Title :
Hybrid algorithm for multi-objective optimization of PMSM using massively distributed finite element analysis
Author :
Krotsch, Jens ; Piepenbreier, Bernhard
Author_Institution :
Dept. of Electr. Drives & Machines, Univ. of Erlangen-Nuremberg, Erlangen, Germany
Abstract :
A hybrid multi-objective optimization algorithm based on partial aspects of evolution strategy combining stochastic and deterministic elements with the aim of a high efficiency and high scalability suitable for massively distributed finite element analysis is investigated. The selection and generation process of solution candidates depend on density in design variable and objective space. New solution candidates are generated by stochastic variation but also systematically using a triangular grid. With respect to effective design analysis an approximation method is used. The proposed algorithm is evaluated using test cases and is successfully applied to dimension and shape optimization of a permanent magnet synchronous motor.
Keywords :
approximation theory; finite element analysis; machine theory; optimisation; permanent magnet motors; stochastic processes; synchronous motors; PMSM; approximation method; distributed finite element analysis; multiobjective shape optimization; permanent magnet synchronous motor; stochastic elements; stochastic variation; triangular grid; Algorithm design and analysis; Approximation methods; Finite element methods; Magnetic analysis; Mesh generation; Scalability; Shape; Stochastic processes; Stochastic systems; Testing; Multi-objective optimization; cogging-torque reduction; design optimization; distributed analysis; grid-based design generation; hybrid algorithm; objective value approximation; permanent magnet motor;
Conference_Titel :
Optimization of Electrical and Electronic Equipment (OPTIM), 2010 12th International Conference on
Conference_Location :
Basov
Print_ISBN :
978-1-4244-7019-8
DOI :
10.1109/OPTIM.2010.5510429
