Innovative Projects Realized

Experimental validation of data enabled predictive control

Developments in machine learning have increased interest in data-driven methods, particularly in predictive control. Data-enabled predictive control has been claimed to provide a data-driven end-to-end solution. It includes robustness guarantees obtained through regularization, borrowed from popular machine learning methods. These novel methods use data in a new framework, where analysis methods used in adaptive control and prior data-driven methods cannot easily be applied. Full understanding of the apparent robustness is lacking. This project aims to analyze the properties of data-driven predictive control and relate the method to prior results.

The student will implement data enabled predictive control for a quadcopter. Based on the theoretical results from this research project, the student will perform experiments to show when the method does and does not work. This experimental work is an integral part of this project and results from this work are expected to be included in a publication.

Modèles d’apprentissage profond pour l’auto-correction des prévisions météorologiques : application à la prévision de la demande sur le réseau électrique québécois et analyse de l’impact des tendances climatiques

A Hydro-Quebec TransEnergie (HQTE), la prevision de la charge (ou « prevision du MW») en operationnel repose sur de; model es parametriques constitues defonctions modelisantes non lineaires. Ces model es sont alimentes par des observations et des previsions meteorologiques, ces dernieres pouvant etre auto corrigees selon la ten dance observee dans Jes derniere;
heures. S’ils ont fait leur preuve et demeurent performants, on constate une lente degradation de leur capacite a prevoir en toute situation Jes besoins quebecois. Cette degradation est expliquee par plusieurs phenomenes: des changements dans le; comportements sociaux, dans la repartition geographique de la population, dans l’equipement en domotique, dans la composition de l’industrie electro-intensive. Les changements climatiques ont aussi leur part de responsabilite. Dans ce proja: de recherche, nous souhaitons etudier !’integration d’une couche neuronale d’auto-correction des intrants meteorologique; bruts, proposer une interpretation physique de ces modeles, puis analyser leur comportement dans un contexte de changement climatique.

Physical Demands Evaluations with Lifebooster Technology

Cargill is looking to improve the data collection and injury risk assessment process by leveraging the Lifebooster technology. The Lifebooster system has been leased and will be used for 12 months as a proof of concept project. The goal of the POC is to evaluate the ability to collect PDD data with wearable technology, and any improvements in accuracy and efficiency achieved through the use of technology versus the current manual/observation-based system.

Contrôle qualité des enrobés en usine par méthodes non destructives

Dans ce projet, le potentiel d’utilisation de l’essai de résonance par impacts pour faire du contrôle qualité en usine sur les enrobés bitumineux sera évalué. Cet essai est un essai non destructif, économique et simple à réaliser et il pourrait permettre de vérifier à la fois les propriétés volumétriques et les performances des mélanges formulés en usine. Dans un premier temps, l’essai de résonance par impacts sera mis en place au sein du laboratoire de contrôle qualité du partenaire industriel et la fiabilité de l’essai sera vérifiée. Par la suite, des essais seront réalisés sur différents mélanges afin de définir clairement la sensibilité des résultats par rapport à différents paramètres (type et teneur en bitume, teneur en vides, etc.). L’approche non destructive sera également utilisée pour optimiser des paramètres d’usinage de certains produits et déterminer des propriétés thermomécaniques des enrobés. Ce projet est une étape clé vers l’amélioration du contrôle qualité en usine des matériaux de chaussées.

Life Cycle Analysis of Electricity Generation and Battery Technologies for use in Ontario

As Ontario’s electricity grid continues to evolve, all technologies must be evaluated on a life cycle basis. This will ensure that potential environmental impacts and economic costs are understood in advance of installation, and will permit fair and comprehensive comparisons amongst alternatives.
This project seeks to investigate the life cycle environmental and cost performance of utility-scale photovoltaic (PV), wind, small modular reactor (SMR), and battery technologies for use in Ontario. Life cycle assessment (LCA) and life cycle costing (LCC) will be employed to examine one project using each technology, as well as combinations of renewables and battery technologies. Analyses will be completed for typical supply chains, as well as hypothetical Ontario-/Canada-centric supply chains.

Consolidating High-Frequency and Textual Data for Optimal Anomaly Detection in Derivative Markets

In the last few years, a high increase in the interest of traders and investors towards financial instruments directly led to an important augmentation of the information received daily by exchanges. Exchange regulators, who constantly monitor markets to unveil potential infractions, traditionally perform their investigation manually and the notable growth in market activity represents an important risk of fraudulent events going unnoticed. In response to that new reality, exchanges around the globe are establishing automated surveillance systems that track market activity. In this project, we set out to design new artificial intelligence algorithms that will detect anything in the Montreal exchange’s market that seems abnormal or fraudulent in the market data and news, so that analysts can focus on these alerts. Such a system could potentially detect fraudulent cases that are currently going unnoticed, while drastically reducing human costs and validation time.

