Optimization of hydraulically amplified electrostatic actuators based on an evolutionary strategy and finite element model to match the performance of the human triceps surae muscle fibres

Abstract

Artificial muscles are required to improve prosthetics, exoskeletons and other industrial robotics applications. Compared to other technologies like pneumatic actuators and shape memory alloys, Peano hydraulically amplified self-healing electrostatic (HASEL) actuators have the most similar stress output, work density and specific energy as biological muscle. This work designs an artificial muscle fibre that simulates human triceps surae muscle fibre using Peano-HASEL actuators. Analytical models are derived to calculate the capacitance and efficiency of the actuator. A three-stage methodology of an evolutionary strategy, finite element (FE) model and a severity analysis is proposed to optimize the geometrical parameters of the artificial muscle fibre, evaluate its performance, and identify potential risks. The finite element model created in this work produces a full multi-physics simulation coupling the electro-mechanical and fluid–solid interactions, which fits any liquid dielectric volume and for the first time simulates an entire activation cycle. The finite element model further simulates the force-length and force-velocity relationships not seen in previous work and facilitates understanding of the dynamics of the Peano-HASEL actuator, including damping characteristics. Plastic deformation and fluid pocket phenomenon due to materials are found during the simulation, which could give new insights for the analysis of electrical breakdown and mechanical stress failure. The three-stage design methodology described in this paper suggested that an artificial muscle with 80 mm width, 0.4 radian initial central angle, 20 um film thickness and 7.6 kV activation voltage would best match the human triceps surae muscle behaviour. The designed artificial muscle fibre can produce 80% of peak maximum strain (∼18.8%), have an efficiency of 22%, generate high loads within 5% strain and restore muscle-like performance beyond 5% strain.