Table of Content

25 September 2025, Volume 53 Issue 9

Previous Issue   

Contents



2025, 53(9):  0. 

Asbtract ( 11 )  

Related Articles | Metrics

Computer Science & Technology

CODS: An Audio-Text Aligned Dataset for Cantonese Opera Vocal Synthesis



LI Yue, HUANG Yihan, PENG Zhengwei, XIE Jixuan, DU Yuye

2025, 53(9):  1-10.  doi:10.12141/j.issn.1000-565X.250134

Asbtract ( 185 )   HTML ( 1)   PDF (3373KB) ( 5 )  

Figures and Tables | References | Related Articles | Metrics

As one of the traditional Chinese arts, Chinese opera culture has unique musical expressiveness. Cantonese opera, as one of the main Chinese opera genres and an important carrier of Lingnan culture, has been indexed in the World Intangible Cultural Heritage List. In recent years, generative artificial intelligence technology has demonstrated its powerful capabilities in the field of content creation. For example, singing synthesis techno-logy can synthesize natural singing based on specified music scores. This provides a new idea for the digital protection and innovation of Cantonese opera. However, the collection and organization of opera data faces problems such as poor audio quality and complex dialect annotation, resulting in an extreme shortage of high-quality opera data sets. Based on this, this paper applied the singing synthesis technology in the field of pop music to the field of Cantonese opera vocal synthesis, and proposed the first Cantonese opera vocal synthesis dataset with phoneme-level annotation and audio-text alignment. Firstly, this paper constructed the CODS dataset through a systematic process. This dataset was derived from 29 original works by four famous performers with a total length of 3.81 hours, which provides important support for the research and digitization of Cantonese opera. Using this dataset, this paper conducted experiments with a deep learning-based method for Cantonese opera voice synthesis, realizing controllable generation in terms of lyrics, timbre, and melody. Finally, this paper established a comprehensive evaluation framework for Cantonese opera synthesis. Both objective and subjective evaluations reached a satisfactory level within the domain, further validating the usability of the proposed dataset. The CODS dataset constructed in this paper successfully filled the gap in artificial intelligence in the field of Cantonese opera vocal synthesis, and strongly promoted the inheritance and innovation of this traditional art.

Deep Reinforcement Learning Recommendation Model Based on Multi-Interest Contrast



LIU Huiting, LIU Shaoxiong, WANG Jiale, ZHAO Peng

2025, 53(9):  11-21.  doi:10.12141/j.issn.1000-565X.240088

Asbtract ( 148 )   HTML ( 1)   PDF (2268KB) ( 3 )  

Figures and Tables | References | Related Articles | Metrics

Deep Reinforcement Learning (DRL) is widely applied in recommender systems to dynamically model user interests and maximize cumulative user benefits. However, the sparsity of user feedback has become a significant challenge for DRL-based recommendation algorithms. Contrastive learning, as a self-supervised learning method, enhances user interest representation by constructing multiple perspectives, thereby alleviating the issue of sparse user feedback. Existing contrastive learning methods typically rely on heuristic-based augmentation strategies, which often lead to the loss of key information and fail to fully utilize heterogeneous interaction data. To address these issues, this paper proposed a multi-interest oriented contrastive deep reinforcement learning recommendation (MOCIR) model. The model consists of two key modules: a contrastive representation module and a policy network module. The contrastive representation module utilizes a Heterogeneous Information Network (HIN) to model the user’s local interests from different aspects while capturing their global interests based on raw interaction data. It then treats the global and local interests of the same user as positive pairs and those of different users as negative pairs for contrastive learning, effectively enhancing user interest representation. The policy network module aggregates user state representations and generates recommendations. The two modules are trained using an alternating update mechanism. Experimental results on three benchmark datasets show that the proposed model outperforms several DRL-based models in recommendation performance, effectively addressing the problem of sparse user feedback in recommendations.

