Simulation-Based Optimization of a Piezoelectric Energy Harvester using Artificial Neural Networks and Genetic Algorithm

Abstract

The problem of finding optimal design parameters for a piezoelectric energy harvester is studied. An accurate iterative numerical simulation model based on Euler-Bernoulli beam theory is used as the basis for defining a simulation-based optimization problem. Due to the complexity of the simulation model, evaluation of the Objective Function (OF) is difficult and computationally expensive. In order to remedy this problem, an Artificial Neural Network (ANN) model is trained based on a dataset obtained from the iterative numerical simulation. ANN is then used during the optimization process instead of the original expensive-to-evaluate simulation model. Performance evaluation for the ANN is performed using a set of test data. Genetic Algorithm (GA) optimization method based on the trained ANN model is further developed and optimum system parameters are obtained for an energy harvester based on piezoelectric patches and cantilever aluminum beam.