Optogenetic Suppression of Early After-depolarisations in a Computational Model of the Human Heart

The heart beats an incredible 2-3 billion times in one’s lifetime. This amazing feat is made possible by highly organised electrical signals passing through the heart tissue. Disturbances of the heart’s electrical activity (known as ‘arrhythmias’) can be deadly, as they impair its ability to pump blood to the body. While various treatments for the prevention or termination of arrhythmias exist, they generally lack in effectiveness and are often associated with harmful side effects, so the development of new therapies is needed. One new technology of interest for treating arrhythmias is proteins that are activated by light, which can be used to control the heart’s electrical activity. The goal of the propose project is to explore the potential for using these light-activated proteins for the prevention of deadly cardiac arrhythmias through computational simulations in a mathematical model of the human heart, to complement my ongoing experimental studies in isolated zebrafish hearts and advance its clinical relevancy.

Development of Mobile and Fixed Models of Hybrid Energy Charging Station for Marine Applications and Waterfront Infrastructures

This project aims to develop two microgrid models based on hybrid energy resources for supplying clean energy to marine applications and waterfront infrastructure. The proposed Hybrid Energy Station (HES) has two models of HES microgrid. The first is a mobile HES (MHES) for supplying offshore power for charging boats and ships and auxiliary services integrated with fast charging technology and smart routing technique. The second is a fixed HES (FHES) for supplying the shore power for facilities for loading and unloading, commercial port facilities, supplying ancillary services, and charging ships and boats at the harbor. The two stations are integrated in the way the mobile station can work as a part of the fixed units or goes back and forth between the waterfront and the onshore area for doing charging and discharging missions. The two models combine many renewable energy sources to reduce greenhouse gas emission.

AI Algorithm Development and Refinement to Support the SR.ai Investment Research Platform

The field of responsible investing is rapidly expanding, with even greater attention on the importance of
responsible investment with each passing year, as seen most recently in the aftermath of the impactful 2021
COP26 summit, where responsible investment was key focal point. Directing our financial resources in a
sustainable direction has the potential to have a massive impact on helping us meet the Sustainable Development
Goals set forth by the UN. Though the importance of better alignment between finance and sustainability is clear,
a notion that now has very strong consensus built around it, investors still lack the right tools to support their
research process in terms of sustainability. They rely on high level data that is largely known not to be reliable,
which leads to massive misallocations of capital. The mission of SR.ai is to bring more rigour to the world of
responsible investment, using our technology backed by peer-reviewed research conducted within the University
of Toronto.

Newmont Nature-Based Decarbonization (scoping study)

Like many of the world’s largest mining companies, Newmont is voluntarily targeting net-zero carbon emissions by 2050. Recent research on technical feasibility of mining operation decarbonization has shown that mining companies may be able to achieve up to 70% carbon emissions reductions through internal initiatives, but may require options to capture and store carbon (either through purchasing offsets on the open market or capturing carbon) to drive net emissions all the way to zero. It is also predicted that the future cost of these offsets on the market will be multiples of what they are today. Nature-based solutions such as capturing carbon in forests, wetlands and soil will become an essential part of carbon capture portfolios for mining companies to stabilize and minimize the total cost of achieving decarbonization commitments. This study will model the different nature-based decarbonization opportunities available to Newmont mining operations in Canada.

Entraînement d’un agent conversationnel en assurance

Koïos Intelligence est à la source de l’agent conversationnel Olivo qui vise à offrir une expérience interactive, guidant l’utilisateur au travers des processus de prévente, de vente et d’après-vente pour tous les types d’assurances. Bien que l’outil soit dans un état avancé tant au niveau de la conversation écrite qu’orale, et ce aussi bien en français qu’en anglais, son amélioration se heurte aux exigences computationnelles lourdes pour l’entraînement des modèles d’apprentissage automatique sous-jacent. De plus, les outils de discussions inte-ractifs sont souvent imprécis dans leur développement et mise en œuvre dans le cadre d’un champ d’application particulier. C’est en particulier le cas en finance, pour les outils de conseils en matière bancaire et assurance. Il est nécessaire pour les rendre plus pertinents d’améliorer leur entraînement en utilisant des bases lexicales spécialisées, dont dispose en interne Koïos Intelligence.

Development of seedless triploid hemp varieties.

Hemp (Cannabis sativa L.) is an important natural source of cannabinoids for consumer products. Mainly grown outdoors, unintended cross-pollination can result in a significant loss of cannabinoid yield via seeding. To overcome this risk, the team intends to develop seedless, triploid varieties of hemp with high minor cannabinoid content, notably cannabigerol (CBG) O cannabichromene (CBC). The team will double the number of chromosomes in starting lines to make tetraploid (4n) hemp or hemp with 4 copies of each chromosome, Tetraploid flowers (4n) will be fertilized with pollen from normal diploid (2n) hemp of the same cultivar to produce 3n or triploid hemp seed. As seen for seedless bananas and watermelon, triploid hemp plants should be infertile when pollinated due to an uneven number of chromosomes. This process known as genome polyploidization has been successful in closely related crops to Cannabis such as hops, with benefits to product quality.