Optimizing piezoelectric vibration-based energy harvester using numerical-analytical method and soft computing algorithms

Abstract

Tuning and optimizing of piezoelectric vibration-based energy harvesters (PVEH) is essential to render sufficient amount of energy. As a modification to the tuning mass and dynamic magnifier of conventional PVEH, a novel integrated multisystem of cantilever-oscillator-spring is proposed in which the vibratory analysis discloses significant mutation in resonance frequency depending on the oscillator mass and spring constant values, showing hyper-tuning capability. Obtaining the maximum extractable amount of electric voltage is the ultimate goal which is an optimization problem with oscillator mass and spring stiffness as design parameters. The extended Hamilton’s principle along with the Galerkin modal decomposition techniques are adopted to find analytical-numerical response of the system undergoing harmonic base excitations. To optimize the voltage frequency response function (FRF), global evolutionary optimization algorithm is adopted. The closed-form voltage function is a hard-to-evaluate and computationally-expensive function. To overcome such issues, soft computing techniques is adopted. Using adaptive neuro fuzzy logic (ANFIS), a regressor model is designed to execute function evaluations in the genetic optimization procedure. Fuzzy inference system (FIS) is developed using 64 fuzzy ules derived from Gaussian and Gaussian-Bell shaped membership functions (MFs). Such a regressor model is utilized in the genetic algorithm launching with 200 iterations and 50 populations. It is observed that using roulette wheel, tournament, and random selection methods; rs= 100, rm= 2 are found as the optimal design parameter values. To validate the correctness of the implemented soft computing algorithm; the optimal voltage FRF is obtained using the closed-form analytical-numerical solution and compared with random case studies. It is shown that the nominated optimal values render the most obtainable amount of voltage. Eventually, it is inferable that the spring-mass subsystem integration with the cantilever energy harvester, drastically improves the amount of harnessed voltage. Furthermore, optimization of such integrated multi-system via soft computing techniques results in the maximum amount of harvestable voltage.