Structural health monitoring based on a physical reservoir with frequency virtual nodes

This study explores the applicability of physical reservoir computing for structural health monitoring by treating the target structure as a physical reservoir layer. By introducing frequency virtual nodes, the proposed method reduces the number of required sensors while maintaining high detection accuracy. Physical reservoir computing is a computational framework that leverages the dynamics of physical systems as computational resources for processing time series data. Frequency virtual nodes are network nodes created by dividing the response of a system into segments across different frequencies, effectively increasing the dimensionality of the reservoir. The structure with virtual nodes acts as a high-dimensional computational resource, enabling damage detection through changes in its physical dynamics. To validate the approach, we conduct numerical simulations on a clamped plate using a single sensor and actuator. As a result, the proposed method achieves comparable damage detection accuracy to conventional neural networks, while significantly reducing the training costs, demonstrating the potential of a novel framework of structural health monitoring based on physical reservoir computing.