Canada Spends Logo

Explorez notre analyse du budget 2026

Grant: Discovery Grants Program - Individual

The University of British Columbia

$50,000.00

Self-Driving Industrial Processes

EF 2024-25

Résumé
It is widely believed that we are at the beginning of a data revolution driven by the internet of things, large data sets, fast computational hardware and major algorithmic advances in machine learning.  This revolution will likely bring a new level of automation in the process industry. Our research group's long-term vision is to create autonomous plants requiring minimal human supervision. In the short-term, we are proposing to work towards this vision by solving the following three subproblems: (1) Data Pre-processing, Visualization and Modelling: Data pre-processing is known to take significant time and effort in the workflow of building a machine learning model. We propose to develop fully automated data-pre-processing algorithms, visualization tools and large-scale modelling algorithms using modern machine learning tools such as deep learning and reinforcement learning. (2) Visual Analytics and Causal Graphs for Fault Diagnosis: We have seen several breakthroughs in image processing using convolutional neural networks. We are proposing to bring the power of these tools to solve fault diagnosis problems in the process industry. We will convert time-series data into images and then apply the powerful image analysis tools. This approach helps us detect faults; however, to diagnose faults, we propose algorithms that extract causal relationships from data. By combining visual analytics with causal graphs, we will be able to achieve superior performance. (3) Meta-Reinforcement Learning for Maintenance-Free Controllers: In the last four years, we pioneered the development of several algorithms for maintenance-free controllers using reinforcement learning. However, we realized these algorithms are not practical as they require a significant amount of live operational data. We propose a new paradigm based on "meta-reinforcement learning" to adapt controllers automatically. The basic idea is to train the agent offline using a distribution of models and then use the same agent online to update the controller parameters. This approach requires much smaller data sets and will have better uptake by the industry.  To the best of our knowledge, this is the first-ever attempt to build a self-driving process with the features described above. Realizing this vision will benefit Canadian industries such as Oil & Gas, Pulp & Paper, mining and pharmaceuticals by making them efficient and giving them a global competitive advantage. In addition, this project will train several highly qualified personnel in an interdisciplinary area that is highly marketable.
Programme
Discovery Grants Program - Individual
Emplacement
British Columbia, Canada
Type de bénéficiaire
Academia
Type
Grant
Source
NSERC's Awards Data

Tel que publié

Application id
RGPIN-2023-05306
Name nom
Gopaluni, Bhushan
Department département
Department of Chemical & Biological Engineering
Organization id
2
Institution établissement
The University of British Columbia
Province en
British Columbia
Province fr
Colombie-Britannique
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2024
Competition year année de concours
2023
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
1511
Committee name en
Materials and Chemical Engineering
Committee name fr
Génie chimique et des matériaux
Area of application code
500
Area of application group en
Manufacturing processes and products
Area of application group fr
Procédés et produits industriels
Area of application en
Manufacturing processes and products
Area of application fr
Procédés et produits industriels
Research subject code
1800
Research subject group en
Chemical engineering
Research subject group fr
Génie chimique
Research subject en
Chemical engineering
Research subject fr
Génie chimique
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
Self-Driving Industrial Processes
Keywords
process modelling; fault detection and diagnosis; process monitoring; process data analytics; process control
Application summary
It is widely believed that we are at the beginning of a data revolution driven by the internet of things, large data sets, fast computational hardware and major algorithmic advances in machine learning.  This revolution will likely bring a new level of automation in the process industry. Our research group's long-term vision is to create autonomous plants requiring minimal human supervision. In the short-term, we are proposing to work towards this vision by solving the following three subproblems: (1) Data Pre-processing, Visualization and Modelling: Data pre-processing is known to take significant time and effort in the workflow of building a machine learning model. We propose to develop fully automated data-pre-processing algorithms, visualization tools and large-scale modelling algorithms using modern machine learning tools such as deep learning and reinforcement learning. (2) Visual Analytics and Causal Graphs for Fault Diagnosis: We have seen several breakthroughs in image processing using convolutional neural networks. We are proposing to bring the power of these tools to solve fault diagnosis problems in the process industry. We will convert time-series data into images and then apply the powerful image analysis tools. This approach helps us detect faults; however, to diagnose faults, we propose algorithms that extract causal relationships from data. By combining visual analytics with causal graphs, we will be able to achieve superior performance. (3) Meta-Reinforcement Learning for Maintenance-Free Controllers: In the last four years, we pioneered the development of several algorithms for maintenance-free controllers using reinforcement learning. However, we realized these algorithms are not practical as they require a significant amount of live operational data. We propose a new paradigm based on "meta-reinforcement learning" to adapt controllers automatically. The basic idea is to train the agent offline using a distribution of models and then use the same agent online to update the controller parameters. This approach requires much smaller data sets and will have better uptake by the industry.  To the best of our knowledge, this is the first-ever attempt to build a self-driving process with the features described above. Realizing this vision will benefit Canadian industries such as Oil & Gas, Pulp & Paper, mining and pharmaceuticals by making them efficient and giving them a global competitive advantage. In addition, this project will train several highly qualified personnel in an interdisciplinary area that is highly marketable.