34254 research outputs found
Sort by
A Reinforcement Learning Approach to Time-Triggered Scheduling Tasks Based on Graph Attentive Network with Variational Information Bottleneck
No full textTime-triggered communication systems are widely used in industries such as aerospace and transportation systems. Time-Triggered Ethernet, for instance, elaborately designs time-triggered mechanisms and network schedules, enabling real-time transmission of time-triggered traffic and ensuring high communication reliability. During the operation of communication systems, abrupt line failures are a common issue. Therefore, the scheduling approach must be flexible to response system failures. However, traditional scheduling algorithms such as linear programming and heuristic algorithms typically have high computational complexity. Consequently, these methods are constrained when applied to large-scale networks. Furthermore, when faced with abrupt line failures, traditional scheduling algorithms struggle to respond promptly. In this thesis, we propose a network scheduling approach that combines Attentive Graph Network (AGN) and Reinforcement Learning (RL) to establish reliable and efficient network schedules. Within the graph network, nodes represent routers, and edges represent transmission paths within the network communication system. We also introduced Variational Information Bottleneck theory to attention network to achieve parameter pruning while enhancing system robustnessandperformance.In RL architecture, we divide agent decisions into a three-staged pipeline: transmission path selection, path phase selection, and time slot selection. During the validation phase, this paper references existing network topology structures, including trains, airplanes, and spacecraft used in space exploration. Through the validation of these real-world scenarios, the proposed scheduling architecture in this paper demonstrates significant improvements in handling incremental scheduling compared to traditional algorithms. Furthermore, when faced with abrupt line failure, the scheduling architecture we have proposed exhibits the ability to maintain stable performance. This robustness further highlights its feasibility and efficacy in real-world environments
Battery Management System for Different Chemistries of Batteries
No full textIn a variety of applications, Battery Management System (BMS) is essential for maximizing the performance and security of battery packs. This thesis focuses the design and implementations of LTC6804-1(BMS), for using different chemistries of batteries with specific Lithium iron phosphate (LiFePO4) and Nickel Manganese Cobalt (NMC) batteries. The research begins by examining the characteristics challenges associated with diverse battery chemistries of LFP and NMC cells, emphasizing their influence on battery performance. To ensure accurate monitoring, the LTC6804-1(BMS) is integrated to precisely measure cell voltages, current, and temperature. Balancing the LFP and NMC cells connected in series and protect them are one of the main topics of this study. For a battery stack to effectively reduce voltage differences between cells at 10mV, passive balancing is essential. Utilizing the LTC6804-1's passive balancing capability, cell voltages are equalized to enhance performance and prolong the life of both LiFePO4 and NMC batteries. BMS has high voltage measurement accuracy with error less than 2mV
A case study of job crafting behaviors among foreign English teachers in Taiwanese cram schools
No full textIn Taiwan, cram schools, known as buxiban, are private, government-registered, for-profit educational institutions. The system offers short-term and supplementary teaching to paying learners, and its existence stems from the high demand among students seeking to quickly improve scores or support other educational goals in a society featuring highly competitive standardized testing. Many English language education cram schools hire foreign, often native English-speaking teachers to attract students. Compared to public schools, English language cram schools have lower threshold entry requirements and offer higher salary, attracting foreigners to temporarily or permanently switch careers and devote themselves to teaching English. However, these teachers are often under-prepared for the realities of teaching, and challenges related to living and working in a foreign country. To adapt, teachers therefore develop and apply various skills while engaging in job crafting behaviors.
