21793 research outputs found
Sort by
Integration of a Solar Energy System at the Future Building Laboratory of Concordia University: Simulation, Analysis, and Testing of Different Operational Modes.
Full text linkThis research explores the development and integration of a solar photovoltaic (PV) system and a vehicle-to-home (V2H) backup power system at the Future Building Laboratory (FBL) of Concordia University, Montreal, Quebec. The study addresses the challenges of rural electrification, proposing renewable energy systems as sustainable and eco-friendly solutions for off-grid applications. A modified electrical power system is designed for the FBL, incorporating critical loads, a subpanel, manual transfer switches, grid-forming and grid-following converters, battery storage, and an electric vehicle with a vehicle-to-home inverter.
The system's performance is analyzed through simulation and experimental testing. Sunny Island (grid-forming/battery charger) and Sunny Boy (grid-following) were simulated in both standalone and grid-connected scenarios, with control strategies ensuring seamless operation in grid-forming and grid-following modes. Experimental results validate the simulations, demonstrating consistent performance of the converters under different operating conditions, including parallel operation.
The research also explores vehicle-to-load (V2L) and vehicle-to-home (V2H) systems, highlighting the capability of EVs to serve as reliable emergency backup power sources. The experimental results demonstrate that integrating V2H technology with renewable energy systems enhances the overall resiliency and flexibility of decentralized power systems. Moreover, this EV to-home integration reduces reliance on traditional backup energy sources, thereby enhancing sustainability.
Thus, this study provides a comprehensive framework for integrating solar PV power and EV technologies into decentralized power systems, demonstrating their potential to create scalable, reliable, and environmentally sustainable solutions for rural and remote electrification needs
Stress and Crack Analysis in PBF-LB IN738 with Heat Treatment
No full textCracking is a significant challenge in components fabricated by powder bed fusion using a laser beam (PBF-LB), occurring due to thermal stresses during or after the printing process. This study investigates crack behavior in two geometries–dog bone and cruciform–fabricated with Inconel 738 (IN738) alloy. Finite element analysis for coupled thermomechanical simulations was used to identify stress concentration locations and predict crack formation based on stress intensity factors. The results showed that stress intensity factors at critical locations exceeded the fracture toughness of IN738, indicating a high likelihood of crack formation. Furthermore, the study examined how post-process sequences, including heat treatment (HT) and base plate removal, influence stress distribution and geometric accuracy. Despite geometric differences, both geometries exhibited similar stress patterns and crack behavior, providing valuable insights into designing geometries to mitigate cracking and highlighting the importance of process sequencing in optimizing additive manufacturing
Leveraging AI for Sustainable Energy Development in Solar Power Plants Operating Under Shading Conditions
Full text linkIn a photovoltaic (PV) system, shading caused by weather and environmental factors can significantly impact electricity production. For over a decade, artificial intelligence (AI) techniques have been applied to enhance energy production efficiency in the solar energy sector. This paper demonstrates how AI-based control systems can improve energy output in a solar power plant under shading conditions. The findings highlight that AI contributes to the sustainable development of the solar power sector. Specifically, maximum power point tracking (MPPT) control systems, utilizing metaheuristic and computer-based algorithms, enable PV arrays to mitigate the impacts of shading effectively. The effect of shading on a PV module is also simulated using MATLAB R2018b. Using actual PV data from a solar power plant, power outputs are compared in two scenarios: (I) PV systems without a control system and (II) PV arrays equipped with MPPT boards. The System Advisor Model (SAM) is employed to calculate the monthly energy output of the case study. The results confirm that PV systems using MPPT technology generate significantly more monthly energy compared to those without MPPTs
A middle to late Holocene paleo-environmental study of L’Anse aux Meadows National Historic Site, Newfoundland, Canada
Full text linkPaleoecological reconstructions offer insight into environmental and climatic conditions of the past, allowing us to understand how changing climate conditions have shaped Canada’s landscapes over millennia. While instrumental data only reaches back a few centuries, paleo-reconstructions allow us to understand past environmental variability over much longer periods of time as well as predict future changes. They also contextualize archaeological sites within the broader context of ecological conditions of the period. Using high-resolution, multi-proxy analysis, this study reconstructs the past 6,000 years of vegetation and fire history around the UNESCO World Heritage archaeological site of L’Anse aux Meadows and provides new understanding into long-term ecological changes and their relationship with regional climate variations and human activity. While many previous analyses have studied samples from directly within the archaeological remains, my research offers the first regional combined charcoal and pollen record, with a well-dated chronology based on a larger suite of AMS radiocarbon dates than previously used, for the area directly downwind of the site. First, I situate this site within the human history of the North Atlantic as well as review previous paleo-ecological work done within the region, including the Northern Peninsula of Newfoundland and that of southern Labrador. Then, I present a reconstruction of pollen, macrocharcoal, and loss-on-ignition analysis based on a 2 m peat core located 300 m east of the archaeological site. The core sequence was dated using 14C dating methods with the bottommost core-section dating to ~4055 BCE (6000 cal BP). The analysis shows transitions from fen- to bog-like environments, punctuated by fire events and shifts in vegetation composition. Early fen conditions (4055–1740 BCE) transitioned to a more bog-like environment, following a significant fire disturbance. The early fen conditions were followed by a prolonged period of low peat accumulation (1095 BCE–50 CE), potentially due drier conditions. Fire frequency increased during the first millennium CE, peaking during the Medieval Climate Anomaly (920–1280 CE), suggesting warmer conditions preceding the onset of the Little Ice Age. The long-term decline in pollen influx aligns with regional cooling trends documented in the other paleo-ecological studies in the North Atlantic, driven by decreasing solar radiation and sea-surface temperature changes. This research contributes to our understanding of Holocene environmental dynamics in northern Newfoundland, situating L’Anse aux Meadows within a broader climatic and ecological context, and explores potential anthropogenic impacts on fire regimes and landscape changes
How do refugee youth experience discrimination both online and offline? What are strategies for encouraging teacher reflection?
