PhD proposal
Forecasting power generation and distribution by ‘intelligent’ machine
learning optimization methods
PhD director and co-director(s)- Albanie: Ligor Nikolla (ligor.nikolla@umt.edu.al ) and Eralda
Gjika(Dhamo) (eralda.dhamo@fshn.edu.al )
PhD director and co-director(s)- France: Didier Georges(didier.georges@gipsa-lab.grenobleinp.fr), Arben Cela (a.cela@esiee.fr)
Context and general objectives
Production of power in Albania is based heavily on renewable energy system (RES) which is highly variable and unpredictable, leading to the need for optimization-based planning and operation in order to maximize economies while sustaining performance. Proposing prediction models based on available data are essential and are being used from government and non-public producers of energy.
The focus of this PhD research will be on elaboration of prediction model of power systems and mainly in: power generation capacity, power demand, prices, wind and solar power generation capabilities at the country and region level. Climate change , their effect in natural sources used for energy production will be at the attention of the studies. Albania has recently been added to the open market of energy sales and purchases in the region and beyond. Its production and export capabilities should be assessed and exploited by optimizing the natural resources that the country has as one of the most favorable energy areas in the region. During the last years the country has focused its attention to solar source as well as hydropower
which have been the main source of energy production [3,4,5].
Recent developments of machine learning within Big Data environment have not been exploited enough in this domain. This PhD aims to answer the question of how to incorporate.
Big Data platform and machine learning into an intelligent system for modelling and prediction of energy demand and production in the context of renewable energy system.
Using different sources, variety, volume, veracity , speed of data we aim at making these data valuable by obtaining useful information for the stakeholders and decision makers. Big data information on energy market ranges from real-time/intraday auctions and continuous trading, day-ahead auctions, weeks, months or even years . Prediction of day ahead energy load, production, prices is studied in the vast majority ( up to 90%) of the research literature [1,2].
But other seasonality are also important for the stability of the economy of a country . For this purpose for more than two decades various methods and computational tools have been applied to energy data. Starting from statistical models and latter passing to machine learning (ML) due to the increase in the complexity of data and the speed of information. In this study we aim at analyzing the stationarity, trend and seasonality of data, their complexity and ability of models for nowcasting and short/medium/long term prediction. Optimization methods in combination to prediction model will be used to achieve an optimal planning of production , demand and price of energy.
Scientific program / Methodology / Expected results
The thesis is articulated around 6 main tasks:
T1) Bibliography on the existing methodologies of modeling and predicting energy from renewable source (hydro, solar termo etc.)
T2) Collect of consolidated data of energy production and consumption (using various sources of
open data and communication with OSHEE, KESH, OST).
T3) Design and validation of a realistic and adaptive model that optimizes the energy production, distribution and price market control.
T4) Formulation and solution of strategies for optimal control of natural sources used for energy production.
T5) Valorization and dissemination of results, in particular through publications, and the development of a simulation and validation tool.
Some references
[1] Arkadiusz Jędrzejewski , Jesus Lago , Grzegorz Marcjasz1 , Rafał Weron (2022) Electricity Price Forecasting: The Dawn of Machine Learning (2022) , Forthcoming in IEEE Power & Energy Magazine May/June 2022 (doi: 10.1109/MPE.2022.3150809)
[2] T. Hong, P. Pinson, Y. Wang, R. Weron, D. Yang and H. Zareipour, “Energy Forecasting: A Review
and Outlook,” in IEEE Open Access Journal of Power and Energy, vol. 7, pp. 376-388, 2020, doi: 10.1109/OAJPE.2020.3029979 .
[3] Gjika E, Basha L, Ferrja A, Kamberi A. Analyzing Seasonality in Hydropower Plants Energy Production and External Variables. Engineering Proceedings. 2021; 5(1):15. ITISE2021-International Conference on Time Series. Granada, 19th-21th July, 2021. Gran Canaria (SPAIN) https://doi.org/10.3390/engproc2021005015
[4] E.Gjika, E. D. Lamberti, L. Basha (2021), Predicting Energy Production by HPP Using Machine Learning Algorithms with Priority Weights, 41st International Symposium on Forecasting, Online, June, 2021. https://forecasters.org/events/symposium-on-forecasting/
[5] Gjika E., Ferrja A., Basha L., Kamberi A., (2019) Probabilistic Forecasting of Electricity Demand using Markov Chain and Statistical Distribution, International Symposium on Forecasting, Greece, Thessaloniki, June 16-19, 2019. https://forecasters.org/events/symposiumon-forecasting/