We study household net load under everyday operating conditions with long-standing, non-notified pricing and relative temperature variation. Using 15-minute data for 25 Pecan Street homes from July 19 to October 31, 2019, we develop an operator-oriented pipeline that combines Difference-in-Differences (ATE), a dynamic Event Study (with explicit pre-trend diagnostics), and a Causal Forest (CATE) to quantify average, temporal, and heterogeneous responses. Price peaks are defined by wholesale-price quantiles to avoid clock confounding, while hot days are defined by relative temperature quantiles to enhance transferability. Under quantile-defined peaks, average price effects are small and statistically unstable, consistent with pre-trend evidence and potential price endogeneity; in contrast, relative hot days yield a clear but short-lived same-day increase in grid imports. Heterogeneity is substantial and interpretable: price, temperature, and household temperature sensitivity dominate, with PV/EV households exhibiting near-zero or slightly negative CATEs and others responding positively. Turning heterogeneity into action, a CATE-targeted design outperforms uniform peak-window tweaks in evening hours, although mild end-of-window rebounds motivate ramping constraints. Overall, rather than relying solely on passive price signals, integrating temperature-driven triggers, heterogeneity-aware targeting, and operational constraints offers a more practical pathway to distribution-level demand response under stable pricing.

This paper focuses on the optimization of the assessment system of comprehensive course design. It uses the switched-mode power supply (SMPS) design competition as an example to discuss using the causal inference method to establish reasonable scoring criteria. This paper reveals the limitations of traditional probability statistics in dealing with complex causal relationships by analyzing the competitors’ performance in different experiments and the “data generating process” behind them and it adopts the causal inference model to assess the competitors’ abilities more accurately. In the research, the competitors’ ability to decompose higher-order problems and the difficulty of experiments are considered potential variables, and the complex interactions between these factors are clarified through constructing causal diagrams, especially in identifying and controlling confounders and mediating variables. On this basis, this paper innovatively proposes an assessment strategy, the Enhanced Causal-Weighted Scoring Method, which aims to account for differences in experimental difficulty while more accurately reflecting competitors’ abilities in higher-order problem decomposition. The advantages of the method in scoring are demonstrated by comparing it to traditional percentage scoring. This method solves the scoring problem and provides a scientific and reasonable educational evaluation strategy for comprehensive course design and related competition fields. In addition, given the current rapid development of artificial intelligence (AI) where AI still lacks causal thinking as well as data generation processes, the results of this research provide the theoretical support and practical orientation containing causal logic for the development of an AI scoring system, which promotes the growth of intelligence and depth of the educational evaluation system.

In response to the requirements of the Ministry of Education for the construction of national first-class courses, and to promote the deep integration of teaching mode innovation and scientific evaluation, this study focuses on the optimization of the BOPPPS teaching mode, which is widely used in engineering and medical education because of its goal-oriented and closed-loop feedback design, but relies on teachers' experience in teaching adjustment and is prone to confuse the correlation and causality between measures and effects. For this reason, this study takes the front door criterion and counterfactual inference of causal inference science and migrates them to be applied to education, mediated elimination of confounding and counterfactual inference are used to optimize the BOPPPS model, and exploring a continuous improvement scheme using an information security course as an example. The results show that the improved model significantly improves the rationality of experimental design and teaching effectiveness, helps teaching decision-making shift from experience-driven to data-driven, meets the national curriculum standard of civic-political integration, and provides an innovative paradigm for the accreditation of engineering education.

In order to enhance the country’s innovation capacity and build a highly-skilled national talent pool, China aims to build a modern education system by 2035. However, current methods for evaluating and improving the curriculum system are often subjective, do not pay sufficient attention to student learning outcomes, and are not based on scientifically valid models. To address this issue, we propose a factor analysis-based method for evaluation and improvement of the curriculum system. First, we applied a confirmatory factor analysis model to assess the current ‘courses – graduate attributes’ support matrix. Next, we performed exploratory factor analysis to provide insights into potential conflicts between the factor loading-based matrix and the original support matrix. Finally, we convened a teaching steering committee meeting that resulted in an improved support matrix that offers more objective, data-driven guidance through human–machine collaboration. A case study of a Digital Electronic Technology course demonstrates the refined matrix’s application in designing course objectives and assessments. By using factor analysis to provide a scientific basis for evaluating and improving the curriculum system, this study assists educators in preparing higher quality self-evaluation reports for engineering education accreditation. While this study focuses on engineering education, the methodology is equally applicable to political science, where it can help identify relationships between curricula and the development of essential skills. This approach not only aligns with national educational goals but also strengthens the overall curriculum system.

