| Title | Keywords | ||
|---|---|---|---|
| Author | Authorship | ||
| Corresponding Author | Funds | ||
| DOI | Column | ||
| Summary | |||
| Timeframe | - | ||
| Title | Keywords | ||
|---|---|---|---|
| Author | Authorship | ||
| Corresponding Author | Funds | ||
| DOI | Column | ||
| Summary | |||
| Timeframe | - | ||
[Objective] To meet the targets set in the “Carbon Peak” and “Carbon Neutrality” initiatives, it is imperative to enhance the energy-saving and environmental protection capabilities of the aluminum industry. In this industry, optimizing the combustion process in aluminum melting furnaces is vital for enhancing thermal efficiency and curtailing NOx emissions, thereby ensuring energy conservation and emission reduction. [Methods] In this study, the multifield coupled combustion process in an aluminum melting furnace is simulated to analyze the combustion characteristics of the furnace. The analysis focuses on the distribution of temperature fields, velocity fields, and component concentration fields in the furnace. The Box–Behnken response surface methodology is employed to design experiments for achieving the highest possible heat flow density(q) on the upper surface of aluminum along with the smallest possible equivalent NO_x emissions(eq-NO_x). The correlation between different influencing factors and the optimization objectives is analyzed, the effect of multiparameters on the optimization objectives is explored, and the optimum condition is proposed. [Results] The results indicate that the magnitude of the correlation with q on the upper surface of aluminum follows the order of burner vertical inclination > oxygen concentration > burner horizontal inclination. Moreover, the magnitude of the correlation with the smallest possible eq-NO_x follows the order of oxygen concentration > burner horizontal inclination>burner vertical inclination>excess air coefficient. A balanced consideration is given to the correlation of each factor with q and eq-NO_x, leading to the proposed optimized conditions of an oxygen concentration of 18%)(volume), an excess air factor of 1.05, a burner vertical inclination of 85.5°, and a burner horizontal inclination of 83.5°. Under these conditions, q on the upper surface of aluminum is determined to be 93 958 W/m~2, while eq-NOx is calculated to be 0.002 22× 10~(–6)/(W·m~(–2)). A comparison of the optimized conditions with the actual operating conditions reveals that the maximum temperature of the furnace and the temperature inhomogeneity in the furnace are reduced under the optimized conditions. Furthermore, the flame stroke of methane combustion is extended, and the oxygen concentration in the combustion flame stroke is decreased, the q on the upper surface of aluminum increases by 1.76%, and the eq-NO_x emissions decrease by 40%. [Conclusions] The employment of Box-Behnken response surface methodology for optimizing the combustion of the aluminum melting furnace facilitates the construction of a regression model, thus enabling the extrapolation of response values from limited test point data to non-test points. The optimization result of the response surface methodology is capable of attaining an arbitrary point that is aligned with the established constraints. Consequently, this optimization solution is rendered more accurate and can serve as a reliable guide to the structural design of aluminum melting furnaces, thereby providing a foundation for production optimization.
