1 March 2018 in Energy Policy
1 December 2021 in Cleaner and Responsible Consumption
25 April 2022 in Energy Reports
1 February 2015 in Proceedings of the Institution of Civil Engineers - Energy
This paper analyses the deviations between nominated and realised gas quantities, for the balancing group consisting of 36 members in the Croatian gas market. The imbalance distribution among the balancing group members regarding the absolute deviation of a single member and the causer–helper principle is presented. In total, 458 calculations using different architectures of artificial neural networks were conducted to complete the research. The combined analysis of accuracy obtained with the best combination of input variables for the defined datasets is presented in this paper. The results are represented according to the multi input multi output (MIMO) and two multi input single output (MISO) architectures used for the day-ahead and intraday forecasts. The results decrease the gap in knowledge in the field of gas consumption forecasting. Furthermore, the accuracy of gas consumption forecasting using different neural network architectures is calculated. There is a clear benefit for balancing group members if they are organised in the balancing group with the aim of calculating the imbalance based on the group effect. Finally, the results show that the deviation tolerances between nominated and realised gas quantities for a single member can be higher compared to the deviation for its balancing group without significant economic penalty for the member. This paper analyses the deviations between nominated and realised gas quantities, for the balancing group consisting of 36 members in the Croatian gas market. The imbalance distribution among the balancing group members regarding the absolute deviation of a single member and the causer–helper principle is presented. In total, 458 calculations using different architectures of artificial neural networks were conducted to complete the research. The combined analysis of accuracy obtained with the best combination of input variables for the defined datasets is presented in this paper. The results are represented according to the multi input multi output (MIMO) and two multi input single output (MISO) architectures used for the day-ahead and intraday forecasts. The results decrease the gap in knowledge in the field of gas consumption forecasting. Furthermore, the accuracy of gas consumption forecasting using different neural network architectures is calculated. There is a clear benefit for balancing group members if they are organised in the balancing group with the aim of calculating the imbalance based on the group effect. Finally, the results show that the deviation tolerances between nominated and realised gas quantities for a single member can be higher compared to the deviation for its balancing group without significant economic penalty for the member.
1 June 2017 in Tehnicki vjesnik - Technical Gazette
U radu je prikazan inovativni pristup u odabiru strategije razvoja plinskog distribucijskog sustava na temelju formiranog modela. U radu je primijenjen analitički hijerarhijski proces (AHP) koji kao jedan od kriterija za odabir...