Evaluations of program educational objectives, graduation requirements, and curriculum systems in engineering education certification are often confounded by multiple interfering factors, limiting their accuracy. Traditional experience-driven approaches, such as simplistic scoring or quantile-based methods, frequently fail to isolate these confounding influences, leading to biased assessments. To address this critical limitation, we propose a novel two-phase framework: Mediator-based Deconfounding and Continuous Improvement (M-DCI). Building upon causal inference principles, specifically leveraging the front-door criterion for path analysis, M-DCI isolates and eliminates confounding factors to accurately assess the true attainment of course objectives and subsequently drives targeted, evidence-based improvements. Using the “Information Security” course in a Communication Engineering program as a case study, we constructed a causal model and applied the M-DCI framework. This study establishes a robust, closed-loop “evaluation-feedback-improvement” mechanism, demonstrating a significant shift from experience-driven to evidence-driven practices. The M-DCI framework provides a precise and actionable methodology for continuous quality enhancement in engineering education certification.

Significant advancements in artificial intelligence (AI) capabilities have been demonstrated recently by OpenAI, with ChatGPT’s educational applications particularly standing out. AI assistance offers individualized learning plans and real-time replies and has been incorporated into power supply design education at every step of the experimental design process, from circuit concept analysis to experimental execution and result analysis. A paradigm shift in classroom evaluation has resulted from this integration, emphasizing students’ higher-order thinking abilities and various assessment points. We used the causal inference approach, particularly the front-door criterion, to evaluate a course and determine how instructional tactics precisely affected course effectiveness. The benefit of the causal inference approach over conventional course evaluation methods is that this approach successfully eliminates extraneous variables that might affect students’ experimental performance, accurately measuring the actual contribution of AI-assisted instructional strategies to achieve course objectives. An accurate way to evaluate the efficacy of this teaching strategy is to compare the changes in course objective achievements before and after removing confounding variables, as well as the differences in course objective achievements between instructional tactics with and without AI assistance. This validation shows how the use of AI assistance has improved students’ ability in all areas of laboratory course study, resulting in a marked improvement in the quality of education.

Detecting low-frequency faults in smart grids is challenging due to the class imbalance in fault data and dynamic changes in grid topology. To address these issues, this paper proposes a knowledge-guided self-attention graph neural network (KG-SA-GNN). First, a domain-specific knowledge graph is constructed by integrating equipment topological constraints, historical fault co-occurrence patterns, and maintenance rules, which enhances the modeling of physical relationships in the power grid. Second, a dynamic self-attention mechanism is introduced that incorporates relationship weights and fault frequency priors to adaptively focus on critical propagation paths and improve classification performance. Experimental results on the IEEE 39-node system show that the proposed method achieves a recall rate of 78.6% for cascading faults, outperforming SMOTE-GNN by 14.9%. The false alarm rate is reduced by 9.5%, and the macro-average F1-score reaches 87.1%. The results demonstrate that KG-SA-GNN provides an effective solution for imbalanced fault diagnosis in smart grids.

Professional accreditation of engineering education has become a crucial tool for ensuring and enhancing the quality of engineering education. How to conduct scientific and effective evaluation and continuous improvement has become a prevailing topic. This study focuses on integrating the causal inference method into a graduation design evaluation system, adopting the causal diagram and the structural causal model as the core analytical tools, and selecting a case study of a university to reconstruct the teaching improvement process and assess the causal effect of the learning process on the quality of students’ graduation theses by gradually introducing the process evaluation. Meanwhile, through multivariate data analysis, the supervising instructor was identified as a confounding variable and this variable was deconfounded to reveal the true causal effects of each link. Continuous improvement measures were then proposed. The results demonstrated that the causal reconstruction process confirmed the instructional improvement measures from the data perspective, and the deconfounding method could effectively shield the effects of confounders. Through the application of the causal inference method, this study offers a multi-perspective and multi-level evaluation approach, overcoming the limitations of the traditional single evaluation method, and provides an empirical basis and strategic direction for the improvement of engineering education quality.

To address the peak load overlapping phenomenon caused by the highly concentrated EV charging in residential areas, this study proposes an orderly charging strategy based on an improved multi-objective particle swarm optimization (MOIPSO) algorithm. A multi-objective optimization model is established to minimize the total system load variance, reduce user costs, and maximize charging volume, while considering EV battery aging costs and system power constraints. Simulation results demonstrate that, compared to the NSGA-II and MOPSO algorithms, the proposed algorithm significantly improves the diversity and convergence of the Pareto solutions through the introduction of dynamic inertia weights and elite crossover and mutation strategies. Moreover, compared to unordered charging, the proposed orderly strategy increases the average off-grid SOC by 3.3%, reduces the total cost by 29%, decreases the load peak by 28.4%, lowers the peak-valley difference rate by 39.3%, and reduces the load variance by 87.2%.

