21793 research outputs found
Sort by
Machine Learning Approaches for Aftermarket Demand Forecasting: Tackling Intermittent Time Series Challenges
Get PDFThis thesis addresses the significant challenge of achieving precise demand prediction within the aviation aftermarket maintenance and spare parts management sector, particularly concerning intermittent parts. These components, characterized by irregular demand occurrences, present a formidable challenge due to the difficulty in accurately estimating their demand and setting appropriate stock levels. Historical approaches, relying on conventional demand forecasting techniques, often yielded inaccurate forecasts, resulting in slow inventory turnover and increased warehousing costs. To address this challenge, a broad spectrum of techniques was examined, ranging from traditional statistical models to modern machine learning and deep learning methods falling under the broader domain of artificial intelligence. Deep learning has garnered substantial attention in time series analysis for its exceptional forecasting performance. Real-world data from an aviation company was used to implement various forecasting models, including traditional methods like the exponential smoothing, and Croston, as well as machine learning models like SVR, Random Forest, and K-nearest neighbour. Deep learning techniques, including LSTM, GRU, and CNN, were prominently featured, with customized error metrics tailored to intermittent demand forecasting. The findings highlight that, on average, deep learning models, especially Gated CNN and LSTM, outperform other models and offer highly accurate forecasts for intermittent demand. This study serves as a reference point for choosing the most effective
forecasting method to support inventory planning in the aviation aftermarket, reducing costs, and enhancing service reliability. Moreover, its relevance extends to various industries dealing with
intermittent demand, offering valuable insights for improved demand forecasting
The Unsung Heroes of Training and Development in Canada, The Administrators: A Content Analysis of Job Announcements
Get PDFResearch that explores competencies needed by Training Administrators is limited; yet the role of Training Administrator is common. The purpose of this study was to define the role of Training Administrator from the industry’s perspective of Training (Learning) and Development with respect to its main roles and responsibilities, soft skills, education, and technical requirements needed to perform the job successfully, and what the typical title of the job is. To determine definitions, 63 job announcements from across Canada were collected from one online job database (LinkedIn.com) over a five-month period in 2021. Following a systematic process of collection, coding, and the measurement of frequency, by which a role and responsibility category, as well as a stated superior-level soft skill, was found within each job announcement, five main role and responsibilities and eight superior-level soft skills emerged. Moreover, the required minimum education, experience, and technical skills were identified from an employer’s perspective. The results suggested that those in the role of Training Administrator were mainly expected to perform the roles and responsibilities of:
1. Learning Management System (“LMS”) Administrator.
2. Logistical Support.
3. Data Analytics.
4. Design and Development, of curricula.
5. Learning Communication Specialist.
The eight soft skills expected at a superior-level skillset were found to be in:
1. Oral and written.
2. Interpersonal.
3. Multi-tasking.
4. Detail-oriented.
5. Time management.
6. Adaptability.
7. Stakeholder management.
8. Self-motivation.
A typical job title for the role as determined by the current study was Learning Coordinator, rather than Training Administrator
Thermo-poromechanical modeling of saturated clay soils considering bound water dehydration.
Get PDFThe non-isothermal deformation of clay soils is a critical concern in energy and environmental-related geotechnics, given the complex microstructure and mineral composition of clay-related geomaterials. Investigating their thermo-mechanical behaviors poses significant challenges that previous studies have often overlooked. Specifically, distinguishing between thermal plastic strain and clay dehydration strain has received little attention.
To address these gaps, a novel constitutive model is proposed for describing the thermo-elastoplastic behaviors of water-saturated clayey soils. The model incorporates the effects of temperature variation and mechanical loading on elastoplastic strains and dehydration behavior. The thermo-mechanical behavior is quantified using thermodynamics laws and unconventional plasticity principles. Additionally, a finite element method (FEM) model is employed to simulate the thermo-hydro-mechanical (THM) responses of water-saturated clay soils. This FEM model accounts for temperature variation effects on bound water dehydration and corresponding thermo-poromechanical strains. By incorporating unconventional plasticity, the elasto-plastic behavior is more accurately described. The validation process for this FEM model involves laboratory results on various clay soils with different geological origins, demonstrating a reasonable agreement between the model's predictions and experimental data. Notably, the numerical results highlight the impact of bound water dehydration on the generation of excess pore pressure in clay soils during heating.
