Flexible piezoelectric force sensors with enhanced signal stability and interpretable data analysis

초록

The realization of flexible, lead-free piezoelectric systems with reliable electromechanical response and meaningful signal interpretation remains a critical challenge for wearable biomechanics and human–machine interface applications. In this study, a BaTiO3-based platform is engineered through controlled lattice modification and morphology tuning using Bi3 + and Mn4+ incorporation via a rapid microwave-assisted citrate–PVP process. This approach yields phase-pure, compositionally uniform spherical structures while retaining the tetragonal ferroelectric phase essential for effective polarization behavior. The coupled modification of A-site and B-site environments produces structural and defect-related changes consistent with reduced charge-screening effects and improved polarization stability, contributing to enhanced electromechanical consistency under repeated mechanical stimulation. When incorporated into a flexible PDMS matrix, the resulting Ba0.9Bi0.1Ti0.9Mn0.1O3 composite demonstrates a sensitivity of 3.18 V kPa−1 over a linear pressure range of 2–17 kPa (R2 = 0.99), with a detection limit of 2.82 kPa and peak voltage outputs reaching 63 V during cyclic loading. The fabricated device generates stable and distinct electrical responses under multiple deformation modes, including bending, compression, torsion, and impact conditions relevant to real-world biomechanical motion. To translate these electrical signals into actionable information, a time-series classification strategy based on the Diverse Representation Canonical Interval Forest (DrCIF) model is employed, enabling accurate discrimination of twelve discrete force levels (0.8–8.4 N) with an accuracy of 0.8486, without dependence on computationally intensive deep learning models.By combining controlled defect engineering, structural uniformity, and efficient data-driven signal interpretation, this work demonstrates a robust framework for high-resolution force sensing and interpretable signal processing in flexible, lead-free piezoelectric systems, with strong potential for wearable health monitoring and intelligent sensing applications.Code available at: https://github.com/Zahid672/Pressure_Sensor_Classification © 2026 Elsevier B.V. All rights are reserved, including those for text and data mining, AI training, and similar technologies.

키워드