A New Intelligence Solution for Power System Economic Load Dispatch

[attachment=17463]
INTRODUCTION
ECONOMIC load dispatch (ELD) is an important task in
the modern Energy Management System (EMS), which
aims to allocate power generation to match load demand at
minimal possible cost while satisfying all the power units and
system constraints [1]. Small improvements in the unit output
scheduling can contribute to significant cost savings.
Furthermore, it provides useful information to fix market
clearing prices in deregulated power system. In the past
different approaches, including linear programming and
nonlinear programming are applied to ELD. Main drawback
of linear programming methods is that they are associated
with the piecewise linear cost approximation although they
are fast and reliable. The nonlinear programming approaches
like sequential quadratic programming [15] have a problem of
complexity.



GRAVITATIONAL SEARCH ALGORITHM
Traditional optimization methods such as those described
in references [8] are by far the most common optimization
tools used in the industry. However, these techniques can
encounter some difficulties such as getting trapped in local
minima, increasing computational complexity, and being not
applicable to certain objective functions. This calls for
developing a new class of solution methods that can
overcome these limitations. Heuristic optimization techniques
are fast nascent tools that can overcome most of the
shortcomings found in derivative-based techniques.


SIMULATION RESULTS AND DISCUSSION
The GSA and other programs were written using
MATLAB 7.0 and simulations were performed utilizing Intel
desktop with dual core processor. Extensive testing was
conducted to tune the parameters of the presented algorithms
in order to reach acceptable convergence characteristics. The
selected tuned parameters are:


CONCLUSION
Gravitational search intelligence technique to solve
economic load dispatch problem with valve point loading and
Kron’s transmission loss has been successfully implemented
and investigated. The simulation results reveal that in small
and medium search space like 3 and 6 unit systems, GSA
outperforms PSO, DE, GA and SQP approaches. Performance
of GS intelligence tool starts deteriorating when applied to
handle large search space problems like 13 and 40 unit
systems because of pre-convergence of movement of masses
in artificial world.