This study employs Tims et al.\ue2s (2012) job crafting theory to explore the job crafting behaviors adopted by foreign teachers in English language cram schools in order to work in the Taiwanese educational context. This qualitative study adopts a case study method, conducting in-depth interviews on job crafting behaviors and practices of four foreign English language cram school teachers and thematic content analysis based on interview results. Analyzing the interview data in relation to ongoing discussions of job crafting reveal that the participants exhibited job crafting behaviors aligned with three theoretical categories (increasing challenging job demands, decreasing hindering job demands, and increasing social job resources), while a fourth category (increasing structural job resources) was not observed in this study context. This study offers a discussion of the interrelationships between these types of job crafting behaviors and their respective relationships with the study context. Based on the findings, this study offers insights for job crafting theory and possible directions for future research that may add depth to the study of job crafting, both generally and in the context of foreign English teaching, and several implications for educational management practice
Design of flight attitude controller for quadcopter UAV based on reinforcement learning
No full textIn recent years, with the rapid advancement of technology, quadcopters have gain popularity due to their simple structure, flexible operation, low cost, and wide range of applications. As a result, they have been widely used in both civilian and military fields. However, quadcopters face challenges in maintaining flight stability due to their inherent high coupling and sensitivity to external disturbances. Therefore, the objective of this study is to enhance the flight stability of quadcopters by dynamically adjusting the parameters of the PID controller for flight attitude control using the Q-learning in reinforcement learning. First, we integrate a microcontroller, an inertial measurement unit (IMU) module, a receiver, electronic speed controllers, brushless DC motors, propellers, batteries, and a frame to form a quadcopter flight system. Next, the quadcopter\ue2s attitude angles during flight are measured using the IMU module. The microcontroller calculates the error between the desired stable flight attitude angles and the current quadcopter attitude angles. The attitude angle error is then used as the input signal for the PID controller. Subsequently, we design a reward optimization strategy using the Q-Learning algorithm to dynamically optimize and adjust the parameters of the PID controller, such as the proportional gain (k_p), integral gain (k_i) and derivative gain (k_d). Finally, the PID controller outputs PWM signals to control the operation of the four brushless DC motors of the quadcopter, thereby controlling its flight attitude angles to ensure flight stability during task execution. Experimental results demonstrate that the reinforcement learning-based PID controller proposed in this paper is more efficient in dynamically optimizing and adjusting PID controller parameters compared to traditional PID controllers. This allows the quadcopter to rapidly converge to stability in the external disturbances
Research on Reliability and Physical Mechanism of GaN High Electron Mobility Transistor
No full textHigh breakdown voltage, high power, and current density are all achievable with high electron mobility transistors based on GaN (GaN HEMTs). GaN HEMTs also provide enhanced interference protection and significant promise in RF microwave and power electronics applications. However, there are still a lot of performance and reliability challenges that need to be resolved, particularly those involving flaws in semiconductor materials. In this research, the operational performance and the dependability of Schottky and metal-insulator-semiconductor high electron mobility transistors (Schottky - & MIS-HEMTs) are explored, and the relevant physical processes are proposed. Additionally, a low-temperature supercritical fluids (SCF) post-processing technique has been created to enhance component performance and reliability, passivate GaN HEMT flaws and efficiently boost the on-current (Ion) of GaN HEMT.
In the research, the mechanism of the kink effect during the drain current-drain voltage (ID-VD) of Schottky HEMTs on SiC is carefully analyzed. When the kink effect occurs, the drain current-gate voltage (ID-VG) curve shows that the drop in current is linked to an increase in the threshold voltage (VT). Through Silvaco's simulated defect localization and lighting studies, it is discovered that ID-VG reacts to UV light but not to blue, green, or red light following the kink effect. As a result, the VT shift-induced kink effect is eliminated since electrons cannot be released from the defect. The reliance of the kink effect mechanism's reliance on the recombination of electrons and holes in the buffer layer has been demonstrated. The major mechanism allegedly responsible for the kink effect is caused by impact ionization (I.I.) and trapped electrons in the buffer layer and the hole recombination.
Continues to study the special current arc nodes in the kink effect of GaN HEMTs. The forward and reverse output curves of ID-VD show that the node tends to arc as the VG increases. Using With long-term voltage stress and the recovery measurement technology at the arc node, it is confirmed that the position of the arc node is related to the degree of I.I. and also represents the number of holes generated. In addition, the C-V measurement confirms that the hole is generated by the I.I. at the gate edge, which will also be verified by the temperature change experiment, and the physical model and the degradation behavior of the kink effect node will be completely proposed.
Lattice mismatch is a common occurrence in the GaN HEMT process, which causes nitrogen vacancy or dislocation defects to emerge during the epitaxial process, lowering the performance and the reliability of the device. In the first two chapters, the issues regarding dependability are also clarified; the defect passivation is critical for GaN HEMTs. Furthermore, will take advantage of the physical and chemical qualities of SCF and the traits of solid, liquid, and gas phases to develop low-temperature SCF post-processing technology, so as to deal with passivation GaN HEMT defects, effectively improve GaN HEMT Ion, reduce RS, and improve component performance and reliability.
