The application research of nuclear power mechanical equipment multi-fault feature extraction technologies based on material, energy and signal flow modeling

Due to the inherent complexities of nuclear power machinery equipment, which include intricate structures, numerous interfaces, and interconnected functions, failures within this equipment often exhibit dispersed, repetitive, and coupled characteristics . To address the challenges associated with extracting fault features and the subsequent difficulties in fault diagnosis and positioning in such mechanical systems, this paper introduces a multi-fault feature extraction method grounded in material, energy, and signal flow modeling. Utilizing condition monitoring data and expert insights, a sophisticated technique for multi-fault feature extraction is developed, incorporating function dependency modeling, fault parametric analysis, and the construction of a fault dependency matrix. This approach establishes a dependency matrix that correlates “multi-fault coupling failure modes and characteristic parameters” of typical mechanical equipment, enabling the systematic extraction and categorization of fault characteristic parameters. The efficacy and applicability of this key technology are validated, and recommendations for its optimization are proposed.