Expanding beyond the realm of theoretical models, a research project is underway to assess the geomechanical performance of a potential borehole thermal energy storage system (BTES) in a Canadian subarctic region. To quantify the poromechanical impact, a two-dimensional finite element model (FEM) is created to simulate a BTES system, encompassing the borehole and the surrounding soil formation. The model aims to analyze the effects of cyclic temperature variations and bound water dehydration on the short-term ground response and pore water pressure development. Our results indicate the importance of considering bound water dehydration in characterizing the ground heave process during the short-term BTES operation in an overconsolidated formation. The simulated ground expansion behavior is due to the high excess pore pressure generated during thermal storage, which is accompanied by the release of in-situ effective stresses. The neglect of bound water dehydration will underestimate the magnitude of ground heave during a short-term BTES operation
Multivariate Change of Measure as Correction Method in Ethical Pricing
Get PDFIn recent years, multiple global events have drawn society's attention to fairness-related issues and various societal movements resulted from them. For many fields, the impact was immediate and substantial, but for others it has been much more timid. Insurance is one of the latter. More specifically, the way fairness is implemented in algorithms used to calculate insurance premiums has not changed in decades due in part to the lack of modernization from regulators and in part to the complexity of the issue. Nonetheless, in preparation for society's growing expectations, researchers have developed many ways to implement algorithmic fairness. An exposition is made on this concept, including qualitative and quantitative definitions of fairness as well as approaches to its implementation found in the literature. In particular, the method developed by Lindholm et al. (2022) is discussed in detail and followed up by the introduction of our own novel approach. This approach is demonstrated on simulated data, and it is shown that it can significantly reduce unfairness according to pre-determined metrics
Visual Servoing-Based Dynamic Accuracy Enhancement of Industrial Robots by Using Photogrammetry Sensor
Get PDFIndustrial robots are defined as robot systems for manufacturing, with feature programmability, certain automation, and capability to move on several axes. However, the insufficient accuracy has limited the industrial robots to many potential applications in aerospace manufacturing. Typical accuracy requirement in aerospace manufacturing, such as drilling and fastening, are or less. Unfortunately, the discrepancy between a virtual-model robot and the corresponding real robot can even reach around due to the deflection of the mechanical structure and tolerances. Therefore, accuracy enhancement is highly significant in order to expand industrial robots to more applications in aerospace manufacturing. Visual servoing is extensively applied to control industrial robots with the help of the visual information feedback, especially for unmodeled environment. In recent decades, using visual servoing to reach the desired pose precisely has attracted the attention of many researchers. In this thesis research, three visual servoing-based control schemes are proposed to target at the accuracy enhancement of positioning and path tracking.
The research work in this thesis includes four parts. First, an adaptive Kalman filter (AKF) is developed to estimate the pose information of the objects in Cartesian space from the measurements of the visual sensor with noises. In order to address the difficulty in obtaining precise process error covariance and measurement error covariance, an adaptive algorithm is proposed to tune the covariance matrices so that the Kalman filter can produce synchronous pose estimations even when the objects are moving at certain acceleration or high speed. In this research, a photogrammetry sensor, C-Track 780 is selected as the visual sensor. The original measurement data from C-Track 780 are contaminated with the noises. The proposed AKF algorithm is developed to process the measurement data from C-Track 780 to obtain smooth pose information for visual servoing.
Second, an effective dynamic pose correction (DPC) scheme for industrial robots is proposed to enhance the pose reaching accuracy for satisfying both position and orientation precision requirement. By applying the DPC scheme, the end-effector of an industrial robot can approach the poses in its reachable workspace with high accuracy. Some experiments are implemented on an industrial robot, FANUC M20-iA, by using C-Track 780. The experimental results demonstrate high pose accuracy ( for position and for orientation).
In the third part, a practical dynamic path tracking (DPT) scheme for industrial robots is elaborated for improving the path tracking accuracy. The proposed DPT scheme is designed to realize 3D dynamic path tracking by correcting the robot movement in real time. By using the proposed DPT scheme, the industrial robot can be controlled to follow the pre-planned path with high tracking accuracy. The dynamic stability for the robot system with the proposed DPT scheme is proved theoretically through Lyapunov function. Moreover, the effectiveness of the proposed DPT scheme is verified by the experiments on FANUC M20-iA with C-Track 780. The experimental results show the high path tracking accuracy ( for position and for orientation) is achieved.
In the last part, adaptive iterative learning control (AILC) in parallel with the proposed DPT scheme is proposed to update the time-varying control parameters along iteration axis and calculate new compensation to adjust the control inputs produced by the DPT module at each time interval based on the memorized data information and current feedback. Three experiments in different situations (without path correction, with DPT control, and with AILC control) are carried out for the comparison. The pose accuracy can be stably confined to less than for position and for orientation. Moreover, the repetitive disturbances can be also overcome within certain iterations so that the vibrations can be significantly reduced. Therefore, the AILC algorithm proposed verified to be effective to further improve the DPT scheme.