Full text linkAbstract for PhD
How do refugee youth experience discrimination both online and offline? What are strategies for encouraging teacher reflection?
Natasha Doyon Ph.D.
Concordia University, 2025
This ABAR (arts-based action-research) dissertation used the restorative aspects of collaboratively driven pedagogies to develop immigrant youth’s critical digital literacy skills around discrimination through creativity, and to develop reflexive teaching practices with the tutors who teach them French. The central research question, “what can we learn from listening to and collaborating with marginalized youth to combat online/offline discrimination?” was pursued through workshops, artistic response, interviews, and surveys with two pools of participants: immigrant youth and teachers. Through a partnership with community-based organization Say Ça, where youth participants shared their experiences of external pressures created by racist and sexist stereotypes and created a counter-narrative that situated themselves as protagonists with agency. Teacher participants increased their awareness of implicit bias, being in positions of power, and teacher identity and responded creatively as to how these factors influenced their teacher identity development and teaching practice. This research demonstrates the positive effect that creative critical pedagogies have in initiating and sustaining difficult, but necessary conversations with refugee youth who are often overlooked and spoken for, and underlines the importance of reflexive practices in community-based organizations
Model Merging and Feature Visualization in Deep Neural Networks
Full text linkLinear mode connectivity (LMC) has recently become a topic of great interest. It has been empirically demonstrated that popular deep learning models trained from different initializations exhibit linear model connectivity up to permutation. Based on this, several approaches for finding a permutation of the model’s features or weights have been proposed leading to several popular methods for model merging. These methods enable the simple averaging of two models to create a new high-performance model. However, besides accuracy, the properties of these models and their relationships to the representations of the models they derive from are poorly understood. In this work, we study the inner working mechanisms behind LMC in model merging through the lens of classic feature visualization methods. Focusing on convolutional neural networks (CNNs) we make several observations that shed light on the underlying mechanisms of model merging by permute and average
Strengthening Canadian Academic Entrepreneurial Ecosystems: A Living Lab Approach
Full text linkAbstract: Canada’s universities are facing unprecedented challenges due to new government regulations, the shift towards a more digital environment and the advent of new digital technologies such as generative AI in education. In order to respond to these challenges, most universities have now moved toward contributing to local and regional development to co-create local value. This can be done through innovation and entrepreneurial activities by universities. To promote entrepreneurship and innovation in academic environments, living labs (LL) have been identified as a possible solution as they bring in actors from all around an innovation ecosystem together to co-create value. In order to better understand the impact and the role LL can play in encouraging academic entrepreneurship the most up to date information was collected on the subject through a literature review that served as the theoretical foundation to conduct an exploratory qualitative case study in Canadian academic ecosystems to see how LL can be an answer to the challenges facing Canadian academic entrepreneurial ecosystems and how LL’s were integrated into these ecosystems based on the five aspects of LLs which are real-life setting, co-creation, active user involvement, multi-stakeholder participation, and a multi-method approach. The findings of this study highlight the adequacy of LL as a way to fulfill the new mission of HEI’s, the mixed impact the pandemic had on LLs, the facilitating role of digital technologies in LL and some best practices and challenges giving insights on how to best set up LL methodologies in academic entrepreneurial ecosystems with qualified leadership
Continual Learning in Constrained Scenarios: Bridging Real-World Needs and Practical Constraints
No full textContinual Learning (CL) aims to enable models to learn from a sequence of tasks without forgetting previously acquired knowledge, an ability that is much needed in real-world scenarios where data and system requirements evolve over time. Traditional machine learning models are typically trained once on a fixed dataset, but CL offers a more efficient paradigm, updating models incrementally as new tasks arise, avoiding the need to retrain from scratch. However, CL faces significant challenges, such as catastrophic forgetting, where models lose performance on earlier tasks when adapting to new ones. This thesis proposes to advance CL by addressing these core challenges and developing methods for highly constrained scenarios, where access to past data is limited or unavailable.
