Anagrafe della ricerca

The dramatic seismic events occurred in Italy in past 15 years, have clearly highlighted that reconstruction and restoration costs of large territories are unsustainable by the governments. Also, cities downtime for reconstruction has a dramatic societal impact, such as depopulation and disaggregation of social tissue. To mitigate the effects of these catastrophic events, the only way known is the prevention by reduction of seismic risk. This is intended as a diffuse incentive to reinforcement and retrofitting of buildings and infrastructure, to bring seismic risk below acceptable thresholds. However, a diffuse seismic risk reduction still would involve significant costs and downtime and also a massive employment of raw material with consequent increase of CO2 emissions. This research project gathers all these aspects, proposing a breakdown way of facing the issue through the aid of artificial intelligence (AI) soft computing techniques. The idea developed in this research proposal is to bring artificial intelligence techniques to become a real aid to the current strategic decision processes (what to do and where). The project aims at implementing an artificial intelligence architecture supporting users (engineers or public administrators) in designing sustainable and strategic seismic retrofitting interventions to protect structures and infrastructures against the consequences of earthquakes. AI algorithms are here proposed as a chance to: Solve large and difficult optimization problems such as retrofitting cost minimization or reinforcing material volume minimization in seismic upgrading design. Provide estimations of seismic risk reduction costs of single structures as a function of simplified inputs. The first goal will be achieved by implementing metaheuristic bio-inspired algorithms, based on the Darwinian concept of natural selection. In a few words a metaheuristic optimization framework will consider a target solution (e.g. the one with the lowest cost) as the strongest individual. The second goals will be achieved by implementing an Artificial Neural Networks (ANNs). ANNs are computational data-driven methods based on the idea to mimic the learning and memory capability of the human brain. The ANN will be trained by data and will allow doing simple estimations of seismic risk reduction costs as a function of a limited number of data (e.g. seismic hazards, number of floors, surface, structural typology etc). Both the goals have relevant fallouts in engineering design practice and strategic decision processes. The challenge to face is making AI-based retrofitting design a stable reality in civil engineering, so that interventions will cost less, will be more effective and will be less invasive, and more and more buildings and infrastructures could be retrofitted with less capitals, less time, less material waste.