasting techniques/methods that were used included multiple linear regression, bagged regression trees, and feedforward artificial neural networks. In a comparison with other forecasting techniques, this research suggests the use of artificial neural networks as the most promising technique while forecasting short term electric loads in utilities in Pakistan.
