Canada Spends Logo

Explore our Budget 2026 analysis

Grant: Discovery Grants Program - Individual

École Polytechnique de Montréal

$50,000.00

Continuously-Adaptable Machine Learning Methods for Structural Health Monitoring

FY 2024-25

Summary
This proposal is centred around the development of probabilistic machine learning methods that are tailored to the needs of structural health monitoring (SHM), with applications on real-time anomaly detection and data-driven deterioration models for visual inspections. Our recent progress has focussed on including Bayesian neural networks into our probabilistic SHM frameworks in order to better predict complex patterns and dependencies that are present in the data recorded on civil infrastructure. Despite the streak of successes we had when validating these approaches on structural condition forecasting and SHM anomaly detection, one key aspect remains to be addressed; Neural networks are known to have a hard time dealing with changing conditions that require continual learning over time without losing track of what was previously learnt. The overarching objective of my research program is to enable large-scale data-driven decision-making for long-term SHM; Scaling data interpretation up to entire infrastructure networks not only involves focussing on prediction accuracy, it also requires a complete data interpretation system with probabilistic predictive models that can differentiate between what they know and what they do not, so that a computer-based decision-making framework can rely upon this information in order for it to make better decisions than human engineers do, and being able to do so at a much larger scale. We are already working toward this goal in partnership with Hydro-Quebec and the Quebec Transportation Ministry through the integration of Bayesian neural networks within SHM anomaly detection, and network-scale infrastructure probabilistic infrastructure deterioration models. This research proposal aims at contributing to this program by developing new methods for increasing the level of automation in order to reduce the requirement of having engineers constantly supervising and maintaining the data-analysis process. More specifically the objectives consist in developing continual learning methods tailored for (1) infrastructure time-series models, (2) decision-making for anomaly detection in SHM, and (3) to validate the novel methods for large-scale deployment on infrastructure data. The methods developed through this research will form the foundations for the next generation of data analysis SHM systems that can operate continuously under changing conditions. The developments made will tackle fundamental aspects that are beyond the scope of our ongoing industrial projects, thus allowing my group's research program to expand in reach while contributing to the formation of highly qualified personnel and transferring new knowledge to the Canadian industry. The methods developed will enable sustainable and economically viable infrastructure management by leveraging monitoring technologies and autonomous data interpretation, to better allocate resources and extend the service life of infrastructure through preventive maintenance.
Program
Discovery Grants Program - Individual
Location
Quebec, Canada
Recipient Type
Academia
Type
Grant
Source
NSERC's Awards Data

As published

Application id
RGPIN-2024-04694
Name nom
Goulet, James-Alexandre
Department département
Civil, geologic and mining engineering
Organization id
64
Institution établissement
École Polytechnique de Montréal
Province en
Québec
Province fr
Québec
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2024
Competition year année de concours
2024
Award amount
50000
Program id
RGPIN
Program name en
Discovery Grants Program - Individual
Program name fr
Programme de subventions à la découverte - individuelles
Group en
DISCOVERY RESEARCH
Group fr
RECHERCHE AXÉE SUR LA DÉCOUVERTE
Committee code
1509
Committee name en
Civil, Industrial and Systems Engineering
Committee name fr
Génie civil, industriel et des systèmes
Area of application code
601
Area of application group en
Construction, urban and rural planning
Area of application group fr
Construction, urbanisme et aménagement rural
Area of application en
Structural engineering
Area of application fr
Génie des charpentes
Research subject code
1000
Research subject group en
Civil engineering
Research subject group fr
Génie civil
Research subject en
Civil engineering
Research subject fr
Génie civil
Field of research list names en
Not available
Field of research list names fr
Non disponible
Socioeconomic objective list names en
Not available
Socioeconomic objective list names fr
Non disponible
Application title
Continuously-Adaptable Machine Learning Methods for Structural Health Monitoring
Keywords
structural health monitoring (shm); anomaly detection; infrastructure deterioration models; time-series; probabilistic machine learning; probabilistic methods; bayesian neural networks
Application summary
This proposal is centred around the development of probabilistic machine learning methods that are tailored to the needs of structural health monitoring (SHM), with applications on real-time anomaly detection and data-driven deterioration models for visual inspections. Our recent progress has focussed on including Bayesian neural networks into our probabilistic SHM frameworks in order to better predict complex patterns and dependencies that are present in the data recorded on civil infrastructure. Despite the streak of successes we had when validating these approaches on structural condition forecasting and SHM anomaly detection, one key aspect remains to be addressed; Neural networks are known to have a hard time dealing with changing conditions that require continual learning over time without losing track of what was previously learnt. The overarching objective of my research program is to enable large-scale data-driven decision-making for long-term SHM; Scaling data interpretation up to entire infrastructure networks not only involves focussing on prediction accuracy, it also requires a complete data interpretation system with probabilistic predictive models that can differentiate between what they know and what they do not, so that a computer-based decision-making framework can rely upon this information in order for it to make better decisions than human engineers do, and being able to do so at a much larger scale. We are already working toward this goal in partnership with Hydro-Quebec and the Quebec Transportation Ministry through the integration of Bayesian neural networks within SHM anomaly detection, and network-scale infrastructure probabilistic infrastructure deterioration models. This research proposal aims at contributing to this program by developing new methods for increasing the level of automation in order to reduce the requirement of having engineers constantly supervising and maintaining the data-analysis process. More specifically the objectives consist in developing continual learning methods tailored for (1) infrastructure time-series models, (2) decision-making for anomaly detection in SHM, and (3) to validate the novel methods for large-scale deployment on infrastructure data. The methods developed through this research will form the foundations for the next generation of data analysis SHM systems that can operate continuously under changing conditions. The developments made will tackle fundamental aspects that are beyond the scope of our ongoing industrial projects, thus allowing my group's research program to expand in reach while contributing to the formation of highly qualified personnel and transferring new knowledge to the Canadian industry. The methods developed will enable sustainable and economically viable infrastructure management by leveraging monitoring technologies and autonomous data interpretation, to better allocate resources and extend the service life of infrastructure through preventive maintenance.
École Polytechnique de Montréal - $50,000 Grant (2024… | Canada Spends