Adaptive ANN-Based MPPT Control for Piezoelectric Vibration Energy Harvesting System

Abstract

Piezoelectric energy harvesting systems often suffer from suboptimal power extraction due to the time-varying nature of mechanical vibrations and the nonlinear impedance characteristics of piezoelectric materials. We propose a real-time artificial neural network (ANN)-based maximum power point tracking (MPPT) controller to dynamically optimize the power transfer from a piezoelectric source to a load. The ANN directly maps the instantaneous piezoelectric voltage to the optimal duty cycle of a buck converter. The proposed method employs a single hidden layer with 10 nodes, ensuring computational efficiency while capturing the nonlinear relationship between the input voltage and the optimal duty cycle. The system integrates a full wave rectifier to convert the alternating-current output of the piezoelectric bender into a direct-current voltage, which the ANN then processes to generate the control signal for the pulse-width modulation (PWM) gate driver. Experimental validation demonstrates that the ANN-based MPPT achieves higher power extraction efficiency than conventional perturb-and-observe methods, particularly under rapidly changing mechanical excitation. Furthermore, the approach stabilises the output voltage while maintaining near-maximum power transfer, making it suitable for low-power IoT applications where energy efficiency is critical. The simplicity and robustness of the proposed solution highlight its potential for practical deployment in real-world energy harvesting scenarios.