[Objective] The implementation of high-concentration cemented filling technology in coal mining has effectively addressed challenges related to coal extraction beneath buildings, railways, and water bodies. However, this approach faces economic and environmental limitations. The use of cement as the primary binding material in high-concentration filling contributes significantly to global CO_2 emissions, accounting for 7% of total emissions, and represents approximately 40% of the total filling cost. Therefore, replacing cement with a more cost-effective and environmentally friendly material carries significant value for mining operations. This study investigates the use of alkali-activated cementing materials as substitutes for cement in coal gangue filling. It focuses on developing a proportioning design for alkali-activated cemented gangue filling materials that balances multiple performance objectives by analyzing their composition and microstructure. [Methods] The research employs the RSM–BBD method for experimental design. Key performance indicators evaluated include the expansion degree, bleeding rate, and uniaxial compressive strength of the hardened filling material. The study explores the influence of slurry concentration(A), cementing material dosage(B), and gangue gradation(C) on material performance. Optimal slurry proportions are determined, and the microstructure of the hardened slurry is analyzed using SEM-EDS. [Results] The results demonstrated the following:(1) The regression model established with the RSM–BBD method can accurately predict relationships among influencing factors and filling material performance, achieving model accuracy greater than 95%. Optimal filling material proportions were found to be a slurry concentration of 76%–78%, an alkali-activated cementing agent dosage of 15%–18%, and gangue crushing particle size of 2–3 mm.(2) The bleeding rate and expansion degree decrease as the values of A and B increase, while compressive strength rises. Conversely, the bleeding rate increases with higher values of C, while the expansion degree and compressive strength first increase and then decrease with higher C. The interaction of A and C, along with B and C, significantly affects 7-day strength,while the interaction between B and C significantly influences 28-day strength. The interaction of each factor has no significant effect on the bleeding rate.(3) The main hydration product in the slurry is C–S–H gel, a critical binding phase that forms a dense network structure by integrating other hydration products and fine particles. Over time, with higher cementing material dosages, the amount of C–S–H gel increases, pore structures are divided and refined, and the material's density and strength subsequently rise. [Conclusions] The alkali-activated cemented gangue filling material prepared by replacing cement with alkali-activated cementing materials demonstrates high strength, excellent fluidity, and a low bleeding rate, meeting filling requirements while reducing both CO_2 emissions and material costs. This approach offers significant benefits for advancing coal mine filling techniques and supports the development of green mines.
[Objective] DNAN4-methylcytosine(4m C) modification plays a crucial role in various cellular processes, including DNA replication, cell cycle regulation, and gene expression, making it an essential epigenetic marker. Understanding and accurately identifying 4mC sites is important for uncovering the mechanisms behind epigenetic regulation in disease and other biological functions. However, traditional 4mC site prediction technologies often suffer from high costs and time inefficiencies, limiting their scalability for large-scale applications. Although several intelligent computing-based 4mC predictors have been proposed over the past decade, their performance remains unsatisfactory. Therefore, developing effective methods to fully utilize the complex interactions within DNA sequences has become a major challenge for improving prediction capabilities. [Methods] A multilevel feature extraction module is introduced, utilizing convolutional layers, bidirectional long short-term memory networks, and an attention mechanism as core components. This setup captures long-term dependencies within DNA sequences, ensuring accurate 4mC site detection. In addition, a multiscale feature extraction module, centered on an improved SENet network, extracts multiscale expressions of location features, improving the model's ability to represent complex sequence characteristics. To further improve feature capture, a parallel feature fusion-based optimization method is proposed. Finally, to address strong imbalances in the number of candidates across different species, the class weights in the cross-entropy loss function are designed to balance the training process. [Results] A deep learning-based dual-path multiscale feature fusion approach is proposed in this work for 4mC site prediction. To validate the structural design of the model, ablation variants were performed with variants, including the SCGF-4mC, SMFI-4mC, and DCMF-4mC models. These experiments demonstrated the structural superiority of the proposed framework. In addition, the model was compared with several advanced 4mC site prediction methods currently available. Results indicate that the proposed 4mC site predictor achieved higher accuracy and stronger generalization ability. Model feature analysis experiments were also conducted using feature matrices generated by four encoding methods as inputs. Comparative evaluations using MCC and ACC metrics on an independent test set confirmed the model's stability and reliability. Meanwhile, spatial distribution calculations of 4mC and non-4mC samples across different species provided compelling evidence of the model's ability to effectively learn and recognize 4mC loci. In summary, the proposed deep learning-based method demonstrated greater accuracy and stronger generalization ability in predicting 4mC sites across six species. [Conclusions] The proposed method demonstrates the capability to identify 4mC sites in a multispecies environment, enhancing predictive performance and offering valuable support for identifying 4mC sites in DNA sequences.
