C. Yildirim, R. E. Lillo, A. M. Franco Pereira
Effective condition monitoring and fault classification are essential for ensuring the reliability of complex engineering systems. Advances in big data, machine learning, and IoT have enabled multi-sensor diagnostics, but handling the multi-fault conditions remains a challenge. This study presents a novel Multivariate Functional Data Analysis (FDA) framework based on Multivariate Functional Principal Component Analysis for fault diagnosis in hydraulic systems. Experimental results show that our approach achieves high classification accuracy using raw multi-sensor data, highlighting multivariate FDA as a powerful tool for condition monitoring and multi-fault classification.
Palabras clave: Functional Data Analysis (FDA), Multivariate Functional Principal Component Analysis, Fault Diagnosis and Classification, Condition Monitoring, Hydraulic Systems
Programado
Tratamiento y análisis de Big Data (TABiDa2)
10 de junio de 2025 17:10
Sala 3. Maria Rúbies Garrofé