Carleton University Institutional Repository
Not a member yet
20816 research outputs found
Sort by
Forward and Back-Propagation with an Analog Neural Network
No full textThe objective of this thesis is to attempt an artificial neural network (ANN) built from simple analog components that can potentially be manufactured on the moon. The ANN should discover a mapping between inputs and outputs via a feedback loop and train weights that produce a desired output. The methodology for the circuits consists of forward propagation, calculating new weights, and backpropagating the weights until trained. It is observed that with the trained weights for a pair of inputs, a new pair of similar inputs can predict desired outputs. As a specific use case, the circuit’s output is used to showcase simulated obstacle avoidance. It is concluded an ANN architecture constructed with simple electronic components can be utilized for robotic control but further work will be required toward building circuits for batch training data that can be manufactured on the moon
Learning Based Resilient Control and Vulnerability Management for Wide Area Damping Control with Cyber and Physical Structures
No full textLow-frequency oscillations (LFOs) limit power transmission between interconnected power systems, focusing especially on inter-area oscillations. While power system stabilizers (PSSs) effectively manage inner-area oscillations, inter-area oscillations remain a significant challenge. This is compounded by cyber uncertainties including uncertain time delays and random DoS attacks in wide-area signal transmissions through remote phasor measurement units (PMUs) operating over wireless channels, affecting both the reliability and security of power system operations. Innovative Reinforcement Learning (RL) algorithms and deep reinforcement learning (DRL) are proposed to secure communication channels and stabilize cyber-physical systems under these conditions, without requiring prior knowledge of the specific cyber uncertainties. A key innovation is the development of a model-free RL based Wide-Area damping controller (WADC), which evolves into a more advanced DRL based version. This version employs the Deep Deterministic Policy Gradient (DDPG) method to refine controller policies. The training process integrates cyber and physical layer data, including random time delays, to enhance the effectiveness of the stabilization process. Additionally, a coordinated WADC scheme is designed to address multi-mode LFOs effectively. For modes with minor disturbances, it proposes a DRL-based WADC equipped with an LSTM delay compensator, allowing the DRL agent to adapt to dynamic power system conditions by designing optimal stabilization voltage action sets. For significant disturbances, it recommends a robust DRL-based WADC that features continuous online training using a minimax DRL algorithm aimed at optimizing performance under worst-case scenarios. This approach not only improves robustness but also utilizes a combination of cyber and physical system metrics within the agent’s state set to enrich training data, ensuring more effective stabilization and reducing training costs. To defend against uncertain cyber attacks on the communication channels from PMUs to WADCs, the thesis proposes an incomplete information stochastic game (IISG) based optimal cyber-layer intrusion detection system (IDS). This IDS employs a Bayesian-based method to update beliefs about the nature of the input signals, distinguishing between genuine PMU signals and potential threats. This strategy is designed to minimize unnecessary defensive actions, to optimize system response to actual threats and to enhance overall security measures within the power systems
A Tunable Diode Laser Absorption Spectrometer for the Detection of Volatile Compounds in Lunar Regolith
No full textA bench-top proof-of-concept tunable diode laser absorption spectrometer (TDLAS) has been developed for the sake of travel volatile compound detection. The objective was to quantify volatile compounds in the mid-infrared (IR) region. The mid-IR region was selected for its habitation of many volatile compound's strong absorption banks. Work done was limited to the detection of H2O(v) at standard conditions in a flushed test chamber. Budgetary and timeline constraints did not allow for the procurement of quality reference cells. The TDLAS targets H2O(v)'s absorption line located at 2701.3844 nm, which has an associated line strength of 4.5408x10^-20. The design employs a second harmonic-based spectral analysis detection method. The instrument was developed in-house, with an FPGA-based MyRIO-1900 microcontroller. The circuitry includes thermal control circuitry and diode/photodiode circuitry. The associated control software is developed in NI LabVIEW. After simulation and experimentation, it was proven that the instrument can detect and quantify H2O(v)
