SENSOR HEALTH MONITORING BASED ON ONTOLOGIES. A SYSTEMATIC LITERATURE REVIEW

Kurzfassung

Flight research relies on precise and reliable measurement data for validation, collected from highly modified research aircraft. These aircraft integrate data from standard onboard systems and experimental sensor networks. Conventionally, the quality of this data has been systematically examined at the user-stage when the data gets processed for further work. Traditionally, the validation of sensor data is a highly customized and aircraft specific process, often requiring manual post-processing before the data can be used for aerophysical, structural, or aeroelastic analysis by domain-specific researchers. As this process requires the domain-specific knowledge of flight test engineers, it can take several weeks before the validation is finalized. Often, the flight test engineers also face obstacles in parameter monitoring in-flight. The extensively equipped aircraft of the German Aerospace Center record up to 3000 parameters. Therefore, the detection of sensor faults immediately during or after an experiment is hardly possible. In addition, the specific characteristics of the different research aircraft limits the ability to transfer fault detection methods across aircraft. The availability of structured system information via ontologies and metadata-modelling is progressively increasing by the introduction of metadata frameworks such as the Common Parametric Aircraft Configuration Schema (CPACS). System information shaped into standardized machine-readable formats can potentially be embedded into the process of detecting faults instead of manually reconfiguring systems for each parameter. This has the potential to significantly reduce development times of Fault Detection Algorithms. The current state of the art in systems to determine Sensor Health of complex systems will be examined in this paper. Existing approaches to metadata-based Fault Detection methods, Sensor Health Quality and Sensor Health Index Frameworks are considered and systematically reviewed using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. The results are then clustered and classified upon different criteria such as their flexibility, standardization and viability for a use case in Predictive Maintenance for Aircraft.