A review in guided-ultrasonic-wave-based structural health monitoring: From fundamental theory to machine learning techniques

The development of structural health monitoring (SHM) techniques is of great importance to improve the structural efficiency and safety. With advantages of long propagation distances, high damage sensitivity, and economic feasibility, guided-ultrasonic-wave-based SHM is recognized as one of the most promising technologies for large-scale engineering structures. However, the propagation characteristics of guided ultrasonic waves in inservice engineering structures are highly complex, which results in difficulties in developing precise and efficient signal feature mining methods. The damage identification efficiency and reliability of existing guided ultrasonic wave methods cannot meet engineering requirements. With the development of machine learning (ML), numerous researchers have proposed improved ML methods that can be incorporated into guided ultrasonic wave diagnostic techniques for SHM of actual engineering structures. To highlight their contributions, this paper provides a state-of-the-art overview of the guided-wave-based SHM techniques enabled by ML methods. Accordingly, multiple stages required for ML-based guided ultrasonic wave techniques are discussed, including guided ultrasonic wave propagation modeling, guided ultrasonic wave data acquisition, wave signal preprocessing, guided wave data-based ML modeling, and physics-based ML modeling. By placing ML methods in the context of the guided-wave-based SHM for actual engineering structures, this paper also provides insights into future prospects and research strategies.

Automated summary

The development of structural health monitoring (SHM) techniques is important for improving structural efficiency and safety. Guided-ultrasonic-wave-based SHM is recognized as a promising technology due to its advantages. However, the complexity of guided ultrasonic wave propagation in engineering structures makes it difficult to develop precise and efficient signal feature mining methods. Existing guided ultrasonic wave methods lack efficiency and reliability. Machine learning (ML) has been proposed to enhance the guided ultrasonic wave diagnostic techniques for SHM. This paper provides an overview of ML-based guided-wave-based SHM techniques for actual engineering structures, including modeling, data acquisition, preprocessing, and ML modeling.