A systematic review on time-constrained ontology evolution in predictive maintenance

Abstract

With the modernization of industry and introduction of IoT, maintenance practices have been moving from reactive to proactive and predictive approaches. The identification of faults often relies on the analysis of real-time data provided by streams and unstructured sources. Ontologies have been applied to the maintenance field in order to add a semantic layer to the data and facilitate interoperability, and combined with other approaches for explainability and fault diagnosis, among others. In such a time-sensitive domain, it is important that ontologies go beyond static representations of the domain and allow not only for the incorporation of time related knowledge, but must also be able to adapt to new knowledge and evolve. This systematic review presents four research questions to provide a general understanding of the state of the art of the representation of time and ontology evolution in the predictive maintenance field. The results have shown that there are several ways of representing the evolution of knowledge that are fairly established and several specific evolutionary actions are discriminated and analyzed. Similarly, there is a diverse group of metrics that can be exploited to measure change and to establish evolutionary trends and even predict future stages of the ontology. Studies on the representation of time show us that it can be done either quantitative or qualitatively, with some approaches combining the two. Applications of these to the problem of ontology evolution are still in the open. Finally, results show that while applications of ontologies to the field of predictive maintenance are plenty, there are not many studies focusing on their evolution or in the effective application of their ability to reason with time constraints. The results obtained in this systematic review are particularly relevant for devising solutions that make use of the ontology's potential for time representation and evolution in the predictive maintenance field.

Development of an ontology-based asset information model for predictive maintenance in building facilities

“…temperature, humidity, CO2) and sensor data. With reference to the effectiveness of predictive maintenance, without clear and welldefined relationships, it becomes challenging to establish the cause-and-effect connections between different entities (Canito et al, 2022). For instance, if there is a failure in an HVAC system, the ontology should allow for tracing back the possible causes, which could range from environmental factors to previous maintenance activities.…”

Revision of prioritized $$\mathcal {E}\mathcal {L}$$ ontologies

“…An interesting result we display in this paper is that revision operation is done efficiently in polynomial time. Revision of prioritized EL ontologies with uncertain and inconsistent input information has several applications in different fields such as ontology reasoning and maintenance [38], semantic web and knowledge graphs [39], artificial intelligence and machine learning [40], Natural Language Processing (NLP) [41], medical informatics [42].…”

Review of fault detection techniques for predictive maintenance

“…All these predictive maintenance techniques were developed as part of industry 4.0, which tries to move all the maintenance activities from reactive approaches into proactive and predictive methods (Canito et al. , 2021).…”

“…All these predictive maintenance techniques were developed as part of industry 4.0, which tries to move all the maintenance activities from reactive approaches into proactive and predictive methods (Canito et al, 2021). Since these predictive maintenance operations deal with real-time data streams, cyber-physical systems can be utilised to communicate the data generated from the individual component sensors to the central system.…”

“…Since these predictive maintenance operations deal with real-time data streams, cyber-physical systems can be utilised to communicate the data generated from the individual component sensors to the central system. Afterwards, all these generated data can be stored inside cloud storage (Canito et al, 2021). These new enhancements lead to new architectural paradigms such as Cloud, Fog or Edge systems (Erhan et al, 2021).…”