期限(时间)
计算机科学
电力系统
可靠性工程
估计
控制器(灌溉)
能源管理系统
控制(管理)
功率(物理)
实时计算
能量(信号处理)
能源管理
工程类
统计
人工智能
系统工程
物理
量子力学
农学
数学
生物
作者
Daniel Trudnowski,J.M. Johnson,Paul D. Whitney
摘要
A fundamental objective of a power-system operating and control scheme is to maintain a match between the system`s overall real-power load and generation. To accurately maintain this match, modern energy management systems require estimates of the future total system load. Several strategies and tools are available for estimating system load. Nearly all of these estimate the future load in 1-hour steps over several hours (or time frames very close to this). While hourly load estimates are very useful for many operation and control decisions, more accurate estimates at closer intervals would also be valuable. This is especially true for emerging Area Generation Control (AGC) strategies such as look-ahead AGC. For these short-term estimation applications, future load estimates out to several minutes at intervals of 1 to 5 minutes are required. The currently emerging operation and control strategies being developed by the BPA are dependent on accurate very short-term load estimates. To meet this need, the BPA commissioned the Pacific Northwest National Laboratory (PNNL) and Montana Tech (an affiliate of the University of Montana) to develop an accurate load prediction algorithm and computer codes that automatically update and can reliably perform in a closed-loop controller for the BPA system. The requirements include accurate load estimation in 5-minute steps out to 2 hours. This report presents the results of this effort and includes: a methodology and algorithms for short-term load prediction that incorporates information from a general hourly forecaster; specific algorithm parameters for implementing the predictor in the BPA system; performance and sensitivity studies of the algorithms on BPA-supplied data; an algorithm for filtering power system load samples as a precursor to inputting into the predictor; and FORTRAN 77 subroutines for implementing the algorithms.
科研通智能强力驱动
Strongly Powered by AbleSci AI