ABSTRACT The rapid adoption of ChatGPT for generating calculus exercises in engineering education faces critical quality challenges, including mathematical inaccuracies (17% symbolic errors) and pedagogical misalignment (41% dimensional inconsistencies). This study establishes an automated quality control framework featuring two innovations: (1) a five‑dimensional assessment protocol evaluating mathematical correctness (SymPy/Wolfram API), engineering relevance (PCK‑based expert scoring), cognitive alignment (Bloom taxonomy), clarity (Flesch index), and cultural safety (rule‑based filtering); (2) an open‑source toolchain MathQAValidator integrating constraint programming for prompt optimization. Analyzing 600 AI‑generated exercises, we find: (a) a 145% relative improvement in the qualification rate (the proportion of exercises meeting all five criteria) from 32% to 79% after prompt optimization; (b) critical failure patterns including 29% vector direction errors; (c) expert strictness increases by 0.38 points per teaching year. Deployed as a Chaoxing plugin, the framework provides visualized quality diagnostics for MOOC platforms. This work establishes a pedagogical‑content‑knowledge (PCK)‑informed validation paradigm for AI‑generated educational content.
ABSTRACT Developing spatial and physical intuition for structural behaviour is a central challenge in undergraduate engineering education, and virtual reality (VR) offers a compelling medium for addressing it through immersive, interactive exploration. However, most published research focuses on the design of VR platforms, with limited attention to how instructors should organise and deliver VR‐based teaching sessions. Without deliberate pedagogical design, the novelty of immersive environments can dominate student attention at the expense of conceptual engagement. This paper presents a three‐phase teaching methodology for VR laboratories in structural engineering, which comprises: (1) an onboarding phase that manages the novelty effect through guided familiarisation and controlled exploration, (2) a group demonstration in which the lead Teaching Assistant (TA) introduces key concepts within the VR environment, and (3) a team‐based problem‐solving activity with formative assessment. In this study, the methodology was deployed across two undergraduate modules in structural mechanics at Imperial College London (Structures I and Structures II), reaching a combined eligible cohort of approximately 300 students per year. The methodology was developed over 2 academic years: an initial year of unstructured deployment that informed the design of the three phases, and a subsequent year of structured delivery that constitutes the evaluation reported in this paper. Evaluation surveys indicate high engagement across both deployments, with interest and enjoyment scores between 8.9 and 9.7 out of 10. Self‐reported understanding scores (7.1–8.6) are more moderate, consistent with the fact that the VR sessions target visual intuition rather than primary conceptual instruction. The methodology is described in sufficient detail for direct adoption by engineering educators at other institutions.
ABSTRACT This research was developed in response to the need for new ways of teaching Chemistry in order to improve the understanding of molecular geometry. As the development of technology advances, it is also considered that it should be applied within the educational environment, as it will prepare students for the future. This work was carried out using an action‐research methodology, and both quantitative (statistical) and qualitative (ideas or opinions) data were analyzed. In addition, Blender and Unity software were used to build the augmented reality (AR) application. Results showed an increase on the levels of creativity, innovation, and motivation displayed by the participating students, as well as a better academic performance in the chemical subject of molecular geometry. Thus, it can be concluded that the teaching‐learning process can be positively influenced using new tools. This research implies that the application of immersive technologies such as AR in education favors the development of a meaningful learning and allows students to better understand the usefulness of knowledge and its applicability in the contexts of their daily lives. Such investigations are especially valuable in STEM disciplines, since they allow the students to enjoy lessons while staying in touch with new technologies that are already native to them.
