National Technical University of Athens (NTUA)
Συμβολή μεθόδων μηχανικής μάθησης στην εκτίμηση και τον έλεγχο δυναμικής ασφάλειας συστημάτων ηλεκτρικής ενέργειας
Abstract
dc:descriptionSecurity is defined as the capability of maintaining the continuous operation of a power system under normal operation and following significant perturbations. As the increase in electric power demand outpaces the installation of new transmission and generation facilities, power systems are forced to operate with narrower security margins; therefore fast and reliable assessment of the system security is necessary. In the field of Dynamic Security Assessment (DSA) much attention has been paid to preventive, as well as corrective control. Preventive control refers to a set of actions that are applied when a potentially dangerous violation is detected through DSA. Corrective actions are applied to offset a security violation after the occurrence of a threatening contingency. The scope of this dissertation is to apply machine learning methods to the dynamic security assessment and control of power systems. Machine learning methods are seen to have features that can bring benefits in comparison to analytical methods. Once developed, they provide solutions very fast. Furthermore they have the ability to recognize, if a system condition has previously occurred and predict its security accordingly. Similarly, if properly designed, they can adapt to new conditions by learning from situations previously seen. Finally they provide a high degree of discovery, i.e. they have the ability to uncover salient, but previously unknown, characteristics of, or relationships in, a system Decision Trees (DT) are applied to propose security rules, which can be used for corrective load shedding strategy as well as security constrained economic dispatch determination. A modified algorithm for the construction of Decision Trees, that utilizes genetic algorithms to formulate characteristic groups of load buses, is also proposed and applied to derive load-shedding schemes. The DT method is further developed and an automatic learning hybrid method for corrective dynamic security is proposed, based on a Self Organized Map (SOM) that classifies the power system's security state according to its load profile. DTs nested in nodes with mixed security states are applied to investigate further their security status. Radial Basis Function Neural Networks (RBFNN) and Support Vector Machines (SVM) are applied on DSA, tested with various feature selection techniques. A hybrid method of RBFNN and Particle Swarm Optimization (PSO) is developed for the definition of remedial actions as well as appropriate preventive control. A Reinforcement Learning framework is proposed for the determination of a spinning reserve policy of an autonomous power system. The algorithm is trained online and compared to deterministic spinning reserve policies. Finally a machine learning method that combines PSO with a modified SOM algorithm is developed for the determination of control islanding of power systems. The methods are applied on realistic models of Cyprus and Hellenic power systems, the MicroGrid and MultiMicroGrid as well as IEEE test systems.
Degree
thesis:*- Grantor dc:publisher
- National Technical University of Athens (NTUA)
- Year dc:date
- 2010
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Voumvoulakis, Emmanouil
- Βουμβουλάκης, Εμμανουήλ
Subjects
dc:subject × 20- Εκτίμηση δυναμικής ασφάλειας
- Μηχανική μάθηση
- Προληπτικός έλεγχος ασφάλειας
- Δέντρα αποφάσεων
- Αυτοοργανωμένος χάρτης
- Βελτιστοποίηση σμήνους σωματιδίων
- Ενισχυτική μάθηση
- Ελεγχόμεη νησιδοποίηση
- Dynamic security assessment
- Machine learning
- Preventive security control
- Decision trees
- Self organizing maps
- Particle swarm optimization
- Reinforcement learning
- Controlled islanding
- Επιστήμες Μηχανικού και Τεχνολογία
- Επιστήμη Ηλεκτρολόγου Μηχανικού, Ηλεκτρονικού Μηχανικού, Μηχανικού Η/Υ
- Engineering and Technology
- Electrical Engineering, Electronic Engineering, Information Engineering
Rights
- Language dc:language
- gre
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/20087