Abstract Pipeline structural health monitoring is critical for global energy security, yet traditional bulk piezoelectric acoustic emission (AE) sensors are inherently bulky, narrow-band, and unsuitable for large-scale, distributed deployment. Here, we present a highly integrated, broadband microelectromechanical systems (MEMS) AE sensor based on ScAlN piezoelectric micromachined ultrasonic transducers (PMUTs) for intelligent pipeline damage monitoring. To address the acoustic impedance mismatch between the silicon-based micro-chip and external media, a composite acoustic matching layer comprising epoxy resin doped with 60 wt% Al 2 O 3 powder was engineered. This packaging strategy significantly enhances acoustic transmission efficiency while ensuring exceptional mechanical robustness in harsh environments. Systematic characterizations reveal that the fabricated MEMS AE sensor exhibits a broad bandwidth with displacement sensitivity exceeding 60 dB across 40–600 kHz, and a peak sensitivity of 88.4 dB at 335 kHz. In standard pencil lead break tests, it demonstrates a signal amplitude approximately twice that of commercial AE sensors. Furthermore, the device maintains stable performance under severe thermal shock cycling from −55 °C to 85 °C. By integrating this hardware with a Short-Time Fourier Transform (STFT) and a Residual Neural Network (ResNet)-based deep learning algorithm, we developed an intelligent pipeline monitoring system. The system successfully captured and classified the time-frequency characteristics of five distinct human-induced destructive behaviors with an outstanding recognition accuracy of 100%. This work provides a scalable, high-performance hardware-software paradigm for distributed structural health monitoring in extreme industrial environments.
Liu et al. (Thu,) studied this question.