ABSTRACT With the rapid advancement of intelligent manufacturing and the digital transformation of vocational education, traditional teaching models in mechanical manufacturing are undergoing a critical shift from physical training to virtual–real integrated learning. This study presents a self‐developed Metal Cutting and Tool Virtual Simulation Software to construct a comprehensive virtual learning ecosystem for mechanical engineering education. Supported by AI‐driven interactive simulation and multi‐module competency‐oriented design, the system integrates virtual workshop operations, tool cognition, and technical drawing interpretation into a unified teaching framework. Multiple rounds of teaching practice provide preliminary evidence that the system is associated with higher self‐reported operational proficiency, learning engagement, and safety awareness, while supporting teachers in implementing data‐informed instructional practices. These findings suggest the educational potential of the proposed platform rather than demonstrating direct improvements in objective machining performance. This paper elaborates on the system architecture, functional modules, AI‐driven mechanisms, and pedagogical reform strategies, providing an alternative integrated teaching framework for mechanical manufacturing education.
ABSTRACT Adaptive Education Systems (AES) have become a key part of modern education. They try to meet the needs of all kinds of learners and address the problems posed by one‐size‐fits‐all educational approaches. In this context, AI‐driven methods have enabled personalization, adaptation, and data‐driven decision‐making in education at the content, learning path, and assessment levels. But using AI in AES is hard because it relies on data, can't be scaled up, raises ethical questions, and lacks sufficient, consistent empirical evidence. This paper offers a thorough, systematic overview of novel concepts, models, and trends in AI‐driven AES. The main objective of this study is to analyze the dominant research directions with a special focus on Pathway Personalization with 32% and Evaluation Evidence with 14% as the two dominant axes of recent studies. The findings show that AI‐driven approaches effectively design adaptive learning pathways, enhance intelligent educational support, and improve the accuracy of educational decision‐making. At the same time, the equity and ethical governance dimensions remain underdeveloped. Altogether, this research, by providing a coherent analytical framework, can serve as a conceptual basis for researchers, educational system designers, and policymakers in advancing targeted, generalizable development of AES.
ABSTRACT To address the limitations of physical field simulation and optimization in the mechanical optimization design course, this study develops a comprehensive set of multi‐physics simulation teaching cases centered on power battery protection structures. The proposed teaching case integrates three core modules: fluid‑thermal coupling simulation, mechanical loading response simulation, and multi‐objective optimization design. It systematically elaborates the corresponding theoretical foundations, modeling workflows, and key parameter configurations, and further refines the teaching design, class hour allocation, and teaching assessment mechanism. Preliminary teaching practice indicates that students report high satisfaction with this case. The case effectively improves students' practical abilities in multi‐physics simulation modeling, structural performance analysis, and multi‐objective optimization design. With strong teachability and engineering reference value, it provides a valuable reference for the construction and reform of similar professional courses.
ABSTRACT To bridge the theory‐practice gap in Water Pollution Control Engineering education, this study introduced an innovative teaching model that deeply integrated the professional simulation software BioWin with a year‐long, real‐world operational optimization project from a full‐scale municipal wastewater treatment plant (WWTP). This authentic case was transformed into a comprehensive virtual simulation project, guiding students through the complete engineering decision‐making workflow. Through a quasi‐experimental study with 80 students, the model demonstrated its effectiveness by translating abstract principles into quantifiable practice, significantly improving understanding of process dynamics (pre/post‐test, p < 0.001), and enabling over 90% of students to achieve high‐fidelity model calibration (e.g., average deviations of 4.6% for MLSS and 0.5 mg/L for NH 4 + ‐N). Crucially, the experimental group outperformed the control group in final exams ( p < 0.001) and complex scenario analysis. The model enhanced students’ competencies in model building, operational optimization, data analysis, and systems thinking, providing a replicable, evidence‐based framework for cultivating data‐driven engineering talent.
