Design Optimization of Smart Laminated Composite for Energy Harvesting Through Machine Learning and Metaheuristic Algorithm

The aim of this paper is to improve the performance of the vibration energy harvester beam made of laminated composite material by conducting an optimization and developing a surrogate model that includes an artificial neural network (ANN) model along with genetic algorithm (GA). The parametric analysis of the laminated composite substrate has been carried out using finite element analysis (FEA) to develop the ANN model. In this novel approach for multi-objective optimization for designing the beam, the objective functions were the maximization of the displacement and minimization of the operational natural frequency of the harvester. The FE model of the composite-laminated beam and the unimorph harvester beam was validated with the results available in the reference paper and considered for further study. The ply angle of the laminated composite, lamina thickness, and the modulus of the composite material were considered as the design variables. The validations of the optimal results, obtained from GA-based optimization using ANN surrogate models developed from finite element analysis as the objective functions, are done using FEA. The results show that the optimal design models for all glass fiber composites, carbon fiber composites, and glass-carbon hybrid composites could produce a greater amount of voltage 133.12 V, 313 V, and 324 V, respectively. They could also produce 13.63 mW, 29.74 mW, and 36.36 mW power at their optimal resistance load conditions, because of the improvements in the displacement and also as they can be operated in low-frequency environments.