Vallabhan, Satish. C. (1995) Application of artificial neural networks for voltage stability evaluation of power systems. Masters thesis, Memorial University of Newfoundland.
PDF (Migrated (PDF/A Conversion) from original format: (application/pdf))
- Accepted Version
Available under License - The author retains copyright ownership and moral rights in this thesis. Neither the thesis nor substantial extracts from it may be printed or otherwise reproduced without the author's permission.
Voltage instability has emerged as one of the most important areas of concern to modern power utilities. Once associated with weak power systems and long transmission lines, voltage instability problems are now acutely felt in highly developed networks. This is because many utilities are loading their bulk transmission networks to their maximum capacity to avoid the enormous capital costs of building new lines, In recent years, voltage instability has been responsible for several system collapses in Europe, Asia and North America. -- Voltage instability is concerned with the ability of a power system to maintain acceptable voltage at all buses in the system under normal loading conditions and after being subjected to a disturbance. A system enters a state of voltage instability when a disturbance, increase in load demand, or change in system condition causes a progressive and uncontrollable decline in voltage. The main reason causing voltage instability is the inability of the power system to meet the demand for reactive power. The other factors contributing to voltage instability are generator reactive power/voltage control limits, load characteristics, characteristics of static var compensators, and action of on load transformer tap changers. -- The study of voltage instability has become an important area of research in the field of power system engineering. The main thrust of research has been to arrive at an accurate and reliable indicator of the proximity of a system to voltage collapse. Such an indicator would be useful to utilities in operating their systems with maximum economy and security. However, for such voltage stability indices to be truly useful to utilities from an operations point of view, they should be implemented on-line in the Energy Management System (EMS). The Energy Management System has become a very important tool in modern power system control and operation and has versatile capabilities for power system control, analysis and monitoring. The major hurdle in the on-line implementation of voltage stability indices in an EMS would be the heavy computational costs involved in terms of time, memory and hardware costs. This is because most methods for voltage stability analysis need repeated solutions of power flows and associated calculations. Thus, for on-line applications, there is a need for tools which can quickly identify potentially dangerous conditions and provide the operator with guidance to steer the system from voltage collapse. Also, in view of the large size of modem power networks, it is important that the memory requirements of the computational tools be as low as possible. -- In recent years, there has been considerable interest in the application of Artificial Neural Networks (ANN) to power system problems. Artificial Neural Networks have the ability to identify and classify complex relationships, which are nonlinear and result from large mathematical models. The main feature of an ANN is the ability to achieve complicated input-output mappings through a learning process, without explicit programming. Once an ANN has been trained, it can classify new data much faster than would be possible by solving the model analytically. ANNs have the potential to play an important role in Energy Management Systems by providing system operators with a fast and reliable indication of the voltage stability of a power system. -- This thesis presents the application of ANNs for evaluation of power system voltage instability. Two popular voltage stability indices are studied and simulations are carried out on the IEEE 24 Bus system and the 39 Bus New England system. The effect of contingencies on the voltage stability of the above two systems was investigated. ANN models were designed to evaluate the voltage stability indices using the system parameters available from the EMS as inputs. For the energy margin based voltage stability index, separate ANN models were used for each contingency. However, for the load margin index, a single ANN model which takes into account the network topology, was used to evaluate the voltage stability. This single ANN model is able to evaluate the voltage stability of a system under normal operating condition (i.e., all lines in service) and also in the event of a line outage. Simulation results are presented on the application of the above indices to both power systems. The performance of the ANN models are presented, which compares the predicted accuracy to the expected value. The thesis also proposes a scheme for integrating the ANN based system into the EMS environment.
|Item Type:||Thesis (Masters)|
|Additional Information:||Bibliography: 113-116.|
|Department(s):||Engineering and Applied Science, Faculty of|
|Library of Congress Subject Heading:||Neural networks (Computer science); Electric power system stability; Voltage regulators|
Actions (login required)