RESUMO
In the detection of slipping anomalies in viscoelastic sandwich cylindrical structures (VSCS), conventional methods may encounter challenges due to the extremely rare and weak nature of slipping signals. This study focuses on normal signals and introduces unsupervised graph representation learning (UGRL) with discriminative embedding similarity for VSCS's detection. UGRL involves data preprocessing, model embedding, and matrix reconstructing. Association graphs are constructed based on sample similarities for yielding adjacency and attribute matrices. Subsequently, the matrices undergo embedding and reconstruction via various network modules to enhance graph data characterization. Detection indicators are derived by calculating embedding similarities and reconstruction errors, and thresholds are constructed using these indicators to enable efficient anomaly detection. The experiments in VSCS slipping dataset effectively indicate the superiority of the proposed method.
RESUMO
This paper investigates one eigenvalue decomposition-based source number estimation method, and three information-based source number estimation methods, namely the Akaike Information Criterion (AIC), Minimum Description Length (MDL) and Bayesian Information Criterion (BIC), and improves BIC as Improved BIC (IBIC) to make it more efficient and easier for calculation. The performances of the abovementioned source number estimation methods are studied comparatively with numerical case studies, which contain a linear superposition case and a both linear superposition and nonlinear modulation mixing case. A test bed with three sound sources is constructed to test the performances of these methods on mechanical systems, and source separation is carried out to validate the effectiveness of the experimental studies. This work can benefit model order selection, complexity analysis of a system, and applications of source separation to mechanical systems for condition monitoring and fault diagnosis purposes.