To meet the demands of the information and intelligent era and cultivate students’ ability to integrate diverse knowledge and skills to solve practical problems, interdisciplinary integrated curriculum design has become an important paradigm in education and teaching. However, due to challenges such as the interference of confounding factors, static evaluation systems, and the lack of dynamic optimization mechanisms, the design, assessment, and improvement of interdisciplinary integrated courses are difficult to be effectively supported. For this reason, based on the theory of causal science, this paper proposes the method of “introducing mediators to eliminate confounding”, adopts the structural equation model and combines counterfactual inference to calculate the causal effect, in order to analyze the intervention path. Taking the cognitive computing course as a case, a causal model including AI-enhanced research, the application of metacognitive strategies and higher-order abilities is constructed, and an empirical analysis is conducted through teaching data. The results show that the overall intervention effect is positive, and teaching intervention mainly acts on the improvement of students’ abilities through mediating variables. The research provides a scientific path for the quantitative assessment of educational intervention effects and offers theoretical and methodological support for the dynamic optimization and precise improvement of the curriculum.

The design of switched-mode power supplies (SMPSs) is a material-electrical engineering interdisciplinary problem. The design of magnetic component materials can significantly affect SMPSs’ performances. In engineering education, the design and development of solutions is an important skill for engineering students. However, the traditional engineering experiments curriculum is shown as validation and single-disciplinary experiments as well as one-sided assessment results. To foster open-ended exploration, enhance students’ material design skills, and improve skills adaptation, students must tackle real engineering problems and develop practical design solutions. This paper proposes an exploratory factor analysis and knowledge graph-based method for evaluating and continuously improving magnetic components material design skills in the context of SMPS design tasks. First, we used the multiple imputation method to address the missing data. Each imputed data was analyzed to extract factors through parallel analysis, ordinary least squares estimation, and target rotation. Then, we identified four sub-skills: efficiency design skill, passive device design skill, power magnetic reduction design skill, and power economy design skill. The RMSEA for the four-factor model is 0.042, suggesting a good fit to the data. We established the relationships between these sub-skills and SMPS performance metrics. Furthermore, the average of the factor scores from each of the imputed datasets and the SMPS design constraints were combined to obtain the cut-off scores to evaluate engineering students’ achievements in these sub-skills. Finally, we constructed an SMPS magnetic components material knowledge graph, which could recommend specific experimental tasks, relevant knowledge areas, and SMPS performance metrics, providing personalized guidance to designers.

Non-intrusive load monitoring,as an essential means for fine-grained management of household electricity consumption,plays a significant role in promoting energy conservation and emission reduction for achieving the dual-carbon goal.However,it is challenging to achieve high-precision load identification using a single voltage-current trajectory image.Therefore,a non-intrusive load identification method based on the fusion of Gramian angular difference field(GADF)image coding is proposed.First,the high-frequency steady-state data collected by the device are preprocessed to obtain a complete base-wave period current and voltage signal.Then,the one-dimensional voltage and current signals are encoded separately using the GADF to generate the corresponding two-dimensional feature images,and load identification is performed via superimposed fusion input to a neural network based on a convolutional block attention module.The public datasets PLAID and WHITED are used for testing experiments to verify the effectiveness of the proposed method.The results indicate that the method has a high recognition accuracy,with average accuracies of 99.45%and 99.24%for the PLAID and WHITED datasets,respectively.

With the continuous promotion of professional accreditation of engineering education, an education assessment method based on students' learning outcomes has been gradually formed at this stage. This is characterized by focusing on hard indicators such as students' grades and achievements and aims to establish a comprehensive curriculum evaluation system for students guided by the framework of graduation requirements of the Washington Accord. However, the design of the curriculum inevitably does not correspond to the actual situation due to various problems. In this paper, we try to explain the relationship between courses in the accreditation of engineering education by combining the assessment of students' abilities and course objectives, and propose ways to improve the design of courses, so as to provide a scientific basis and reference for the formation of a more reasonable and better engineering education system.

