Modeling and Forecasting of Power Plant Generation using Machine Learning Approach

Abstract

Electricity usage planning is a main concern for electricity stakeholders in a country. The exhaustible resources are not enough to address energy demand in our country. It comes with varied problems such as price, environmental hazards and availability of resources. Renewable energy resources, mainly hydropower energy is an ideal solution to energy deficiency in Pakistan. However, to meet the compelling demand for electricity and to deal with different uncertainties involved in this process, development of sustainable policies through proper planning is becoming increasingly challenging. To overcome the above-mentioned problems, we propose to study electricity generation trend of Tarbela Power Plant. Historical data of last 5 years on a daily resolution of generation is used to develop regression models and predict energy generation. This study shall help in analysis of future supply of electricity produced by the Powerhouse. A prediction-based model for electricity production will be useful in forecasting future energy generation and therefore will play a significant role in electric energy planning. It will allow stakeholders to visualize operational excellence of the Plant and incorporate many influencing factors such as decision making on tariffs, policy regulations, investments, available resources and the environmental factors in the country. Once the proposed model is validated using historical data, it will be accredited as a decision support tool for planning future generation needs