ABSTRACT Modular construction is increasingly adopted in industry; however, undergraduate civil engineering students often struggle to comprehend assembly sequencing and structural connection logic when instruction relies primarily on two‐dimensional (2D) drawings and text‐based materials. These limitations hinder the development of spatial visualization and procedural cognition, skills essential for modular construction planning and coordination. To address this instructional gap, this study designed and evaluated an augmented reality (AR)–supported Building Information Modeling (BIM) learning environment named AR‐ModuConnect aimed at enhancing students' conceptual and procedural understanding of modular steel building connections. The instructional approach enables interactive three‐dimensional visualization of Intra‐Module, Inter‐Module, and module‐to‐foundation connections, supported by animated assembly sequencing. A quasi‐experimental implementation was conducted with civil engineering undergraduates to compare learning outcomes between the AR‐supported instruction and conventional methods. Learning effectiveness was assessed through measures of conceptual understanding, spatial visualization, and procedural clarity, while usability was evaluated using the System Usability Scale (SUS). Results indicate that students exposed to the AR–BIM instructional approach demonstrated significantly improved comprehension of connection systems and assembly processes. The SUS score of 69.46 suggests acceptable usability for integration into engineering curricula. These findings contribute empirical evidence to immersive technology–enhanced learning in construction education and highlight the potential of AR‐supported BIM environments for strengthening procedural learning in modular construction courses.
ABSTRACT This review examines the evolution of undergraduate aerospace engineering (AE) laboratories from 1990 to 2025, focusing on course formats, pedagogy, technology, and assessment. Once centered on traditional demonstration‐based exercises, AE laboratories have increasingly shifted toward hands‐on, project‐based, and hybrid physical‐virtual models that better connect theory with practice. The COVID‐19 pandemic accelerated the adoption of remote and online laboratories, which expanded access but also raised questions of authenticity and engagement. Alongside these pedagogical changes, technological advances have reshaped laboratory instruction: wind tunnels, strain gauges, and flight simulators remain central, but computational fluid dynamics (CFD), additive manufacturing (AM), and modern techniques such as particle image velocimetry (PIV), pressure sensitive paint (PSP), and digital image correlation (DIC) are playing growing roles. Emerging approaches include digital twin frameworks that couple real‐time data with simulation, virtual and augmented reality platforms that enhance immersion, and applications of artificial intelligence for automated analysis and adaptive control tasks. Sustainability has also become a driver of laboratory design, with new experiments emphasizing electrical propulsion and aeroacoustics. Despite these advances, assessment practices remain dominated by lab reports with limited innovation. The review concludes that systematic evaluation and strategic integration of advanced technologies are essential to sustain the relevance and impact of AE laboratory education.
ABSTRACT The engineering profession faces global challenges that require graduates with strong practical and cognitive skills, yet traditional curricula often lack real‐world and interdisciplinary preparation. Immersive Reality (IR) technologies, including Virtual, Augmented, and Mixed Reality, are increasingly explored in engineering education to bridge this gap. This systematic literature review examines how IR supports learning outcomes, categorising studies by technology, instructional design, and assessment methods. Using Bloom's Taxonomy, the findings reveal that most interventions target the ‘Apply’ level, with limited focus on higher‐order cognitive skills. A key gap identified is the limited use of pedagogical frameworks such as Bloom's Taxonomy and Constructive Alignment (CA), which hinders evaluation of instructional depth. To address this, the paper presents two illustrative designs that demonstrate how immersive Teaching and Learning Activities and Assessment Tasks (ATs) can be designed using CA principles. These cases offer practical guidance for embedding IR into engineering curricula in a pedagogically coherent and outcome‐aligned manner.