[Objective] Adopting an efficient UAV formation can leverage wake surfing technology to improve transportation efficiencyand the utilization of low-to-medium altitude airspace, reduce energy consumption, and support the growth of the low-altitude economy.Therefore, choosing a reasonable formation is an important issue in UAV formation flight. To solve this problem, this study designs twoformation types, namely “V” and “I,” for three fixed-wing UAVs of the same type, inspired by the flight characteristics of geese migrationin nature. The goal is to improve aerodynamic efficiency by simulating nature's strategies for improved UAV formation performance.[Methods] Using large-scale finite element analysis software CFD and k–ω SST turbulence model, detailed 3D models of the twoformation types were created. The air flow field of UAVs during the flight was simulated. Based on numerical simulation results,comparisons were made to evaluate changes in wake vortex intensity, pressure distribution on the upper surface of the wing, and variationsin the aerodynamic parameters of the rear aircraft for both formation types. These analyses provided insights into the optimal formationand the spatial position of front and rear aircraft. [Results and Conclusions] The experimental results indicate that the tail vortex strength,wing pressure distribution, and aerodynamic parameters of the front aircraft remain consistent in both formations. However, the rearaircraft experiences significantly higher tail vortex intensity and wing pressure distribution compared to the front aircraft. Furthermore, therear aircraft's lift coefficient and lift-to-drag ratio increase and then decrease as the longitudinal distance grows but exhibit a continuousdecline with increasing vertical distance. For both formation types, the optimal aerodynamic efficiency was observed at a longitudinaldistance of 2.50 m, a vertical distance of 0 m, and a lateral distance of 1.89 m between the rear and front aircraft. At this configuration, thelift-to-drag ratio improvement for the rear aircraft reached 16.77% in the “V” formation and 13.41% in the “I” formation.
[Objective] The signal of thin coatings overlaps in the time domain, making it challenging to directly apply the time-of-flightmethod for thickness detection. The accuracy of model-based methods combined with optimization algorithms depends on both theprecision of the coating's optical parameters and the modeling precision. These methods must also address the anisotropic properties ofcarbon fiber composite materials. However, in practice, the propagation path of terahertz signals changes only at the interface, and thesample response to terahertz signals can be regarded as approximating a linear system. [Methods] Since the front part of the terahertzsignal waveform already contains critical interface information of the coating, and the response of the sample behaves as an approximatelinear system, the propagation path of the terahertz signal changes only at the interface. Therefore, signal sparse decomposition combinedwith the time-of-flight method was employed to detect thin coating thickness on carbon fiber composite substrates. First, a comprehensiveanalysis of compressive sensing theory was conducted, identifying the spectral projection gradient algorithm with L1 norm as suitable, asit has been validated on isotropic bases. Experimental evidence clarified that the reflective terahertz time-domain systems are preferablefor detecting coatings on substrates. Subsequently, based on practical coating thickness measurements, the principle of the SPGL1 algorithm was derived, and a perception matrix was constructed using reference signals. A coating thickness detection scheme combiningsignal sparse decomposition and time-of-flight method was proposed by analyzing the conditions required for solving convex optimalproblems.[Results] Experimental data demonstrated that when coating refractive indices differ significantly and the thickness is greaterthan or equal to 100 μm, the results, even with added noise, align well with ideal expectations. When coating refractive indices are similarbut the thickness is greater than or equal to 100 μm, the SPGL1 method exhibits strong anti-interference capability. [Conclusions] Thiscomplete experimental design promotes a deeper understanding of fundamental theories and methods, such as time-domain spectroscopy,sparse decomposition, and the time-of-flight method. It bridges theoretical knowledge and practical application, fostering students' abilityto connect the two while cultivating their interest and skills in scientific research.
