The transition to CBM (Condition Based Maintenance) requires a collaborative effort on a massive scale and will require new tools, and embedded diagnosis systems. The transition to CBM involves the construction of data-centric, platform-operating capabilities built around carefully developed algorithms. This will allow maintenance workers in the field, support analysts, and engineers the ability to simultaneously, and in real-time, translate conditional data and proactively respond to maintenance needs based on the actual condition.
Computerized Maintenance Management Systems (CMMS) are the core of traditional maintenance record-keeping practices and often facilitate the usage of textual descriptions of faults and actions performed on a component as part of a system. Sensors are capable of directly monitoring component parameters; however, attempts to link observed CMMS events to sensor measurements have been limited. To integrate the two-disparate data types a relational database which tags events from both data sets with location, severity, and rarity parameters. This metadata should be extracted from the CMMS in textual descriptions, and sensor records must be processed using statistical analyses.
Computerized Maintenance Management Systems (CMMS) records information such as usage, failures, servicing or repairs, and inventory control. The underlying structure is typically heavily regulated, allowing for a large base of consistently structured data. These systems are the core of traditional scheduled maintenance practices and rely on historical data to make modifications to regulated maintenance actions.
Condition sensors collect quantitative data to assist in the identification of imminent or already occurred faults. Sensors are combined with a data acquisition system. Currently, there exists no standardization in the way data is collected across platforms or vendors. CBM will only works if there are means by which these data sets can be consistently and reliably merged.
A full integration requires a form of interfacing between observed maintenance and sensor data. The overall goal is to automate the process of linking events to datasets. This begins with: historical data collection, data abstraction and data analysis in an attempt to bridge the gap between the data types by allowing for the proximity, severity, and rarity of events across datasets to be evaluated. This allows for an objective determination of parts prone to failure and an evaluation of sensor effectiveness in monitoring those regions.
Modern MMS information is stored in relational databases. This format is appropriate for an integration investigation since there are a large number of software tools available to query and investigate the tables. For the historical analysis, only certain fields are required. Importing sensor data into a relational database is challenging, since each type of sensor generates different data classes, sampling rates, and number of compiled indicators. Furthermore, each manufacturer stores the collected information in unique proprietary formats, requiring platform-specific importation software to be written.
When both the CMMS and the sensor data co-exist within a single database they only have two fields in common: tag number and date. Given a maintenance fault or action can only be accomplished through the compilation of overlapping metadata. For these components, further analysis must be performed on the raw data outside the CMMS to discover new algorithms for condition indicator