Urban railway infrastructure predictive maintenance system based on monitoring of vibro-acoustic track properties
Sustav prediktivnog održavanja kolosijeka na gradskim željeznicama zasnovan na praćenju vibro-akustičkih svojstava
Overview
The URITMIS (Urban Railway Infrastructure Maintenance System) project involves the development of an innovative system for predictive maintenance of tram infrastructure based on the recording of vibroacoustic track properties and machine learning methods.
The project also forms a research team of young scientists gathered around the described topic. Through the project, the competencies and knowledge of the research team are strengthened through a series of activities with the aim of overcoming the challenges in the development of the innovative URITMIS system.
Through three years of research, a highly specialized multidisciplinary team of young experts trained in the areas of acquisition and analysis of track vibro-acoustic properties, modelling of track structures and predictive maintenance systems based on machine learning methods will be formed.
On the basis of the developed URITMIS system, a significant saving of valuable resources will be possible for managers of tram infrastructure maintenance.
The project is financed through National Recovery and Resilience Plan 2021-2026 (NPOO) of Republic of Croatia (NPOO.C3.2.R2-I1.06).
Goals
Develop the URITMIS system
- Urban Railway Infrastructure Maintenance System
- Innovative system for predictive maintenance of tram infrastructure
Develop a strong research unit at Faculty of Civil Engineering
- Knowledge of vibroacoustic analysis of track
- Using monitoring of tramway track properties
- Predictive maintenance modelling
- Numerical modelling and model updating
- Machine learning
Objectives
Detect locations where increased vibration levels occur (in a separate strip of the tram line), which occur as a result of irregularities or damage to the track
Examine the connection between irregularities and/or damage on the track with the characteristics of the recorded vibration signal (vibration amplitude, spectrogram of frequency bands) and the application of the connection for the purpose of detecting irregularities and/or damage on the track
Detect a repeating pattern of the frequency response of individual irregularities and/or damage in the frequency spectrogram (combined irregularity)
Using the wheel roughness test and the combined roughness data, calculate the roughness of the rail running surface
Determine certain frequency responses for various irregularities and damages and their associated degree of degradation
By direct (manual) measurement of surface roughness and geometry at defined locations, check the results obtained by processing vibration acceleration signals
Create numerical models of the tram track with defined vibro-acoustic parameters (transfer functions of wheels, track, rail contact surface)
Use machine learning methods to create a predictive track maintenance model
Activities
A1. Project management
Ensuring effective team collaboration and the successful execution of planned activities.
A2. Establishment of the research team
Focuses on creating optimal conditions for a high-performing interdisciplinary research group.
Includes meetings and participation in specialized workshops to enhance and acquire new skills.
A3. State of the art
A comprehensive review of relevant literature to expand the research team’s knowledge base.
A4. Vibration data acquisition and analysis of the track condition using conventional measuring methods
Identifiying key defects along a test tram track section using standard manual measuring equipment.
Preparing a catalog of documented defects to improve and validate the machine learning algorithm in A5.
A5. Data analysis with the purpose of correlating recorded vibrations with track irregularities
Analyzing collected vibration data using machine learning to classify and detect track irregularities.
A6. Forming and developing of numerical models for condition monitoring and predicting the degradation of the tram track
Developing and enhancing numerical models for real-time condition monitoring and damage detection in tram track components.
Continuously updating numerical models using AI and machine learning with data from A4 and A5.
A7. Development of a predictive maintenance model
Creating predictive maintenance and classification plans based on processed data.
Data presentation in a GIS system for easy identification of problematic areas along the tram network.
A8. Development and testing of a GIS system for predictive maintenance of the tram track
Development of an architectural GIS system for displaying and managing predictive maintenance data.
A9. Market analysis
Investigation of the market potential of the URITMIS system for commercialization.
Identification of the interested stakeholders such as tram infrastructure managers, railway maintenance solution providers, and infrastructure data analysis companies.
A10. Dissemination of research results
Publication of the research findings in scientific journals and conferences to enhance international visibility and recognition of the group and URITMIS system.
A11. Registration for other funding sources
Ensuring project and research group sustainability by submitting three competitive project proposals for future funding.