ABSTRACT In recent years, materials engineering education has undergone significant reform due to the transformative impact of computer simulations. This shift towards a more digital curriculum requires accessible and intuitive software for both educators and students. This study evaluates the use of a free and user‐friendly data mining and machine learning platform (Orange Data Mining) as a possible pedagogical tool to teach basic concepts of artificial neural networks (ANNs) in materials engineering education. The platform enables the creation of customized workflows and simplifies complex data analysis without requiring programming skills. As a practical application, the study investigates the prediction of inorganic compound solubility in water (412 compounds) and the dissolution rates of inorganic glasses (295 entries from 19 compositions) using experimental data obtained from the literature. Input features comprised elemental composition, temperature, and pH conditions. The best‐performing model consisted of an ANN with 30 neurons in a single hidden layer, using the ReLU activation function—selected for its computational efficiency and suitability for nonlinear modeling—and 400 iterations, chosen to balance model capacity and generalization in a teaching‐oriented scenario. The results demonstrated high predictive accuracy, with a mean squared error (MSE) of 0.5 for solubility and 0.03 for dissolution rates. Additionally, the study illustrates both classical and modern interpretations of ANN behavior, including overparameterization effects. Overall, the findings indicate that Orange is an effective and accessible tool for teaching machine learning concepts in introductory classrooms, allowing the exploration of complex chemical and physical relationships using simple computational resources.
The integration of chatbots and generative artificial intelligence (AI) tools into engineering education is rapidly changing the way students learn, interact, and solve complex problems. These technologies offer new opportunities for personalized learning, real-time feedback, and enhanced student engagement. However, a comprehensive understanding of their implementation, pedagogical value, and limitations in engineering education remains limited. This structured literature review examines how chatbots and generative AI tools are integrated into engineering education and evaluates their educational impact and associated challenges. The review focuses on their effects on student learning outcomes, engagement, and skills development, as well as the challenges associated with their implementation. Following PRISMA guidelines, literature was identified through Scopus, Google Scholar, Taylor & Francis Online, and ScienceDirect. After applying predefined inclusion criteria, 16 studies were included in the final review and were analyzed thematically. The findings show that chatbots and generative AI tools can improve learning outcomes in engineering education by promoting student engagement, supporting conceptual understanding, encouraging self-directed learning, and contributing to the development of problem-solving and creative thinking skills. However, the review also identifies challenges related to ethical concerns, accuracy limitations, and the risk of student overreliance, which can impact the development of deeper learning and collaboration.
Simulating transient heat transfer is crucial for computational thermal analysis in manufacturing processes such as arc welding, laser powder bed fusion, and directed energy deposition. These processes feature moving heat sources that generate steep spatial and temporal temperature gradients, which in turn strongly influence microstructure formation, residual stress development, distortion, and final mechanical properties. As a result, accurate and efficient simulation tools are invaluable for both research and industrial applications. This tutorial presents a streamlined, Python-based finite element (FE) framework for simulating three-dimensional transient heat conduction with a moving Gaussian heat source over a parallelepiped domain. Distinct from complex commercial or open-source FE codes, this approach uses the lightweight and transparent scikit-fem library for spatial discretization. This enables efficient assembly of system matrices and clear correspondence between mathematical formulation and code implementation. Time integration is handled using the generalized theta -method, covering Forward Euler, Backward Euler, and Crank-Nicolson schemes; the Crank-Nicolson option ( theta = 0.5 ) is chosen for its unconditional stability and second-order accuracy. Nonlinear boundary conditions due to combined convective and radiative heat losses are included via Newton-Raphson iteration with an analytically derived Jacobian. Each Newton linear system is solved using a conjugate gradient algorithm preconditioned by algebraic multigrid (AMG), ensuring efficiency for moderately fine three-dimensional meshes. The complete simulation code is concise, at roughly 99 lines, making it well-suited for teaching, self-study, or rapid prototyping. The resulting temperature fields are benchmarked against published FEniCS results for the same moving heat source scenario, exhibiting a relative field difference of only 1.5%.