The research work in this thesis explores various schemes to enhance the positioning and path tracking accuracies for 6-DOF industrial robots. The proposed schemes, DPC, DPT and AILC, are proved to be effective on some FANUC robots which can be representative in 6-DOF articulated industrial robots for manufacturing
Linear quadratic control using reinforcement learning and quadratic neural networks
Get PDFThis thesis focuses on the application of reinforcement learning (RL) techniques to design optimal controllers and observers for linear time-invariant (LTI) systems, namely linear quadratic regulator (LQR), linear quadratic tracker (LQT), and linear quadratic estimator (LQE), utilizing measured data. The closed-form solution and wide-ranging engineering applications of the linear quadratic (LQ) problems have made it a preferred benchmark for assessing RL algorithms. The primary contribution lies in the introduction of novel policy iteration (PI) methods, wherein the value-function approximator (VFA) is designed as a two-layer quadratic neural network (QNN) trained through convex optimization. To the best of our knowledge, this is the first time that a convex optimization-trained QNN is employed as the VFA. The main advantage is that the QNN’s input-output mapping has an analytical expression as a quadratic form, which can then be used to obtain an analytical linear expression for policy improvement. This is in stark contrast to available techniques that must train a second neural network to obtain the policy improvement. Due to the quadratic input-output mapping of the QNNs and the quadratic form of the value-function in the LQ problems, the QNN is a suitable VFA candidate. The thesis designs the LQR and LQT without requiring the system model. The thesis also designs the LQE correcrtion term provided that the system model is given. The thesis establishes the convergence of the learning algorithm to the LQ solution provided one starts from a stabilizing policy. To assess the proposed approach, extensive simulations are conducted using MATLAB, demonstrating the effectiveness of the developed method. Furthermore, the proposed observer is designed for a nonlinear pendulum with a given linearized model and it is shown that the proposed observer is improved over utilizing only linearized model. This shows the adaptability for nonlinear systems
Bayesian Parameter Estimation of Probabilistic Models for Information Retrieval and Clustering in Discrete Data Spaces
Get PDFIn the contemporary era, a substantial amount of data is generated, prompting a critical need to effectively model data for thorough analysis and extraction of meaningful patterns. This is particularly crucial in various real-world applications, with natural language processing standing out as an area urgently requiring data analysis. Tasks such as document retrieval, spam email filtering, smart assistant applications, and sentiment analysis exemplify the extensive scope of natural language processing (NLP) and text mining. Addressing this context, various Bayesian models have been developed to aptly model data and extract essential information by considering latent topics. These models, grounded in probabilistic graphical models like Bayesian networks, capture the probabilistic dependencies between variables. Their ability to incorporate evidence from previous user knowledge enhances retrieval performance significantly. Furthermore, Bayesian network models exhibit effectiveness and generality surpassing classical information retrieval models like boolean, vector, and probabilistic models. This versatility positions Bayesian models as valuable approaches in information retrieval. Topic modeling, a valuable technique in text mining, plays a key role in uncovering concealed thematic structures within document collections, facilitates the identification of clusters of "topics" or co-occurring words, and aids in understanding underlying themes and patterns from data. The unsupervised classification of documents, akin to clustering in numeric data, allows for the discovery of natural document groups, even in the absence of predefined topics.
However, significant challenges persist, including the management of queries not present in data collection, sparsity within datasets, especially in the age of big data, and addressing correlations between observations. This thesis suggests innovative Bayesian extensions for data modeling, utilizing the Generalized Dirichlet distribution and the Beta-Liouville distribution as prior probability distributions to incorporate new queries into the topic space. Furthermore, these priors are integrated into a probabilistic clustering-projection model to evaluate their impact on both clustering and projection jointly. Lastly, in addressing the issues and hurdles associated with data sparsity, the Generalized Dirichlet distribution and the Beta-Liouville distribution are advocated as prior probability distributions to confront these challenges.
