PSI - Issue 44

Agnese Natali et al. / Procedia Structural Integrity 44 (2023) 2012–2019 Agnese Natali et al./ Structural Integrity Procedia 00 (2022) 000–000

2017

6

For the Level 0, the form with the data to be collected is at disposal. The computerized form has the same structure of the one provided in paper format from the Guidelines, but with an intuitive organization of the data. All the data necessary for the further steps are automatically selected by the system algorithms. It is also possible to attach files, when necessary (depending on the type of data). For the Level 1, the software is equipped with all the forms to support the visual inspections for all the risks to be analyzed. The structure of the forms is the same of those provided by the Guidelines. In line with the Operating Instructions, it is possible to make pictures to be directly assigned and connected to the object of interest. Referring to the forms about the damages of the components – to be filled for the evaluation of the conservation status and the level of damage of the structure for the vulnerability of the structural risk – they provide smart buttons to allow the correct determination of the level of defectiveness (for the next level), in line with the procedure provided by the Operating Instructions. The Level 2 is totally automated, but the algorithm for the evaluation of the WaC work only if all the data from the previous levels are inserted. If some necessary data are missing, the WaC is not evaluated, so the operator is mandatorily called to insert all the data to finalize the procedure. At the end of it, the software provides, besides the WaCs for the defining parameters and the risks, the full development of the logical flows, in order to highlight the relevance of each parameters in the final evaluation of the WaCs. Basically, the operator has to provide to the software the entry data only, and the post-process is fully handled by the software algorithms. This guarantee the fully correct and objective application of the procedure. • Conditions of use and responsibility issues. The scientific content is not the only relevant aspect to be considered in the software. Indeed, the software considers also that the process of classification may involve different persons (users), both from the organization side and considering the different scientific fields involved. Starting form the first point, INSPICIO gives the possibility to create different users with different functions: the users manager , the activities manager and the operator . The users manager is the supervisor of the whole process and he is only one that can create all the other users, so assigns roles and tasks. The activities manager is the one that prepares all the forms for the execution of the Level 0 and Level 1 activities. He also assigns each form to the person that will have to perform the activity (the operator ). The operator is responsible only of the activity and of the form assigned. Each form can be modified only by the operators that are assigned to this activity. All the other users (both at equal and higher level) can only see the filled forms but not put any modifications. It is possible to assign more operators to a single form, considering that different expertise may be required for the same form (different sources of risk are involved in the classification process), and in this case the form can be also filled simultaneously by the different operators. This structure is made to accomplish different company organisations and to respect the different and vast know how needed for the execution of the activity. Indeed, companies can more complex, meaning that supervision, management, assignation and execution are divided, or be smaller, meaning that all these activities may be handled by just one person. This also implies sectorialisation and right attribution of responsibilities, which is necessary for the kind of activity to be performed. All these aspects are handled by a structured technological implementation and by means of an eXplainable Artificial Intelligence (XAI) algorithm, a new expert system that overcomes the usual pitfalls of conventional “black box” approaches. This improves transparency and trust in its results and consequently in the decision-making processes, as explained in the following paragraph. 4. Technical background and development of the software 4.1. Technological implementation The development of the software adhered to the best practices in the implementation of distributed mobile and cloud-based applications, with special consideration to the ease of data entry on the field, and to the need for explainability of the results of the algorithms prescribed by the Guidelines. The application consists of two separate modules: a web-based User Interface (UI) and a backend acting as a server to persist and share the data between multiple users. The UI is a single page web application implemented