Fast and smart approach for buildings Risk Assessment using open accessible geostatistical data and data-driven classification methods - SmartRA

• Applicants: Ehsan Harirchian, Ph.D.; Professor Dr. Tom Lahmer
• Subject Area: Structural Engineering, Building Informatics and Construction Operation ; Applied Mechanics, Statics and Dynamics ; Methods in Artificial Intelligence and Machine Learning
• Term: since 2025
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 555132871

The advancement in technical achievements and big data accessibility, allows the assessment of damages (hazards) and associated challenges after natural hazards (e.g., floods or earthquakes) and the reduction of their impacts. An appropriate assessment method to detect vulnerable buildings makes it more manageable to investigate the seismic risk of buildings in a territory. Additionally, it helps with better city planning, retrofitting strategies, and preparation for pre-disaster rescue services. Contrary to advanced structural, dynamical analyses and time-consuming Finite Element Simulations, rapid visual screening (RVS) allows assessing a series of buildings with regards to their vulnerability due to natural hazards in a short time. With the help of such screening methods, one can prioritize buildings much faster, affordable, and simpler concerning their need for a comprehensive assessment. Additionally, an efficient, reliable, and fast classification of different levels of damage after a significant impact is possible. The immediate results of the RVS allow, in particular, the saving of resources and the reduction of risks and associated economic losses for the user and authorities such as crisis management or governmental agencies. The proposed project develops an automated workflow and digital method that is practical, quick, and reliable. A framework will be created that collects and evaluates building data from different sources like drone images, own developed mobile applications, and street-view images, e.g., from Google API. In this regard, we integrate image analysis and artificial intelligence, merely in the context of classification and uncertainty / multicriterial assessment methods, into a novel toolbox. The strategies to be developed and implemented will be evaluated at the end of the project w.r.t their costs, time consumption, scalability, and consistency compared to classical methods. The probability-theoretic methods and those of the data science, so-called soft computing (SC) approaches, are in particular characterized for their ability to a rapid assessment, reduction of human efforts, and resulting wrong interpretations. The direct results of the project will therefore eliminate the influence of human cognition and subjectivity while reducing effort, time, and costs.

• DFG Programme: Research Grants