The selection of a suitable prior is crucial in Bayesian data modeling, and these distributions are explored for their ability to model various non-Gaussian data and overcome the limited covariance structures of other distributions like the Dirichlet distribution. Following the determination of prior probabilities, the next step involves estimating optimized parameters for the distribution and model. An iterative parameter estimation model, utilizing the Expectation Maximization algorithm, is developed to maximize data likelihood. The simplicity of the proposed iterative algorithms allows these models to successfully handle real-time data, making them applicable across a broad range of practical scenarios
Le Salon de la Refusée: une résilience de A à Z
Get PDFChaque outil typographique a déterminé un contexte de création et dans une certaine mesure, conditionné l’esthétique des lettres d’une époque. Aujourd’hui, les logiciels mettent en place une manière de faire qui condamne les yeux à être au centre du processus, laissant ainsi la main et ses multiples savoirs de côté. Les ressources théoriques quant à elles, observent la dimension gestuelle de l’écriture sans véritablement prendre en compte l’aspect tactile. Afin de combler ces insuffisances, la recherche propose des contextes de création non conventionnels, définis par l’action et la matière. Elle aborde la discipline avec un regard alternatif et observe des ressources contre-culturelles, philosophiques et sur la typographie expérimentale. Ainsi, de quelle manière un contexte de travail défini par l’action et le toucher pourrait-il interroger la création de caractère et déplacer les standards de la discipline ? Le processus revisite les conventions de la création de caractère pour créer un cadre de travail conceptuel inventif et inspirant. Les formes se mesurent en surfaces, les styles typographiques se pèsent en grammes et l’espace de travail se calcule en degrés. La structure de la lettre est revisitée, fabriquée d’imprévus et bricolée de raffinements. Ses lignes synthétisent et matérialisent le processus dans une ligne éloquente, expressive et organique. La lettre porte en elle la subjectivité de l’auteur et les matériaux qui la composent, elle devient une lecture du processus. Le contexte de création devient une signature qui inspire de nouvelles manières de faire
Cognition of common mammal mesopredators and implications for their management
Get PDFAn animal’s cognitive abilities can modulate its interaction with humans and exacerbate conflicts. Mesopredator mammals demonstrate innovation and learning through their behaviour, especially in a generalist and widespread species like the common raccoon (Procyon lotor). The aim of this thesis is to combine wildlife management with the study of cognition to provide better coexisting conditions between humans and mesopredators. I first conducted a narrative synthesis to characterize the contexts in which conflicts occur with the raccoon, the red fox (Vulpes vulpes) and the striped skunk (Mephitis mephitis), and a meta-analysis to rigorously evaluate the efficacy of the mitigation techniques in reducing the intensity of conflicts. Although lethal interventions are regularly applied with relatively high efficacy, many nonlethal options are also effective. Many methods are based on a profound understanding of animal behaviour and cognition. Shifting toward cognitive studies, I experimentally tested problem-solving and learning performances of wild raccoons in three Québec national parks. I demonstrated innovative problem-solving in raccoons, and that task difficulty level has a clear effect on success probability and time to solve the problem. Higher exploratory diversity was linked to success, but not persistence. I also found evidence of learning, by an improved performance in term of success probability over consecutive trials. Raccoons living in a zone of the park more affected by the human presence also present more pronounced learning performance, which likely relates to their strong propensity to forage on human food. There are also indications that the improved performance gained through learning is retained over the winter season. Indeed, we found the success rates of the last trial from a summer to be similar to that of the first trial of the following summer. Basing mitigation interventions on scientifically proven methods and better integration of animal behaviour, may improve mesopredators management. Expanding our knowledge of cognition in common species contributes to our appreciation and tolerance toward wildlife. Overall, my findings could facilitate reaching a balanced coexistence between humans and mesopredators
Morality and Meaning-Making: How Mothers Make Sense of Their Own Transgressions and Those of Their Adolescent Children
Get PDFThis study examined mothers’ constructions of meaning about transgressions, and whether the way mothers make sense of their own moral transgressions is related to how they make sense of those of their children. The sample consisted of 89 mothers of adolescent children (children’s age range = 12-15 years; 43 boys, 46 girls). Each mother was asked to choose a moral value that was most important to them and to write about past experiences wherein they and their child acted out of alignment with this value. Written narratives were coded reliably for references to growth, choice, remorse, negative evaluation, and negative characterological attribution. Mothers also answered a series of related closed-ended follow-up questions on Likert scales. The first research aim was to examine how different aspects of meaning-making were interrelated within the mothers’ written narrative accounts. Results indicated that, in narratives of their own transgressions, mothers’ negative evaluations were positively related to their negative characterological attributions and remorse. Regarding their narratives of their child’s transgressions, negative characterological attributions were positively linked to negative evaluations and growth. The second aim was to examine similarities and differences between mothers’ accounts of their own and their children’s transgressions. Contrary to expectations, results showed that mothers discussed growth and remorse more for themselves, and choice and negative characterological attribution more for their children. The third aim was to examine associations between the types of meanings mothers made regarding their own transgressions and those of their children. Results revealed negative correlations between mother’s choice and child’s growth and mother’s negative evaluation and child’s choice. Negative evaluations of the mother and child were positively correlated. Findings based on the follow-up Likert scales did not consistently reflect the patterns revealed in the narratives. From a scholarly perspective, this study provides new information about the processes involved in moral socialization, and how parents come to conclusions about their children’s wrongdoings. Implications for parenting are discussed