Multiobjective Optimal Design of Symmetrical Langevin Transducers

The optimal design of symmetrical Langevin transducers using multiobjective evolutionary algorithms (MOEAs) is studied. The design problem is formulated as a constrained multiobjective optimization problem (MOP) involving continuous design variables (dimensions of the transducer) and discrete design variables (the number of piezoelectric rings and material types). The maximum vibration amplitude and the minimum electrical input power are considered as optimization objectives. The MOP is then solved using four MOEAs. In order to determine the preferred MOEA the four MOEAs are evaluated based on the obtained non-dominated solutions. The result shows the strength Pareto evolutionary algorithm (SPEA) has a better overall performance and should be preferred in the optimal design of the transducers. Furthermore, the size of Pareto-optimal set can be reduced by the clustering technique and the preferred design can be determined according to the high-level information.