Diagnostic Analyzer for Gearboxes (DAG): User's Guide. Version 3.1 for Microsoft Windows 3.1.
NATIONAL AERONAUTICS AND SPACE ADMINISTRATION CLEVELAND OH LEWIS RESEARCH CEN TER
Pagination or Media Count:
This documentation describes the Diagnostic Analyzer for Gearboxes DAG software for performing fault diagnosis of gearboxes. First, the user would construct a graphical representation of the gearbox using the gear, bearing, shaft, and sensor tools contained in the DAG software. Next, a set of vibration features obtained by processing the vibration signals recorded from the gearbox using a signal analyzer is required. Given this information, the DAG software uses an unsupervised neural network referred to as the Fault Detection Network FDN to identify the occurrence of faults, and a pattern classifier called Single Category-Based Classifier SCBC for abnormality scaling of individual vibration features. The abnormality-scaled vibration features are then used as inputs to a Structure-Based Connectionist Network SBCN for identifying faults in gearbox subsystems and components. The weights of the SBCN represent its diagnostic knowledge and are derived from the structure of the gearbox graphically presented in DAG. The outputs of SBCN are fault possibility values between 0 and 1 for individual subsystems and components in the gearbox with a 1 representing a definite fault and a 0 representing normality. This manual describes the steps involved in creating the diagnostic gearbox model, along with the options and analysis tools of the DAG software.
- Computer Programming and Software