The Truncated Tropomyosin Kinase B Receptor Influences Feeding Behaviour, Cholesterol Production and Susceptibility to Stress in a Sex-Dependent Manner
No full textBrain-derived neurotrophic factor (BDNF), is known to promote neuroplastic changes that facilitate many behaviors and may mitigate the impact of stressor exposure. BDNF signals through the full-length tropomyosin kinase B receptor (TrkB.FL) on neurons to activate pathways that upregulate synaptic proteins and promote neuroplasticity. There are several possible isoforms of TrkB, including a truncated form, TrkB.T1, which acts to repress TrkB.FL/BDNF signaling in neurons. Moreover, TrkB.T1 it is the sole isoform found on astrocytes, where its function is not yet well understood. Astrocytes are a critical component of the tripartite synapse, and limited work suggests they play an important role in BDNF sequestration and release. To better understand the role of the truncated TrkB receptor and how it might influence susceptibility or resilience to stress we assessed behavioral and neuronal signalling in male and female mice with TrkB.T1 constitutively knocked out that were exposed to chronic unpredictable stressors. Additionally, since enhanced BDNF signaling in the hypothalamus has been shown to create hypophagia, we also assessed the impact that the lack of TrkB.T1 has in terms of microstructural changes affecting feeding behaviour. We hypothesized that removing the “brakes” on full-length TrkB signaling in the prefrontal cortex and hippocampus will enhance stress-resiliency but would result in an anorexic-like reduction in feeding behavior. Finally, to elucidate the underlying mechanisms behind the role of TrkB.T1 on glia, we engaged in a parallel series of in vitro experiments to better understand this receptor function with respect to cholesterol production and exogenous BDNF sequestration/release in astrocytes. Our data clearly indicated that TrkB.T1 deficiency produced a unique phenotype that enhances behavioural vulnerability to stressor exposure in a sex-dependent manner, with males mostly affected. TrkB-pCREB signaling was clearly disrupted in TrkB.T1 knockout mice and these mice also displayed disordered feeding and weight loss that was accompanied by sex-based changes in cholesterol. Taken together, our findings prompted us to suggest that astrocytic TrkB.T1 signalling might be important for mediating both cholesterol processing and feeding behaviour which has important implications for anorectic or other stressor-related illnesses
Analysis of the 2 Flavour SU(5) Georgi–Glashow Model
No full textWe study the supersymmetric and non-supersymmetric versions of the SU(5) Georgi-Glashow model with 2 flavours of 5* and 10. In the non-supersymmetric model we identify six possible gauge invariant bound states. Like in the 1 flavour model, 't Hooft anomaly matching results in non-trivial solutions potentially indicating that no global symmetry is broken and the existence of massless fermionic composites in the infrared (IR). In the supersymmetric model, we identify six (different) composites but show that anomaly matching is impossible and thus analyze possible flavour symmetry breaking patterns. We then look for supersymmetry breaking and found a possible flat direction, unlike in the 1 flavour model. We also identify an ADS-like effective superpotential in the IR suggesting a possible runaway behavior for the VEV indicating that the theory might not have a stable vacuum
The Therapeutic Potential of AI-Generated Art in Short-Term Stress Management
No full textArt therapy has been used throughout history to aid in maintaining the mental well-being of individuals. We aim to explore the potential benefits of generative AI tools, such as stable diffusion,in helping reduce short-term stress, as well as compare them with traditional art methods. Thirty participants were randomly assigned to one of four condition groups based on the stress-inducing activity (without the stress-inducing activity in a control group and with the stress-inducing activity in an experimental group) and the art activity (AI art generator or traditional sketching task). Stress levels were measured throughout the studies using a Visual Analogue Stress Scale (VAS). The results of the experiment reveal that, while not as effective as the traditional art method, AI art generators can help reduce short-term stress, which could be attributed to the tool allowing participants to express their thoughts and feelings and experience the feeling of catharsis