《华盛顿协议》毕业要求框架规定了学生应掌握的知识、具备的素质和能力,各专业依据毕业要求反向设计课程体系,并对课程体系设置的合理性进行评价,实际实施过程中的主观因素过多,客观性不足.本文运用计算思维和教育心理学,通过大数据分析探寻内部规律,将毕业要求对应的能力特征稳定性转化为各门支撑课程学生成绩分布的一致性度量,构建专业课程体系对毕业要求支撑关系的评价算法,以学生综合成绩为主要数据来源,依据计算结果分析毕业要求层面或课程层面支撑关系的短板,给课程体系设置合理性评价提供定量的参考依据.三所高校工科专业课程体系评价案例结果表明,该方法能有效发现专业课程体系存在的课程对毕业要求支撑不当的问题.

在线上课堂火热开展之际,教师如何保证线上教学与线下课堂教学效果实质等效,实现基于产出导向的翻转课堂是值得探究的.文章以湘潭大学2017级通信工程专业信息安全课程的第四次探究式学习作业为例,介绍"以学生为中心,产出导向"的线上课堂如何融合"互联网+""智能+"技术实现专业与工程伦理学习任务的演进与交互,结合学生信息安全违规事件证据推理实验的学习成果,展示教师如何构建面向产出的课程教学"智能+"评价.

Since teachers typically focus on the quantitative analysis of the data when writing the analysis report on the achievement of course objectives, there is no guarantee that the final calculation result is valid. They should give equal consideration as to whether the source of the data is also reliable. This paper combines the emerging computational psychometrics and causal inference science, and mainly solves two engineering teaching problems. First, this paper proposes to use computational psychometrics as an instrumental variable to explore its deconfounding effect in teaching assessment, which can eliminate the influence of confounding factors in engineering experimental teaching assessment. Secondly, the scientific method of causal inference is used to calculate the causal effect factor of the experimental results on the test scores from the observation data. Then, characterize the influence of the experimental scores on the test scores, thus solving the cross-modal problem of the process data participation calculation. The method proposed in this paper cannot only ensure the reliability of the data source but can also unify the calculation mode so that the degree of achievement of the course objectives can be more accurately calculated, which is helpful for teachers to continuously improve the teaching level.

Experimental courses in engineering majors in colleges and universities are suffering from insufficient class hours, single course experiment contents, and difficulty assessing the students' knowledge mastery. Based on the power supply experiment course as a platform, this paper designs a prototype system that assesses students' mastery of course-related knowledge and provides a personalized experiment push assistant. At the same time, the system can give teachers course feedback in the form of clearer data. This paper first starts with the experimental result data of senior students who have completed the experimental course and constructs the original RIMER student ability assessment model based on the teacher's experience. Secondly, it uses the knowledge graph to record the relevant information of the experiment and the ability information of the students. Finally, through the algorithms of word segmentation, part of speech tagging, and template matching, an experimental push assistant is constructed to help students analyze the weak parts of their knowledge in the form of human-computer interaction and push relevant experiments. The system will record the use of students and provide teachers with necessary curriculum feedback. The practice results show that the accuracy of the system is high, and after using the system, students' knowledge mastery has increased in varying degrees.

At the moment of constant changes in measurement research methods, causal inference methods have made great progress in social science research. The complementarity and integration of causal inference methods and education evaluation schemes have become an urgent concern for researchers in the field of engineering education. In order to explain the causal effect of professional courses supporting the requirements of non-technical graduation, and thus enhance the efficiency of education governance, this paper focuses on causal inference, The transfer is applied to the field of engineering education to provide guidance and reference for teachers' teaching plans, and the tool variable method is used to remove the influence of mixed factors and to build a causal inference model. STATA software was used to empirically analyze and study the causal effect of the communication ability of electromagnetic microwave technology and antenna bilingual teaching supporting graduation requirements of professional courses, and on this basis, relevant suggestions are put forward to regulate the curriculum teaching plan and increase the output of results. The research results show that this method can guide and regulate the course teaching plan, thereby helping to make education governance intelligent.

It is very important to innovate teaching evaluation tools. In this paper, we build a circuit design rule library in the switching power supply, and exploit deep learning tools to establish an intelligent diagnostic network model for evaluating possible errors in the circuit design process. We also implement a preliminary evaluation system to make it easy for students to use. This method we proposed can provide reference and improve efficiency for students and teachers. Firstly, we take input/output capacitance and inductance as the research object to analyze the ripple changes of input current and output voltage in different states. Furthermore, we summarize three important circuit design rules and collect experimental data through the WEBENCH simulation tool. Then the fast Fourier transform (FFT) operation is performed on this data, and the corresponding phase spectrum is calculated. Additionally, characteristic waveforms with discernability are selected to construct circuit design rule library as training samples of a neural network. Finally, the intelligent diagnostic network model is proposed and trained offline. The result shows that the method is feasible to some extent.