This study addresses the limitations of traditional instruction in the "Intelligent Systems" course by developing an integrated teaching model that connects theory, methodology, practice, and competency development. The reform converts surface interest into deep drive while furnishing a scalable knowledge scaffold and hands-on readiness for frontier domains. To achieve this, the course content has been redesigned to incorporate contemporary concepts from knowledge management systems and core algorithms for intelligent information processing. At the practical level, a project-based learning approach is adopted, centering on motor speed and position control as an applied scenario. A progressive, hands-on experimental framework guides students from open-loop control through closed-loop control to intelligent control. This case-driven design bridges theoretical principles and engineering practice, fostering integrated abilities in intelligent sensing, computing, and actuation. Learning is assessed through a multi-component evaluation system that considers in-class performance, laboratory work, and final examination results. Teaching practice demonstrates that the model enhances students' ability to transfer knowledge, design systems, and innovatively solve complex engineering problems, with noticeable improvement in their practical engineering literacy and higher-order thinking. Although the study is limited by its sample size and scope as an elective-course experiment, it provides a transferable pedagogical prototype.
As intelligent manufacturing systems become increasingly complex, effective instructional tools are critical to support learners' understanding of intelligent production and operational control concepts. This study presents an educational virtual simulation system for teaching the operation and control of intelligent production lines. Developed around real industrial workflows, the system provides an interactive three-dimensional learning environment that facilitates comprehension of production line structure, process planning, workflow design, and motion control. The simulation incorporates task scheduling strategies and control logic to model multi-variety, small-batch production scenarios, enabling learners to explore process routing, takt time adjustments, and responses to abnormal conditions. Instructional evaluation was conducted with 62 students in a controlled pre-test/post-test experiment and 178 questionnaire respondents. ANCOVA showed that the experimental group achieved significantly higher adjusted post-test scores than the control group, F(1, 59) = 53.10, p < 0.001, partial eta(2) = 0.474. The experimental group also completed operational tasks faster than the control group (7.85 +/- 2.65 min vs. 10.00 +/- 2.55 min) and achieved higher operational accuracy (95.28% +/- 3.71% vs. 90.62% +/- 4.25%). These findings suggest that the proposed system can enhance students' conceptual understanding and operational performance in intelligent manufacturing education.
This article presents a low-cost data acquisition architecture for control engineering education, enabling integration with new pedagogical systems or leveraging legacy professional setups. The proposed system uses Arduino boards together with a custom shield with digital to analog converter (DAC), encoder interface, and analog modulation, and integrates with MATLAB Simulink for real-time control. A prototype was implemented and tested on a rotary flexible link and a Furuta pendulum, demonstrating its reliability and performance. The same prototype is compatible with other platforms and programming languages, such as Python and C, and can operate across different operating systems. The control of cyber-physical systems serves as a case study under which the experimental results are presented.
A new approach, 'Variable Time Fixed Result (VTFR) Assessment', which shifts the focus from time-bound assessments to mastery-based learning to accommodate students with varying learning abilities and levels of preparedness is presented in this paper. In this approach, the correctness of problem solving is not only verified for the final answer of the problem, but the intermediate steps and their correctness is also ensured. This approach was applied for assessments of four challenging concepts in Electrical Engineering, Mechanical Engineering and IT Engineering. The results obtained indicate that the student performance after VTFR assessment has significantly improved, thereby confirming better learning of the concept and also its retention. Around 70% of the students have indicated preference for VTFR assessment as compared to the traditional approach. Opportunity to retry problems with reduced anxiety levels during assessment, has been the main reason for this shift in preference.
Understanding how students conceptualize and apply thermodynamic models remains a challenge in chemical engineering education. This study addresses how a regression-based activity influences conceptual understanding, develops model-based reasoning, and shapes student engagement. The study was conducted in a third-year undergraduate course (n = 38) using Aspen Plus. A four-stage inquiry-based activity guided students through parameter regression using literature-reported liquid-liquid equilibrium data, model validation, and reflection. Learning was evaluated using questionnaires, performance-based tasks, and thematic analysis of student work. Results show that 97.4% of students reported the ability to interpret binary interaction parameters, 92.1% understood model assumptions, 89.5% could evaluate predictive accuracy, and 100% demonstrated competence in parameter regression. Qualitative findings indicate active engagement in comparing predictions and identifying model limitations. These results suggest that data-driven modeling activities can support conceptual understanding and model-based reasoning, while further validation across contexts is needed.