Finally, the atypical two-stage degradation of VT under MIS GaN-HEMT off-state operation is thoroughly covered. The VT shifts briefly positively and then negatively during the off-state operation. In contrast, the gate leakage (IG) keeps getting smaller as in the off- state stress process. The experiment outcomes show that electron-hole pairs are created when a transverse electric field is strong enough. The created electrons and holes are then trapped on the Si3N4 gate insulator (GI), where they work to offset the VT tendency to shift in one direction. Additionally, it has been established through the temperature change experiment that the I.I. is a dangerous condition that is brought on by carriers from the gate during the off-state. Finally, a model is verified and suggested with various GI thicknesses
Supercritical CO2 surface treatments for interface passivation of perovskite/Si solar cells
No full textDue to the greenhouse effect and global warming, various studies about renewable energy have been researched, among them, the research concerning solar cells get the most attention. This study proposes the application of supercritical \uf0\ub6\uf0 2 surface passivation technology in the perovskite/Si 2 terminal tandem solar cells. Perovskite has emerged as a popular solar cell material in recent years, with advantages such as
high absorption of light, tunable bandgap, and lower cost. Research on
perovskite/silicon tandem cells has also been researched in recent years. This study aims to use supercritical \uf0\ub6\uf0 2 fluid carrying the oxidant ( H2O2) to repair defects on a silicon by hydrogen and oxygen bonds, growing a passivating layer (SiOx ) . Device stacking and measurements were performed, efficiency of the device will be discussed.
This study investigates the minority carrier lifetimes of supercritical fluid (SCF) passivation. After comparing with hot nitric acid wet passivation, the optimized parameter process under SCF passivation shoes the minority carrier lifetime of 26.31 \uc2\ub5s, hi gher than 22.89 \uc2\ub5s with nitric acid. Additionally, the stability of supercritical passivation was found better than hot nitric acid wet passivation. Subsequently, we test the spin coating parameters of perovskite on the SCF passivation layer, confirming t he feasibility of spin coating perovskite on the SCF passivation layer.
This study presents an application of SCF passivation in solar cells and explores potential future directions, representing a highly forward looking research direction
Design an Ultra-Smooth Surface Printing Approach for Vat Polymerization Additive Manufacturing
No full textNowadays, the vat photopolymerization technique in 3D printing methods has become popular and progressed rapidly. This is primarily due to its advantages of high resolutions, low cost, and easy operation. However, despite conducting the high-resolution LCD panel, the printing quality still has been limited by the defects of surface texture. The texture defect was observed on the print model, which was due to the black matrix structure of the LCD panel.
To obtain the high-resolution with an ultra-smooth surface, a novel dual LCD system was introduced. This system aims to eliminate the surface defect on the printed model. The smart liquid crystal film in the system was designed to adjust the direction of light by the voltage-modulated liquid crystal, which leads to light scattering. During the printing process, the scattered light path covers the black matrix area and achieves a continuous and smooth printed surface without losing resolution.
In this research, we confirmed that the new LCD 3D printing approach has operated distinct removal of the black matrix defect during the printing process. The experimental results demonstrate the successful quality of the ultrasmooth printed models without losing resolution by the dual LCD system. Additionally, the modified LCD 3D printer nearly eliminates shrinkage and improves mechanical properties, addressing the crucial challenge of LCD photopolymerization 3D printing
Using Geometric and Cutting Error Models to Enhance the Milling Precision
No full textIn the cradle type (AC) design of five-axis machine tools, when error sources are combined, they form Position-Dependent Geometric errors (PDGEs). These primarily encompass errors related to positioning, straightness, and rotation. To precisely acquire this error data, a laser interferometer was employed for measurements. Different mirror assemblies were used to measure various geometric errors, and the CARTO software was applied for data analysis.
For geometric error prediction, the laser interferometer was again used for measurements. The Least Squares Method (LSM) was utilized to establish a model related to coordinates, predicting the geometric errors of specific coordinates. The kinematic error model was adopted to correct machine accuracy, calculating the overall error caused by various error terms. The primary error sources were identified, particularly those related to machine positioning errors. The CAD/CAM software (MasterCAM) was employed to plan the tool path. Furthermore, the laser interferometer was utilized to collect spatial point coordinates, providing accurate coordinate values in three-dimensional space.
In the milling process, the three partitioned forces can be vector-decomposed into three primary directional forces: the Main Cutting Force (F_c), the Feed Force (F_f), and the Axial Force (F_a).
