Anticipatory fuzzy automatic generation control scheme using very short-term ANN ACE forecasting
Load frequency control can be challenging for a control area with a large number of widely fluctuating steel mill loads under the new NERC(North American Electric Reliability Council) control performance standards. Of particular interest in this thesis is the question as to whether a redesign of the load frequency controller with anticipatory ability could improve the area's control performance with respect to CPS1 and CPS2, and to reducing the tie flow fluctuations and unit reversals. In formulating the control strategy, security and reliability of the interconnection should take precedence over economic considerations. The control objective is two fold, first to comply with CPS1 and CPS2, only then attempt reduction in tie flow fluctuations and unit reversals. This thesis begins with a description of a recursive method to forecast the ACE over a very short-term for purposes of load frequency control using Artificial neural network. For on-line control purposes, the forecast has to be reasonably accurate and of sufficient time resolution. The technique was tested on data previously collected from a utility with widely fluctuating steel mill loads. Along with the results obtained is an assessment of the technique's viability and performance. Following which we investigated the application of very short-term forecasting in an anticipatory fuzzy Automatic generation control scheme to improve the area's CPS1 and CPS2 control performance and to reduce tie flow fluctuations. Results of the investigation are given and discussed. Also given in this thesis is an on-line method to estimate the frequency response characteristic of a control area. ^
Major Professor: Chee-Mun Ong, Purdue University.
Engineering, Electronics and Electrical