Title :
The usage of neural networks and time series in pattern recognition and forecasting of power consumption profiles
Author :
Torres, H. ; Duarte, O.G. ; Cajamarca, G.A. ; Gallego, L.E. ; Pavas, F.A. ; Urrutia, D.F.
Author_Institution :
Nat. Univ. of Colombia, Colombia
Abstract :
Most of the power quality disturbances such as unbalances, sags, swells and harmonic distortion are ultimately related to power consumption variations. In this manner, with the aim of identifying electrical problems causing PQ disturbances, a suitable knowledge of these power consumption profiles is required. These profiles can be obtained by either a continuous monitoring or by using some tools capable of representing the behavior of the power demand. This paper presents a comparison between two analytical tools, one is an artificial intelligence approach by means of neural networks, and the other one uses statistical techniques such as time series analysis. These techniques not only can represent power consumption profiles, but also may predict them allowing the customer to make a suitable planning of the electrical facilities.
Keywords :
artificial intelligence; monitoring; neural nets; pattern recognition; power engineering computing; power supply quality; power system faults; time series; artificial intelligence approach; continuous monitoring; electrical facilities; neural networks; pattern recognition; power consumption profiles; power consumption variations; power demand; power quality disturbances; statistical techniques; time series analysis; Artificial intelligence; Artificial neural networks; Energy consumption; Harmonic distortion; Monitoring; Neural networks; Pattern recognition; Power demand; Power quality; Time series analysis;
Conference_Titel :
Power Engineering Society General Meeting, 2005. IEEE
Conference_Location :
San Francisco, CA, USA
Print_ISBN :
0-7803-9157-8
DOI :
10.1109/PES.2005.1489724
