Large civil engineering structures, such as hydro electrical power plants, bridges, and airports, play an important role in today’s society, providing it with multiple services and capabilities. Therefore, failures in those structures can produce catastrophic events with wide-spread impact on human life, the environment, and on the economy. Reducing the risk of such failures requires proactively acting, which implies being able to continuously assess their safety, threats and vulnerabilities.
For example, structural assessment depends on accurate information about the concerned structure, which noways can be supported by updated information provided by sensors and other means to monitor dynamics. Structural monitoring relies thus on using a set of sensors to periodically acquire data as a means of measuring properties of a structure (e.g. displacement, extension). Originally, data acquisition was carried out manually using mechanical or electrical devices, which produced a discrete and limited set of values that could then be correlated with long-term behaviour series. Later, the development of electronic data loggers and programmable controllers capable of automatically acquiring and transmitting data, led to the deployment of permanent monitoring systems that generate large volumes of data. This data is fundamental to understand the dynamics of a structure, to detect deviations to its expected behaviour, and, to act accordingly in case of need.
Structural monitoring manages multiple types of different sensors deployed on a structure, which can vary in number from a few dozens to thousands. Thus, sensors must be efficiently characterized due to their diversity and quantity. Moreover, structural monitoring not only deals with the operation of sensors in terms of data acquisition, but must also consider dependencies between sensors, and the impact of supporting processes, such as maintenance, replacement, and calibration. Finally, considering the expected long lifetime of these infrastructures, the long-term preservation of data and the definition of assessment methods and techniques that can deal with multiple generations of technology are other relevant challenges.
However, critical infrastructures also can be monitored in a wider perspective, considering not only their intrinsic information, but also other contextual information, making it possible to design and apply more global risk management processes (considering for example threats due to extreme weather conditions, accidents, human events, etc.).
Semantic technology represents computational techniques that can used to address some of the problems behind risk management, safety assessment and structural monitoring, as they enable:
We have been exploring the application of semantic technology (including graph databases and ontologies) to analyze sensors performance, sensor dependencies, as well as determining their role in structural monitoring. We have, for example, applied these solutions to the analysis of a scenario provided by the Portuguese National Laboratory for Civil Engineering (LNEC). The application uses ontologies to represent deployed sensors, and logical reasoning to analyze correlations and dependencies between sensors, and to check their compliance against a set of requirements.
Structural monitoring can be classified as a “big data” problem due to
managed and integrated.
Big-data techniques provide the means to perform structural monitoring while handling the volume, velocity and variety of the underlying data. In particular, structural monitoring benefits from exploiting big-data storage (e.g. Cassandra) and processing (e.g. Spark/RSpark and Hadoop/rHadoop).
Check more publications here.