The pursuit of sustainable green energy has intensified the demand for self-powered materials, which are capable of efficiently converting ambient mechanical energy into electrical signals. Herein, polyvinylidene fluoride (PVDF) fabricated by electro-assisted fused deposition modeling (E-FDM) 3D printing is a firm candidate due to the exceptional electroactivity and excellent flexibility. Unlike conventional fabrication methods struggling with dimension control, a physics-informed neural network (PINN) framework was developed for the first time to predict and control PVDF fibers. PINN successfully integrates physical interpretability with data-driven flexibility, achieving high precision predictions (R2 = 0.97, MSE = 0.0016, MAE = 0.034) with improved morphological control. In particular, the results showed that PVDF 50 μm fibers exhibited more than 3-fold improvement in piezoelectric β-phase content when compared to the 3D-printed counterparts without the external electric field. Correspondingly, the PVDF sensor achieved an excellent sensitivity of 130.9 mV/kPa and a fast response time of 35 ms. The PINN-based approach offers a scalable manufacturing strategy for the fabrication of PVDF sensors with tailored electromechanical properties, paving the way for flexible electronics and intelligent sensing systems.
Wang et al. (Mon,) studied this question.