setting an ideal alternative (which will never be reached) for each of the
criteria as the objective
446
and measuring the fulfilment of this objective via the fitness function.
447
In addition, like
other metaheuristic techniques, Particle Swarm Optimization, is distinguished by
448
its capacity to find an optimal solution (unknown until that moment) for complex, real-world
449
problems. Therefore, the ideal or anti-ideal solutions have not been found prior to running the
450
PSO algorithm, and they are probably not found in any case.
For this reason, the developed
451
methodology includes an initial step in which the hypothetical positive-ideal and negative-ideal
452
solutions are generated artificially (Luna et al., 2019b). To do so without incurring a high
453
computational cost, a hypothetical solution is generated each time whose aim is to exploit
the full
454
potential of the farm; i.e. seeding as soon as possible and harvesting on the last day for each feed
455
alternative. This hypothetical solution is then multiplied by a supplement of ±75%, assuming that
456
the PSO can find an alternative with better results, but not as good as 75% better.
457
In the
present case, the results shown in Table 3 were found in the initial step and “+ideal” and “-
458
ideal” were estimated from these results as explained previously.
459
Criteria
Do'stlaringiz bilan baham: