| Summary: | 碩士 === 國立中山大學 === 電機工程研究所 === 84 === State Estimation (SE) is the basis for real-time monitoring
and control. There are many ways to obtain the SE solution
, and the Weighted Least Square (WLS) method is a most basic
approach. In the Energy Management System (EMS) implementation
, another important idea is the "fast decoupled" concept which
makes the real-time applications possible. The Measurements
commonly used are bus injection, line flow and bus voltage
magnitude measurements. Owing to the limit of computing
resources, the traditional WLS based methods tried to make
use of the sparse algorithm to limit the size of the Jacobian
matrix.SE runs on the real-time measurement at prespecified
time interval, generally 10-20 minutes, and an
"observability" module has to be executed beforehand.
Computing resources are becoming cheaper and cheaper
nowadays. For example, the price of memory storage is low, and
it is worthwhile to trade the memory space for a better
performance. Besides, the 10-20 minute time interval will
degrade the essence of the "real-time operation"
substantially, the meaning of "real-time" deserves a new
definition. Moreover, if the observability test could be
co-processed by the estimation procedure, the performance of
SE could be further enhanced. This research presents the
framework of a novel Recursive Least Square (RLS) state
estimation method. This method is a static estimation method,
based on the recursive way to process all the measurements.
The proposed method could avoid the problem of a
separate observability module, the divergence problem would
no longer exists,and the essence of quasi-dynamic for real-time
is further implemented. All the above performance can be
obtained by the cost of adding a full-size matrix. Tests will
be conducted to show that the new method is very robust, fast,
and is easy to implement. Bad data process will also be
discussed.
|
|---|
