Multi-objective evolutionary optimization of vibration-driven energy harvesting and structural integrity in piezoelectric systems

A simulation-optimization system is established to improve performance of piezoelectric-based cantilever-based energy harvesters in realistic vibration environments. This is aimed at maximizing electrical energy production and structural integrity and wide range of operation. The approach is a multi-physics electromechanical modeling, nonlinear dynamic simulation, and multi-objective optimization with the help of NSGA-II, and surrogate models, including Artificial Neural Networks and Gaussian Process Regression, to enhance computational performance. Latin Hyperbole Sampling was used to analyze 1024 design configurations with harmonic (10–120 Hz) and stochastic excitations (0.1–1.8 g). The findings show that the power output rose to 4.85 mW and the stress rose to 112 MPa as beam length was increased whereas power output rose to 0.52 mW and stress to 38 MPa, thus performance tradeoffs were observed. Nonlinear effects increased bandwidth, which increased between 6.2 and 25.9 Hz. The highest power of 5.12 mW and fatigue life up to 2.8 × 107 cycles were obtained by optimization. The optimal design chosen produced 4.36 mW at 85 MPa stress, 24.8 Hz bandwidth and safety factor of 1.75. The accuracy and applicability were demonstrated by validation within a range of +7% in real-world situations.