Automatic Code Generation of Data Visualization for Structural Health Monitoring

Abstract

Structural Health Monitoring (SHM) aims at detecting, localizing, and characterizing damages in civil, mechanical, and aerospatial structures, which are hardly detectable in visual inspections. The collection, analysis, and visualization of data captured by sensors installed on these structures can be strongly supported by modern techniques of Data Science. In particular, the visualization of these data provides valuable help to experts on structural health and decision makers on preventive and corrective maintenance. Unfortunately, existing systems of data visualization still demand those stakeholders for a high level of software programming skills to take full advantage of visual and interactive exploration of data that sensors capture and output. This work introduces a model-driven approach to develop data visualization in the domain of structural health monitoring, in particular, of bridges. This approach is based on the definition of a Domain Specific Language (DSL) that describes the main concepts of an infrastructure of sensors typically used in SHM, along with common graphics and visual alternatives of data visualization. This DSL is instantiated by a modeling language, composed of a metamodel, a visual representation of concepts, and a set of model-to-text transformation rules. In this way, non-programmers can implement their own data visualization from a graphical and intuitive design, by automatically generating the corresponding code. This approach was implemented in a modeling and code-generation tool, called Vis4bridge, whose usability and output were successfully evaluated through the development of tasks and case studies.