[Objective] As an important protection and control device in the power system, the switchgear inevitably experiences different types of partial discharges because of its harsh working environment. Thus, the accurate identification of fault types is crucial to ensure the safe and stable operation of the power system and prevent equipment damage. However, at the present stage, the characterization of fault information during partial discharge fault identification in the switchgear cabinet is difficult, and the accuracy of partial discharge fault identification is low. In this study, we propose the automatic optimization of successive variational mode decomposition(SVMD) and support vector machine(SVM) parameters based on the IIVY algorithm to realize efficient identification of different partial discharge types. [Methods] First, we develop three multistrategy fusion methods using spatial pyramid matching chaotic mapping for initialization parameters, adaptive t-distribution for decision updates, and dynamic adaptive power selection for mode updates. On this basis, we propose the IIVY algorithm. Second, we develop a partial discharge feature extraction strategy based on IIVY-SVMD-MPE. This uses the IIVY algorithm to adaptively select the SVMD penalty factor α, combine it with correlation coefficients to filter the three largest IMF components, extract the multiscale permutation entropy(MPE), and construct the multidimensional fusion feature dataset. Third, we establish a switchgear localized discharge fault identification model based on IIVY-SVM using the IIVY algorithm to select the three largest IMF components for the MPE and construct a multidimensional fusion feature dataset. Finally, we establish a fault identification model based on IIVY-SVM for efficient identification of partial discharge types. The IIVY algorithm adaptively optimizes the penalty parameter C and kernel parameter σ in SVM, resulting in a fault identification model with the optimal parameter combination. [Results] This study combines the experimental data, compares the 10 fault identification models, and draws the following conclusions:(1) The IIVY algorithm proposed in this study is more advantageous than the three original optimization algorithms in the hyperparameter adaptive optimization under the same conditions, which proves the high efficiency of the proposed improvement strategy.(2) The pattern recognition model SVM is more suitable for partial discharge fault identification than BP and ELM.(3) MPE can be used to extract the fault features carried by the signal more comprehensively.(4) The adoption of a single signal processing or feature extraction method has a large impact on the accuracy of fault recognition, and the model proposed in this study can efficiently process the original signal and extract fault features.(5) Overall, the comprehensive recognition accuracy of the fault recognition model proposed in this study reaches 98.8%, in which the recognition accuracies of the pin–plate discharges, discharges along the surface, suspended discharges, and air gap discharges are 100%, 100%, 100%, 95% and 100%, 95% and 100%. [Conclusions] By establishing a multi-strategy fusion method based on spatial pyramid matching chaotic mapping, adaptive t-distribution, and dynamic adaptive weighting based on the IIVY algorithm, we propose and establish a partial discharge feature extraction method based on IIVY-SVMD-MPE and a partial discharge fault identification model based on IIVY-SVM, utilize the IIVY algorithm adaptive optimization of the SVMD penalty factor α with the penalty parameter C and kernel parameter σ in SVM, and realize the fault recognition model. The test results showed that the fault identification model established in this study has an identification accuracy of 98.8%, which effectively improves the fault identification accuracy and stability and provides a reference for partial discharge fault identification in the switchgear.
[Objective] Cloud desktop laboratories are essential in supporting experimental teaching in colleges and universities. However,managing computer hardware faults in large-scale cloud desktop laboratories remains a challenge due to the high number of terminal faults.Existing fault diagnosis methods are often inefficient and fail to provide timely prewarning for hardware faults(physical faults), such asissues with CPU, memory, motherboards, hard drives, and power supplies. To address the uncertainty of downtime caused by hardwarefaults in such environments, an intelligent prewarning model for computer hardware faults in cloud desktop terminals is proposed, utilizingcloud desktop technology and machine learning algorithms. [Methods] This study leverages a hardware status perception system designedfor cloud desktop laboratories with VDI, VOI, and IDV fusion architecture. The perception system collects data from terminal computers,including cumulative usage time, utilization rates, energy consumption, usage frequency, repair history, hardware changes, and variouswarnings such as CPU high temperature or load, memory high load, disk IO high load, insufficient hard disk space, graphics card hightemperature, abnormal crashes or blue screens, and network issues. Furthermore, the system records whether hardware faults haveoccurred. The collected data is used to train and evaluate the intelligent prewarning model with machine learning algorithms, includingKNN, decision tree, support vector machine, and XGBoost. The data set, sourced from real production environments provided by a cloudcomputing company, undergoes transformation, cleaning, and normalization to enhance training accuracy. The data size is reduced from9 850 × 18 dimensions to 9 630 × 15 dimensions. The training set and test set account for 80% and 20% of the total data, respectively. Thetraining set includes 7 704 samples(6 576 nonfault data and 1 128 fault data), while the test set contains 1 926 samples(1 644 nonfaultdata and 282 fault data). To evaluate the robustness and generalization ability of the prewarning model, the positive and negativeproportions of the data samples are adjusted, and the performance indicators, namely precision, recall, F1-score, accuracy, and AUC, arecalculated for all four models. [Results] Experimental results show that the XGBoost-based prewarning model demonstrates superiorrobustness and generalization compared to other models. [Conclusions] This intelligent prewarning model for large-scale cloud desktoplaboratory terminal computer hardware faults, built on machine learning, offers significant economic and practical benefits. It notablyimproves fault prewarning accuracy, reduces costs related to extensive computer updates, strengthens the refined management oflaboratories, and advances the overall level of laboratory construction management.
