MOST - Spoke 4
Type
PNRR
CN4 - MOST (Mobilità Sostenibile)
Description
The Task 3.1.2 ((Algorithms and procedures for monitoring and smart predictive maintenance of main subsystems of railway vehicles using both model-based and artificial intelligence approaches), is focused on developing innovative data-driven and model-based methods for monitoring, diagnostics and prognostics of railway vehicles to improve sustainability and maintenance management. Modern Information Technologies and, in particular, Industrial Internet of Things, Fog/Edge computing, and Artificial Intelligence/Machine Learning (AI/ML), are key factors in the digitization of tramway and railway transport systems, with the objective to improve the effectiveness of vehicles monitoring and management techniques, to improve predictive maintenance and failure analysis techniques. In this context, applications must adhere to stringent safety, security, and real-time requirements, imposed by regulations, market, and users’ expectations. The railway industry is required to provide evidence about application isolation properties (temporal, memory, fault), with the aim to avoid temporary denial of service and possible catastrophic consequences. An effective way for isolation is represented by mixed-criticality systems, able to consolidate several domains of different criticality levels, on top of common hardware platforms, while reducing hardware SWaP-C factors (size, weight, power, and cost) and virtualization is expected to play a crucial role in this context. Indeed, combined with the availability of modern embedded platforms, virtualization is becoming a prominent way to easily reduce SWaP-C factors by consolidating multiple complex software on the same system-on-a-chip (SoC) in a flexible way. The first line of UNINA research aims to leverage virtualization technologies in the context of the digitization of tramway and railway transport systems, will include:
- The application of AI/ML techniques in diagnosis of Remaining Useful Lifetime prediction and unveiling main causes of failures in the context of tramway and railway transport systems.
- The use of a proper virtualization approach and solution for integrating tramway/railway-relevant scenarios. These include integrating legacy applications, developing AI/ML engines and novel monitoring solutions for predictive maintenance.
- The assessment of deployed scenarios in terms of safety, isolation and real timeliness requirements. In some cases, the model-based approach may be a preferable methodology for Smart Predictive Diagnostics (SPD). This occurs mainly when the mathematical models of the subsystems are sufficiently approximated and when the very rapid temporal variation of the quantities of interest give rise to large amount of data, not easy to handle in a shorter time intervals. Therefore, it becomes necessary the development of a second line of research, where the activity will provide the development of the mathematical models of carriages and propulsion units during the most common faults, identifying a clear pre-fault indicator, able to detect and isolate a certain fault. The mathematical models developed will be validated using some experimental tests carried out on scale simulator. Starting from model-based algorithms, the SPD aims to develop local pre-processing techniques capable of:
- predicting with some accuracy the failure and the time interval in which it could occur, giving the possibility to assess an increasing of device lifetime.
- synthesizing and filtering the information to be transmitted to the central control unit, to reduce the amount of data to be processed.
Start date: 01/09/2022 End date: 31/08/2025
Partners
Università degli Studi di Firenze
Univertità degli Studi di Napoli Federico II
CNR
Univertisità degli Studi di Parma
Università degli Studi di Roma “La Sapienza”
Accenture
Atos
People
Stefano Russo