On the Margins of Socio-Political Extremes: An Analysis of Conservative Albertans’ Facebook Posts and Comments
No full textThis study investigates Albertans' expressions of social, political, and economic discontent to describe to what extent these expressions reflect right-wing populist outlooks and the characteristics of recent right-wing populist movements. This question was investigated using a critical theoretical orientation and an ethnographic content analysis methodology to analyze content collected from two popular, Alberta-based, socio-politically conservative public Facebook pages: Alberta Proud and Alberta Stronger. Analysis revealed the central elements of populism with social decline being attributed to corrupt elites and associated ‘others’ and social revival being attributed to ‘the people’. Overarching themes of discontent are consistent with the characteristics of recent right-wing populist movements including distrust in institutions, elected officials and experts, and perceived economic threat. Ultimately, I theorize that expressions of right-wing populism observed here reflect the political imaginaries of a settler colonial population that remains committed to economic liberalism and liberal individualism despite outwardly rejecting ‘liberalism’
Development of Panelized Exterior Retrofit Solutions for Commercial and Institutional Buildings
No full textThe total energy used by commercial and institutional buildings in Canada is equivalent to over 10.7 million households. Exterior energy retrofits are over-cladding assemblies that are applied to the building envelope. These retrofits can drastically improve energy efficiency without the time consumption and disruptions of a conventional interior retrofit. This thesis evaluated two potential building envelope solutions for commercial and institutional buildings. One solution used primarily biogenic materials, and the other is based around an all-in-one building enclosure system. Test assemblies of both retrofits were built full-scale for long-term experimental testing. Both solutions provided significant improvements of thermal resistance compared to the original building assembly with little risk for failure at all measured locations. Similarly, modelled and experimental data showed overwhelming evidence that neither solution has potential for moisture damage and mould growth. The results indicate that either solution is viable for commercial and institutional building retrofits
Development and Qualification of a Drone-Based Anemometry Platform for Air Risk Assessment in Urban Environments
No full textWind flow patterns around building features are crucial for the safe operation of drones and urban air mobility missions. Fixed anemometer stations and computational fluid dynamics simulations face limitations in measuring airflow behind buildings. To address this, a drone platform was developed for an ultrasonic anemometer, offering high-resolution measurements and flexibility in positioning. A wind tunnel study using this low-cost drone-mounted ultrasonic anemometer was conducted. The study includes an evaluation of the anemometer's accuracy and acceptance range in terms of orientation, angles, namely, pitch, roll, and yaw. Measurements under uniform flow conditions, with reference instruments, validate the anemometer's performance in wind speed and direction measurements. The study also explores the anemometer's capabilities in measuring wake characteristics behind a cylinder in turbulent flow while mounted on a quadrotor mimicking realistic flight conditions. The results demonstrate that using compact ultrasonic anemometers for drone-based anemometry in urban environments can yield accurate measurement
A Meta-analysis of Recidivism Rates Among Individuals Who Commit Child Sexual Exploitation Material (CSEM) Offending
No full textChild sexual exploitation materials (CSEM) offences constitute the majority of online sexual crimes. Using a meta-analysis of 30 non-overlapping samples with CSEM offences (N = 25,978), we summarized recidivism rates and assessed the impact of moderators on sexual recidivism. Fixed-effect estimates showed 5.9% sexual (95% CI = [5.6, 6.3], k [studies] = 21, N = 19,112), 1.5% contact sexual (95% CI = [1.4, 1.7], k = 20, N = 18,543), and 4.1% CSEM recidivism (95% CI = [3.8, 4.4], k = 21, N = 13,522). The rates of contact sexual offences among CSEM-Exclusive individuals are notably low compared to Mixed-offending individuals who present a higher risk profile; however, both subgroups exhibit high rates of nonsexual reoffending, underscoring the heterogeneity within this offending population. Our findings highlight the need for risk assessment tools to evaluate the likelihood of both sexual and general recidivism in the CSEM population