[Objective] The study addresses the challenge of dynamic obstacle avoidance for robot manipulators operating in unstructured environments. Traditional motion planning algorithms often struggle with real-time adaptability and responsiveness to dynamic changes, especially in scenarios involving nonstatic obstacles and targets where the ability to adapt quickly and accurately is crucial for safe and efficient operation. Therefore, this research aims to develop an advanced algorithm based on deep reinforcement learning(DRL) that effectively balances dynamic obstacle avoidance and target tracking, ensuring the safe and efficient operation of robot manipulators in complex, unpredictable scenarios. [Methods] To achieve this goal, a DRL framework using the soft actor-critic(SAC) algorithm was designed. The SAC algorithm, known for its suitability in continuous control tasks, uses neural networks to handle high-dimensional tasks without requiring precise environment modeling. The robot manipulator learns optimal control strategies through trial-and-error interactions with the environment. The proposed method incorporates a comprehensive reward function that balances critical factors, including end-effector and body obstacle avoidance, self-collision prevention, precise target reaching, and motion smoothness. This comprehensive reward function guides the learning process by providing clear feedback signals that encourage the agent to develop efficient and safe behaviors. The state space provides a comprehensive representation of the environment, incorporating crucial details about the robot manipulator, obstacles, and target. It includes joint angles, joint velocities, end-effector positions and orientations, as well as key points on the manipulator's body. This holistic representation of the environment ensures that the agent has all the necessary information for making accurate and efficient decisions. The action space is defined by joint accelerations, which are transformed into planned joint velocities and communicated to the manipulator for control. This control strategy effectively eliminates motion singularities, enabling smooth and continuous operation. [Results] The algorithm is trained in a simulation environment that leverages Python and the PyBullet simulator, providing a realistic and efficient platform for agent training. This environment is encapsulated using the Gym framework and integrated with the Stable-Baselines3 library to facilitate smooth agent–environment interactions. Extensive simulations demonstrate the algorithm's ability to learn effective dynamic obstacle avoidance strategies, with average reward and success rate curves showing noticeable improvement and eventual stabilization. These results indicate that the model achieves a relatively stable state, capable of navigating complex and dynamic environments. The trained model is subsequently deployed on a real robot manipulator equipped with a visual servoing system. This setup includes a Realsense D435 camera and an Onrobot gripper attached to a UR5 manipulator. The visual servoing system employs ArUco markers for detecting obstacles and targets, while OpenCV handles image processing and pose estimation,enabling real-time environmental perception and precise manipulator control. Experimental results validate the algorithm's practical effectiveness, as the robot successfully avoids dynamic obstacles and reliably reaches target positions regardless of the direction of obstacle motion. Quantitative analysis reveals that the end-effector's position error with respect to the target converges to zero, and joint velocities remain smooth throughout the operation. These results validate the algorithm's precision and reliability. [Conclusions] This study successfully develops and validates a DRL-based dynamic algorithm for obstacle avoidance in robot manipulators. By utilizing the soft actor-critic algorithm and a well-structured reward function, the proposed method demonstrates superior performance in navigating complex, dynamic environments. Deployment of the trained model on a real robot manipulator, integrated with a visual servoing system, further validates the algorithm's practical applicability. These results highlight the potential of DRL in enhancing the autonomy and adaptability of robot manipulators, paving the way for future research in intelligent robotic systems.
