Optimal Design of Tidal Power Generator Using Stochastic Optimization Techniques
Particle Swarm Optimization (PSO) and Genetic Algorithms (GA) are used
to reduce the cost of a permanent magnet synchronous generator with concentrated
windings for tidal power applications. Reducing the cost of the electrical
machine is one way of making tidal energy more competitive compared to traditional
sources of electricity.
Hybrid optimization combining PSO or GA with gradient based algorithms
seems to be suited for design of electrical machines. Results from optimization
with Matlab indicate that hybrid GA performs better than Hybrid PSO for
this kind of optimization problems. Hybrid GA shows better convergence, less
variance and shorter computation time than hybrid PSO.
Hybrid GA managed to converge to an average cost of the generator that is 5.2
% lower than what was reached by the hybrid PSO. Optimization results show
a variance that is 98.6 % lower for hybrid GA than it is for hybrid PSO. Moving
from a pure GA optimization to the hybrid version reduced the average cost
Parallel processing features are able to reduce the computation time of each
optimization up to 97 % for large problems. The time it took to compute a
GA problem with 2500 individuals was reduced from 12 hours to 21 minutes
by switching from a single-processor computer to a computer with 48 processor
cores. The run time for PSO with 400 particles and 100 iterations went from
18.5 hours to 74 minutes, a 93 % reduction.
Place, publisher, year, edition, pages
Institutt for elkraftteknikk , 2014. , 29 p.
ntnudaim:11263, MTENERG energi og miljø, Elektrisk energiteknikk
IdentifiersURN: urn:nbn:no:ntnu:diva-27239Local ID: ntnudaim:11263OAI: oai:DiVA.org:ntnu-27239DiVA: diva2:762912
Nilssen, Robert, ProfessorRøkke, Astrid