Integrating Vibration, Motor Current, and Wear Particle Analysis with Machine Operating State for On-line Machinery Prognostics/Diagnostics Systems (MPROS)

Integration of system process information with expert system vibration analysis to improve the accuracy of rotating machinery fault detection in on-line condition monitoring systems is the main focus of this paper. An overview of the condition based maintenance system MPROS (Machinery Prognostics/Diagnostics System) being developed for the US Navy is presented. Details are provided regarding several currently operating prototype systems installed on both shipboard and land based air conditioning plants. Finally a case study is presented which utilizes data from a prototype system to examine several of the relationships between air conditioning compressor vibration signatures and system process parameters. Understanding and utilizing these relationships will enable accurate diagnostics and provide the in-depth data that is needed to support prognostics. Introductionare many systems available today that provide on-line or continuous monitoring of rotating plant machinery for Condition Based Maintenance (CBM). These systems range from relatively simple vibration switches which can provide alarm and machine shutdown information, to very sophisticated multi-channel monitoring systems which provide spectral processing, feature extraction, and expert data analysis to identify specific fault conditions and assess fault severity. It has been found that current on-line CBM systems do not effectively utilize process information when evaluating machine condition from vibration, oil or motor current data. Many systems accept basic process information such as a signal from a PLC (Programmable Logic Controller) which indicates the machine is operating in a predefined state, usually at full system load. But what happens when the machine being monitored only occasionally is operated at full load, or it is subject to a wide range of operating states that have a dramatic impact on the vibration and motor current signatures of the machine? Periodic and trend analysis of the machine under these conditions usually results in a false diagnosis of a fault condition or in an inaccurate severity assessment for an existing fault. A primary goal of the MPROS project is to accurately evaluate machine condition, identify fault severities, and predict machine remaining life under a broad range of machine operating states. It is believed that this can be accomplished by identifying the relationships that exist between a machineâs process and the vibration and motor current signatures of the machine. This information can then be integrated into the machine fault models during fault identification and for adjustment of a fault severity assessment.