AI4SUSCOMP

ABSTRACT

Rehabilitation of existing buildings is one of Europes commitments on its roadmap for the progressive decarbonisation of the industry. Rehabilitation techniques such as strengthening extend the service life of a building and, in turn, minimize the amount of waste generation from complete demolition, thus reducing the consumption of natural resources required for building redevelopment. Furthermore, suitable selection of durable and environmentally friendly materials for building strengthening will contribute to optimize economic resources and reduce the environmental impacts. Likewise, an efficient strategy of the strengthening + sustainable materials combination should extend the useful life of the structures under different conditions as much as possible satisfying the safety requirements. The achievement of this purpose will be only possible by automatically analysing measured data obtained from monitoring devices in the strengthened structure to identify anomalies almost in real-time. According to the previous paragraph, the project underlies in the synergetic implementation of four points:

 a) exploration of a novel generation of innovative strategies for structural strengthening with composite materials in order to extend the life of concrete structures and optimize the use of the materials employed

 b) application and study of the performance of more sustainable materials for the strengthening

 c) study of environmental effects to enhance the durability of the proposed strengthening systems

 d) development of advanced artificial intelligence (AI) methods able to monitor and analyse through different approaches the performance of the proposed strategies in a) and b) under different environmental and loading conditions.

These methods should trace the structural behaviour by providing insights into issues we are not aware of.

Its main focus is the investigation of innovative strengthening techniques of concrete structures using composite materials, under different loading and environmental conditions. Simultaneously to this objective and with the purpose of contributing to its achievement, a deepening and advancement of knowledge on novel advanced deep and machine learning-based structural health monitoring, when applied to a challenging and complex problem like this must be performed. Only through the deep knowledge of these strengthening systems, their performance might be optimized with the purpose of extending more and more their application with safety. Together with the development of novel machine and deep learning techniques to analyze massive data coming from different sources to identify patterns, and detect & predict potential failures of the tested structures, the use of transfer learning techniques will be explored to evaluate the possibility of adapting classifiers built on well-known labeled data sets (experimental and numerical) to new scenario including FRP strengthened elements with unlabeled data set.

It is also expected that some of the conclusions derived from this project can also be extended to the monitoring and analysis of other types of structures, as well as to other thematic areas in which analysis and diagnosis capacity is required.