In this paper, a novel optimization method, in which composed of the genetic algorithm, particle swarm optimization (GA–PSO) and improved gradient descent algorithm, are used to conduct a double-objective optimization for the U-shaped actuator. In the procedure of optimization, two objectives, i.e. force and displacement, and four main sizes are utilized. Before, the deep reactive-ion etch (DRIE) technology is applied to the fabrication of the U-shaped actuator. When different voltages are applied on the actuator, the displacement obtained from numerical calculation always shows a good agreement with that from experiment by edge detection algorithm. Similar phenomenon can be also seen when an external force supplied by the nanoindentation system FemtoTools in the experiment is loaded on the actuator. Based on the validated simulation model of the U-shaped actuator, the improved gradient descent method ensures its displacement very close to 50 µm (target displacement) while the GA–PSO algorithm is used to maximize the output force. In this procedure, the hybrid optimization method implemented by Matlab is incorporated into ANSYS simulation. Preliminary analysis shows that the displacement and force of the particles in each iteration concentrate together with the iteration growing. Fine convergence, whose velocity only depends on the number of particle in the algorithm, is also found in each optimization. Furthermore, the optimized actuators have homologous value of the size variables. At 15 V voltage, the displacement and largest output force of the U-shaped actuator are 2 mN and 50.1 µm, respectively. Finally, an actuator with 30% improvement of the output force is obtained when the displacement condition is meet. According to the optimization result and further parametric scanning simulation analysis, the design range and fabrication error of the sizes of the U-shaped actuator are obtained.
Read full abstract