[Objective] The rapid development of large language models(LLMs) has significantly advanced code generation tools. Totackle issues such as the gap between theory and practice, lack of experience, and the high complexity of programming in the experimentalteaching for the Principles of Compiler course, this study designs and implements an experimental teaching system powered by afine-tuned StarCoder2 LLM. The system offers personalized learning support during experimental sessions in the course, helping studentsdevelop skills in design, implementation, optimization, verification, and programming. [Methods] This paper designs and implements anexperimental teaching system based on fine-tuned StarCoder2 LLM with a progressive teaching approach, allowing students to build asolid foundation in each stage of learning. By processing step-by-step through the design and implementation stages, students can deepentheir understanding of compiler systems. The StarCoder2 model is fine-tuned specifically for the Principles of Compiler course, withprogramming tasks and theoretical concepts tailored for this purpose. This fine-tuning ensures the model specialization and effectivenessin code generation and explanation, enabling real-time, personalized support for students at each stage of their learning process. [Results]This progressive teaching approach helps students gradually grasp complex concepts and skills related to the principles of the compilercourse. The fine-tuned StarCoder2 LLM not only increases students' engagement and motivation but also offers tailored guidance andsupport, enhancing their programming and problem-solving abilities. Through interactions with the fine-tuned model, students achieve abetter understanding of the compiling process through interaction, completing experimental tasks more efficiently. This elevates theiroverall comprehension of compiler principles. [Conclusions] By designing and implementing this experimental teaching system based onthe fine-tuned StarCoder2 LLM, this paper effectively addresses common challenges in the teaching of compiler principles. Theprogressive teaching approach deepens students' understanding of the subject while enhancing their programming skills andproblem-solving abilities. This teaching system presents opportunities and ideas for the principles of the compiler course and fostersfurther development and improvement. It is worthy of broader application and promotion in other courses as well.
[Objective] The multiencoder and single decoder(MESD) architecture has demonstrated success in change detection tasks. This method employs Siamese encoders to independently extract features from bitemporal images, effectively preserving the target features of each temporal image and enhancing the ability to capture changes in input data. Typically, MESD architectures use convolutional neural networks(CNNs) or transformers as encoders. CNNs are highly effective in capturing complex and contextually relevant features, and they can extract local features effectively. However, CNNs struggle to capture global dependencies. Conversely, transformers, with their self-attention mechanism, excel at handling complex spatial transformations and capturing long-range feature dependencies, thereby providing robust global representation. However, transformers may occasionally overlook local feature details, which reduces the clarity of foreground and background elements. This limitation can propagate through the fusion process, resulting in blurred object boundaries and false changes in the final change map. Using CNNs or transformers separately as encoders presents challenges in terms of balancing global and local feature extraction, potentially hampering feature representation. In addition, most MESD methods employ a single decoder that can fully leverage the diverse and rich information provided by multiple encoders, thereby influencing the accuracy and precision of the change detection results. To address these issues, this study proposes an interactive multiencoder and multidecoder network(IMEMD-Net). [Methods] The proposed IMEMD-Net employs Siamese interactive encoders to extract local and global features from bitemporal images. The interactive encoder comprises a parallel structure that combines CNNs and transformers to maximize the retention of local features and global representations. The local features extracted by CNNs and the global features extracted by transformers are enhanced using a feature communication module(FCM), which continuously resolves semantic discrepancies between the two. The fusion process significantly improves the global perception of local features and integrates local details into global representations. Next, global and local difference decoders process the multiscale local and global features extracted from the bitemporal images, with each decoder focusing on specific types of difference features. This targeted processing reduces the computational burden and improves overall performance. Finally, a spatial channel difference fusion(SCDF) module adaptively fuses the difference features obtained from both decoders across spatial and channel dimensions, effectively enhancing relevant changes while suppressing false positives. [Results] The algorithm's performance is evaluated using metrics such as precision, recall, F1-score, intersection over union(IoU), and overall accuracy. Compared to existing state-of-the-art MESD-based methods, such as ChangeForm, the proposed algorithm achieves F1-score improvements of 1.14%, 5.68%, and 6.5%, along with IoU score increases of 1.92%, 9.66%, and 10.73% on the LEVIR-CD, WHU-CD, and DESIF-CD datasets, respectively—without requiring pretrained backbone feature extractors. [Conclusions] This approach effectively integrates CNNs and transformers through the FCM, enhancing interaction to ensure the extraction of global and local features from bitemporal images. The use of dual decoders alleviates the limitations of single-decoder architectures by distributing the task of processing local and global difference features. In addition, the SCDF module improves the method's ability to extract more discriminative change features, which significantly enhances the efficiency of change detection in remote sensing images.