Abstract
The great scales, nonlinearities and uncertainties in modern power systems mean that they are among the most intractable problems in dynamic control. In the present paper, direct neural dynamic programming (direct NDP) is introduced for a real time supplementary control application. Direct NDP is an on-line learning control paradigm that learns to improve system performance by following the computed gradient toward meeting the overall learning objective. As such the method makes use of on-line measurements to generate proper control actions. This feature is of critical significance when dealing with dynamic systems that are difficult to model or model precisely. In this paper, a static var compensator (SVC) supplementary damping control in a 4-generator 2-area system is implemented using direct NDP. The self-learning and adaptive abilities of direct NDP are analyzed in the MATLAB environment. Simulation results demonstrate the advantages of direct NDP over conventional control.
Original language | English (US) |
---|---|
Title of host publication | IEEE International Symposium on Intelligent Control - Proceedings |
Pages | 270-274 |
Number of pages | 5 |
State | Published - 2004 |
Event | Proceedings of the 2004 IEEE International Symposium on Intelligent Control - 2004 ISIC - Taipei, Taiwan, Province of China Duration: Sep 2 2004 → Sep 4 2004 |
Other
Other | Proceedings of the 2004 IEEE International Symposium on Intelligent Control - 2004 ISIC |
---|---|
Country/Territory | Taiwan, Province of China |
City | Taipei |
Period | 9/2/04 → 9/4/04 |
Keywords
- Direct neural dynamic programming
- Dynamic stability
- Power system
- Svc supplementary control
ASJC Scopus subject areas
- Hardware and Architecture
- Control and Systems Engineering