Page 299 - Advances in Renewable Energies and Power Technologies

272 CHAPTER 8 Hybrid PV/Batteries Bank/Diesel Generator

FIGURE 8.10
Ambient temperature of a day in March in Tunisia.

3.2.6 Effect of Number of Population and Iteration on PSO Algorithm
Conversion
Fig. 8.11 illustrates the effect of number of population on the conversion of the PSO
algorithm. It is clear that the optimization algorithm does not converge at population
value that equals to 10. This can be deduced from the fact that the fitness function is
not varying during changing the iteration number. However, it converges at popula-
tion values of 20 and 30, where the fitness function varies during changing the iter-
ation number and then settles to a specific value. The population value of 20 is used
in the hybrid system sizing because there is clear variation of the fitness function at
different iteration numbers, which is not the case for population value of 30.
Fig. 8.12 shows the effect of maximum number of iterations on the conversion of
the PSO algorithm. A maximum number of 30 is used in the hybrid system sizing
because there is clear variation of the fitness function occurring at different itera-
tions, which is not there for the other choices of maximum number of iterations
of 20 and 35. The system sizing results differ for each chosen iteration and popula-
tion values. An excel spreadsheet was created to monitor the results of the hybrid
system sizing. The optimal parameters chosen for the different parts of the hybrid
energy system are summarized in Table 8.4 after choosing an iteration value of 30
and a population value of 20.

3.2.7 PSO Algorithm Results Summary
The population and iteration values affect the results achieved for the PSO algo-
rithm. Table 8.4 summarizes the optimal solutions achieved for an iteration value
of 30 and a population value of 20 when the optimization constraint is either mini-
mum emitted CO 2 pollution or overall system cost. It is clear that the overall cost of
the hybrid energy system when optimized using minimum pollution criterion is
higher by about 0.3% compared with the system when optimized using minimum