Abstract
Optimal power flow is a widely used tool in power system planning and management. Due to the complexity of the power system both in terms of number of variables, degrees of freedom and uncertainty, there is a continuous effort to find more efficient computational methods to solve optimal power flow problems. This article presents a novel method to speed-up the solution of a security constraint optimal power flow problem. An unconventional warm start based on the training of a neural network is investigated as an option to improve the computational efficiency of the optimization problem. The principle of the method and the validity of the approach is demonstrated by different analysis performed on the IEEE14 test grid and based on a linearized mathematical formulation of the problem. The results show the effectiveness of the method in reducing the number of iterations needed to converge to global optimum.
Zusammenfassung
Die Berechnung eines optimalen Leistungsflusses ist ein Standardinstrument bei der Planung und dem Betrieb von Energiesystemen. Aufgrund der Komplexität des Energiesystems sowohl in Bezug auf die Anzahl der Variablen als auch auf die Freiheitsgrade und Unsicherheiten werden viele Versuche unternommen, effizientere Berechnungsmethoden zur Lösung dieser Probleme zu finden. Dieser Beitrag stellt eine neuartige Methode vor, um die Lösung eines optimalen Leistungsflussproblems (Security Constraint Optimal Power Flow) zu beschleunigen. Basierend auf dem Training eines neuronalen Netzes wird ein unkonventioneller Warmstart der Optimierung zur Verbesserung der Berechnungseffizienz untersucht. Das Prinzip der Methode und die Gültigkeit des Vorgehens werden durch verschiedene Analysen am IEEE14 Test-Grid Netz und mittels einer linearisierten mathematischen Formulierung des Problems demonstriert. Die Ergebnisse zeigen die Effzienz der Methode zur Reduzierung der Berechnungszeit.
About the authors
Marco Giuntoli received the M.S. degree in electrical engineering and the Ph.D. degree in electrical and thermal energy from University of Pisa, Italy, in 2008 and 2012, respectively. His Postdoc was focus on Power System optimization and planning. During his carrier, he had collaborations with universities research and consulting institute and, finally in 2017, he joined the Corporate Research Center in Ladenburg Germany (Now Hitachi ABB Power Grids). His expertise includes Power System analysis and modelling and mathematical optimization.
Veronica Biagini (born 1983) is head of the German Research Center at Hitachi ABB Power Grids. She holds a M.Sc. and Ph.D. in Electrical Engineering from the University of Pisa, Italy. In 2011 she joined ABB Corporate Research working as a Scientist and focusing on the design and optimization of mechatronics systems. She held different positions in Corporate Research before joining Hitachi ABB Power Grids in 2019. Her work is focused on digital transformation of power grids through the development of concepts and solutions for grid planning, operation and control.
Moncef Chioua is an Assistant Professor of Chemical Engineering at Polytechnique Montréal. His research focuses on process data analysis with applications in industrial processes monitoring and process operators support systems. Prof. Chioua was a member of the process data analysis and optimization research group at ABB Corporate Research Center in Ladenburg, Germany between 2008 and 2020. He was a member of the process control research group at PAPRICAN, Pointe- Claire (now FP-Innovations) from 2003 to 2004. Prof. Chioua holds a PhD. from Polytechnique Montréal, an engineering degree from the École Supérieure des Sciences et Technologies de l’Ingénieur de Nancy (ESSTIN) and a Diplôme d’Études Approfondies (DEA) from the National Polytechnic Institute of Lorraine (INPL).
References
1. Biagini, V., M. Subasic, A. Oudalov and J. Kreusel (2020): “The autonomous grid: Automation, intelligence and the future of power systems,” Elsevier: Energy Research and Social Science, 65.10.1016/j.erss.2020.101460Search in Google Scholar
2. Biskas, P. N. and A. Bakirtzis (2004): “Decentralised security constrained dc-opf of interconnected power systems,” IEE Proceedings – Generation, Transmission and Distribution, 151, 1350–2360.10.1049/ip-gtd:20041063Search in Google Scholar
3. Capitanescu, F., J. L. Martinez Ramos, et al. (2011): “State-of the art, challenges, and future trends in security constrained optimal power flow,” Electric Power Systems Research, 81, 1731–1741.10.1016/j.epsr.2011.04.003Search in Google Scholar
4. Fioretto, F., T. W. Mak and P. V. Hentenryck (2020): “Predicting ac optimal power flows: Combining deep learning and lagrangian dual methods,” Proceeding of AAAI 2020 conference.10.1609/aaai.v34i01.5403Search in Google Scholar
5. Giuntoli, M., V. Biagini and K. Schönleber (2019): “Novel formulation of ptdf and lodf matrices for security constrained optimal power flow for hybrid ac and dc grids,” Innovative Smart Grid Technologies Conference Europe (ISGT Europe), IEEE PES.10.1109/ISGTEurope.2019.8905672Search in Google Scholar
6. Glorot, X., A. Bordes and Y. Bengio (2011): “Deep sparse rectifier neural networks,” Proceedings of the 14th International Conference on Artificial Intelligence and Statistics (AISTATS).Search in Google Scholar
7. Goodfellow, I., Y. Bengio and A. Courville (2016): Deep Learning, The MIT Press.Search in Google Scholar
8. Kingma, D. and J. Ba (2014): “Adam: A method for stochastic optimization,” International Conference on Learning Representations.Search in Google Scholar
9. Molzahn, D. K., F. Dörfler, et al. (2017): “A survey of distributed optimization and control algorithms for electric power systems,” IEEE Transactions on Smart Grid, 8, 2941–2962.10.1109/TSG.2017.2720471Search in Google Scholar
10. Muter, I., S. I. Birbil and K. Bülbül (2012): “Simultaneous column-and-row generation for large-scale linear programs with column-dependent-rows.”10.1007/s10107-012-0561-8Search in Google Scholar
11. Overbye, T., X. Cheng and Y. Sun (2004): “A comparison of the ac and dc power flow models for lmp calculations,” Proceedings of the 37th Annual Hawaii International Conference on System Sciences.10.1109/HICSS.2004.1265164Search in Google Scholar
12. Pan, X., T. Zhao and M. Chen (2019): “Deepopf: A deep neural network approach for security-constrained dc optimal power flow,” arXiv preprint arXiv:1910.14448.Search in Google Scholar
13. Paucar, V. L. and M. J. Rider (2002): “Artificial neural networks for solving the power flow problem in electric power systems,” Elsevier: Electric Power System Research, 62, 139–144.10.1016/S0378-7796(02)00030-5Search in Google Scholar
14. Wood, A. J., B. F. Wollenberg and G. B. Sheblé (2013): Power Generation, Operation, and Control, Wiley.Search in Google Scholar
15. Xavier, A. S., F. Qiu and S. Ahmed (2019): “Learning to solve large-scale security-constrained unit commitment problems.”10.1287/ijoc.2020.0976Search in Google Scholar
16. Xavier, A. S., F. Qui, F. Wang and P. R. Thimmapuram (2020): “Transmission constraint filtering in large-scale security-constrained unit commitment,” IEEE Transactions on Power Systems.10.1109/TPWRS.2019.2892620Search in Google Scholar
17. Yalcinoz, T. and M. J. Shor (1998): “Neural networks approach for solving economic dispatch problem with transmission capacity constraints,” IEEE Transactions on Power Systems, 13, 307–313.10.1109/59.667341Search in Google Scholar
18. Yindong, S. and N. Yudong (2008): “Column generation approaches to large driver scheduling problems,” IEEE Chinese Control Conference.Search in Google Scholar
© 2020 Walter de Gruyter GmbH, Berlin/Boston