The first research focus is on understanding catastrophic forgetting, with a particular emphasis on how neural representations evolve as new tasks are introduced. This work investigates the reliability of current metrics, such as Centered Kernel Alignment (CKA), in tracking representation changes and proposes novel methods for more accurate measurement of forgetting.
The second area of study involves developing CL methods for restricted scenarios, particularly in cases where replaying past data is not feasible due to privacy concerns or proprietary restrictions. This research introduces Model Breadcrumbs, a method that merges pre-existing fine-tuned models into a multi-task model without requiring access to their original training data.
Lastly, this thesis introduces prompt migration as a new challenge in large language model (LLM) based products. Prompt migration focuses on adapting prompts that work well for one LLM to a different LLM, without re-optimization or access to internal model parameters. Drawing parallels with CL, prompt migration is crucial for maintaining performance across LLMs as businesses increasingly switch between providers, LLM version, and architectures. The research explores how CL principles, such as incremental adaptation without retraining, can be applied to solve the problem of efficient prompt migration.
By addressing these interconnected challenges, this thesis aims to contribute novel methodologies that extend the applicability of CL to real-world constrained scenarios, improving both computational efficiency and adaptability in dynamic environments
Achilles Tendon Stiffness of the Dominant and Non-Dominant Jumping Leg of University Basketball Athletes: Relation with Performance, Range of Motion, and Injury
No full textAchilles tendon (AT) stiffness plays a key role in athletic performance and injury risk in basketball athletes. While increased stiffness may enhance performance, it may also raise injury susceptibility. Shear-wave elastography (SWE) is commonly used to assess AT stiffness, but its relationship with performance, range of motion (ROM), and injury risk remains unclear. This study examined these associations in university basketball players.
A total of 32 university athletes (12 females, 20 males) were assessed. SWE measured AT stiffness in the dominant jumping leg (DJL) and non-dominant jumping leg (NDJL). Functional assessments included single-leg vertical jump, heel raise test, and ankle dorsiflexion ROM. Paired and independent t-tests compared AT stiffness across limbs, between injured and non-injured athletes, and between sexes. Pearson correlations evaluated associations between AT stiffness, performance, and ROM.
Mean AT stiffness was 455.4 ± 72.4 kPa in males and 411.7 ± 48.5 kPa in females (p = 0.185). No significant differences were found between the DJL and NDJL. Injured male athletes had significantly lower AT stiffness than non-injured males (p = 0.040), while injured females showed more symmetrical stiffness across limbs (p = 0.027). No significant correlations were observed between AT stiffness and performance or ROM.
Although AT stiffness did not differ by limb or sex, injury history influenced tendon characteristics. These findings suggest a potential link between injury history and tendon properties, though no direct relationship with performance or ROM was found. Longitudinal studies are warranted to further explore SWE’s value in monitoring injury risk and performance
String: A novel programming language with applications to genetic programming and protocell model simulation
No full textString is a new computer language designed specifically for the implementation of ‘ribozymes’, the active entities within a new (highly simplified) model of proto-cellular life. The purpose of the model is the study of the abstract nature of simple cellular life and its relationship to computation. This model contains passive and active entities; passive entities are data and active ones are executable data (or programs). All programs in our model are written or evolved in String. In this thesis, we describe String and provide examples of both hand-written and evolved String programs belonging to different functional categories needed for cellular operation (e.g., transporter, transformer, replicator and translator). Results from the evolutionary runs are presented and discussed, where almost all ribozymes reached their optimum fitness. The latest measures of robustness and evolvability are presented, and are applied to String programs. Next a 2-state 3-symbol universal Turing machine is written in String, to prove its Turing-completeness. Finally, we present a complete ProtoCell model with an artificial metabolic cycle at its core. This cycle is simulated using a Gillespie algorithm, where String-encoded ribozymes act the ‘enzymes’ catalysing the different chemical reactions. The genomic subsystem is integrated with ProtoCell with a changing concentration of each ribozyme using self regulating differential equations, while membrane subsystem consists of a permeable membrane diffusing atoms and energy while transporting molecules between the cytoplasm and the environment. After independent stable simulation of the individual subsystems, we combined all three subsystems to have a continuously running stable ProtoCell simulation