Information Technology Reference
In-Depth Information
are willing to significantly decrease their utility with respect to the improvement of the
global welfare of the system, we encountered an expected global increase in utility, but
a considerable variation in the allocation of utilities in the system. Instead, when only
lower decrease in performance is accepted by each coalition the results obtained are
plotted in Figure 2. The diagrams of Figure 2 illustrate a histogram representation of
the coalitions' utilities discretised in increasing order of their worth. It can be seen that
a 20% increase of
reduced significantly the number of coalitions operating at high
efficiency denoted by the first column of the histogram, while the number of coalitions
operating at lower levels of efficiency has been increased. The results emphasized that
the best performance
5
μ
in the vicinity of
0.2 . Somewhat surprisingly, what the experiments show is that being willing to accept
lower efficiencies in the benefit of the global performance is only advantageous to a
certain extent. In actuality, there is a trade-off to be taken into account. Although the
overall system utility increases, the ratio between the number of coalitions with low
utility and those with high utility is increasing as well. So, for assesing the efficiency
of the system not only should we be interested in the global utility, but also in having a
uniform distribution of high utilities for the majority of the coalitions.
Future work will further look to a greater extent at the electrical features of the power
system and incorporate in a more factual form load-flow computation analyses, that
verifies for contingencies and maintain the system within its operational limits. Also,
taking into account that at present the proliferation of DERs in the grid is still yet to
achieve an adequate level, we intent to address in our future work, at a more granular
level, the most suitable techniques for efficiently deploying such devices throughout the
topology of the grid.
of the system was obtained for values of
μ
6
Conclusions
As a proof of concept, our work has introduced a dynamic coalition-based model de-
ployed in distributed environments of negotiating agents. The adaptation mechanism
introduced performs an open-ended adaptation of groups of organizational agents, con-
verging towards stable configurations. In particular, we have highlighted the applicabil-
ity of this approach through the design of a distributed adaptive scheme for the smart
electricity grid. This process resulted in virtual partitions of the grid that would be able
to commit to a steady and robust generation profile requiring less energy from real
power plants especially during high-demand periods and providing a mechanism able
of reducing the complexity of the management process.
5
In our acceptation of best performance we restrict the results to a number of threshhold values.
In terms of utility distribution, namely we would like the number of coalitions pertained by the
highest utility class(last column of Figure 2) to represent a minimum of 50% of all coalitions,
while the remaining classes to be below the limit of 10%. Moreover what we have achieved
is to maintain the inferior fraction of lower utility classes, each below 5% of the total num-
ber of coalitions. In terms of average percent increase in social welfare we impose an 80%
improvement.