The cutting force model established in this study is based on the theories of Victor and Kienzle. It considers the force-receiving material area and the combination of specific cutting force products for analysis. In terms of specific cutting force, when considering a cutting cross-sectional area of A=1 mm\ue21 mm=1\ue3 mm\ue3^2, the material constant determines the magnitude and direction of the force during cutting. Factors such as the tool's rake angle, deformation of the tool during cutting, cutting speed, and wear coefficient were also considered to establish correction factors. This allowed for adjustments in the machining force based on the characteristics of the tool and workpiece during the process, refining the model.
The study adopted the least squares method and a kinematic error model to analyze the accuracy changes before and after compensation for the X360, X280, and X200 line segments. The results showed that the accuracy in the X and Y directions significantly improved after compensation. As the initial error in the Z direction was low and the impact of gravity was significant, no compensation was carried out in the Z direction. Using the TC50 3D probe and a stiffness model for verification, the predicted main cutting direction cutting force (F_c) achieved an accuracy improvement up to 98% under various cutting depths and feed rates. However, the axial stiffness of the tool and issues with chatter led to inaccuracies in axial measurement
Inclusion\uef\ubcExclusion\uef\ubcInterpretive Policy Analysis of Inclusive Education in Kaohsiung City Elementary Schools
No full textInclusive education originates from a reflection on the educational placement of students with disabilities. The proportion of students with disabilities being placed in regular schools in our country has been increasing year by year, but in the process of implementation, we have encountered many difficulties. This study uses interpretive policy analysis to explore the perspectives and experiences of different groups in local elementary schools. The results will use as a reference for future inclusive education policy development.
This study focuses on learning disabilities, intellectual disabilities, autism, emotional disturbances, and hearing impairments. The selection of target schools focus on which have been recognized by parents of students with special needs for their implementation of inclusive education in Kaohsiung City, while also including representatives from disability groups outside of the research sites. This study interviewed 12 parents (including group representatives) and 3 regular class teachers, and observed the social interactions of 4 cases in both regular and resource classes. The research findings are as follows\uef\ubc
I. In inclusive education, parents need to turn anxiety into action, expect teachers to be their children's helper, understand their children better than anyone else, and face the challenge of limited participation for most special education students.
II. The inclusive education actions taken by school teachers include (1) understanding parental anxiety through diverse communication channels. (2) actively promoting peer companionship. (3) acceptance of emotional disturbances needs cross-domain collaboration among parents, teachers, and community organizations. (4) adopting a fair stance through the use of third-person perspective.
III. Different groups have recommendations for inclusive education policies, including several key issues. (1) both parents and teachers emphasize the cooperation between general and special education to create comprehensive effects. (2) both parents and teachers recognize the need to enhance special education professionalism for general education teachers (3) both parents and teachers advocate for seamless transitions across educational stages as a starting point for creating a friendly environment. (4) teachers believe that students learn how can interact with students with special needs is beneficial for peer acceptance. (5) parents affirm that the involvement of social groups can create a friendly environment. The measures that are highly recognized by social groups include (1) providing psychological support to parents. (2) enhance society's understanding of the disabled community. (3) helping to establish self-identity for students with disabilities.
Finally, the conclusion of this study can be summarized as (1) Recognition of the vision of inclusive education policies, but exclusion still exists, and there are differences due to the category of disabilities. (2) Anxious parents combined with NGO to urge the implementation of inclusive education. (3) The key to inclusion lies in feeling empathized . The practice of inclusive education relies on the construction of supportive environments by families, schools, and society, by the process of continuously breaking through existing frameworks and gradually implementing the spirit of inclusive education
Non-metallic Visible Light Induced Ring Strain Release for the all-Carbon Formation of Azetidine Study
No full textMolecules with the structure of azetidine have become increasingly important in recent years. From the discovery in natural substances to the introduction of concepts such as bioisostere in pharmacokinetics, which can accelerate human metabolism and drug activity, all indicate the necessity of azetidine research. However, in the past synthetic processes, various limitations have hindered the smooth progress of small-ring nitrogen heterocycle research.
We optimized the methods provided in the literature and identified an efficient departure mechanism to construct highly strained azabicyclo[1.1.0]butane. In the subsequent ring strain release reaction, we eliminated the original metal-catalyzed pathway and developed a photo-driven synthetic route to construct azetidine structures with double substitution at the beta position