Lane Line Detection Algorithm Based on Deep Learning



YUE Yongheng, ZHAO Zhihao

2025, 53(9):  22-30.  doi:10.12141/j.issn.1000-565X.240609

Asbtract ( 288 )   HTML ( 0)   PDF (2888KB) ( 1 )  

Figures and Tables | References | Related Articles | Metrics

Aiming at the problem of lane detection accuracy of intelligent vehicles in complex scenes, this paper proposed a lane line detection algorithm which incorporates a multi-scale spatial attention mechanism and a path aggregation network (PANet). The algorithm first introduced the pre-anchored frame UFLD lane detection model and incorporated a feature pyramid enhancement module PANet with depthwise separable convolution to achieve multi-scale feature extraction of images. Next, a multi-scale spatial attention module was designed in the network framework and a SimAM lightweight attention mechanism was introduced to enhance the focusing ability on target features. Then, an adaptive feature fusion module was designed to perform cross-scale fusion of feature maps output from PANet by intelligently adjusting the fusion weights of feature maps at different scales, so as to effectively enhance the network’s ability to extract complex features. Finally, the application of TuSimple dataset detection proves that the proposed algorithm achieves a detection accuracy of 96.84%, representing a 1.02 percentage point improvement over the original algorithm, and outperforms conventional mainstream algorithms. Experimental results on the CULane dataset demonstrate that the proposed algorithm achieves an F₁ score of 72.74%, outperfor-ming conventional mainstream methods with a 4.34 percentage point improvement over the baseline. Notably, it exhibits significant performance gains in extreme scenarios (e.g., strong illumination and shadows), confirming its superior detection capability in complex environments. In addition, the real-time test shows that the model infe-rence speed reaches 118 f/s, which meets the real-time demand of intelligent vehicles.

Traffic Congestion Prediction Based on Dynamic Adaptive Gated Graph Convolutional Networks



WANG Qingrong, GAO Huanyi, ZHU Changfeng, HE Runtian, MU Zhuangzhuang

2025, 53(9):  31-47.  doi:10.12141/j.issn.1000-565X.250003

Asbtract ( 229 )   HTML ( 0)   PDF (7533KB) ( 2 )  

Figures and Tables | References | Related Articles | Metrics

With the continual rise in the number of motor vehicles in urban areas, traffic congestion has become increasingly severe, adversely affecting environmental protection and urban operational efficiency. Consequently, it is of critical importance to accurately predict traffic congestion for traffic management and optimization. However, existing research still faces limitations in modeling the dynamic, time-varying characteristics of traffic flow and the complex interactions among road segments. To address these challenges, a gated spatiotemporal convolutional network model based on graph neural networks was proposed to more effectively capture and predict traffic congestion. Firstly, an improved K-means clustering algorithm was employed to divide the raw data into multiple congestion-state categories, which are then incorporated as auxiliary features to enhance feature representation. Next, a gated temporal convolutional network was introduced to capture the temporal properties and dynamic dependencies in traffic data, and a dynamic adaptive gated graph convolutional network was constructed to achieve feature fusion and dynamic weight allocation through a signal generation module and a dual-modulation mechanism, thereby facilitating effective extraction of spatiotemporal features. Finally, residual connections were incorporated to improve training stability, and skip connections were utilized to integrate multi-level and multi-scale features. Experimental results on real-world PeMS08 and PeMS04 datasets demonstrate that the proposed model achieves superior prediction accuracy compared with other baseline methods.

Design and Optimization of Small-Batch Matrix Multiplication Based on Matrix Core



LU Lu, ZHAO Rong, LIANG Zhihong, SUO Siliang

2025, 53(9):  48-58.  doi:10.12141/j.issn.1000-565X.240498

Asbtract ( 132 )   HTML ( 0)   PDF (3143KB) ( 2 )  

Figures and Tables | References | Related Articles | Metrics

General Matrix Multiplication (GEMM) is one of the most important operations in linear algebra, serving as the backbone for numerous applications in machine learning, scientific computing, and signal processing. In particular, FP16 batch GEMM has become a core operation in deep learning frameworks due to its efficiency in training and inference. However, current implementations on AMD GPUs (e.g., CDNA/MI200 architectures with Matrix Cores) suffer from suboptimal memory access and low compute utilization, limiting performance in high-throughput scenarios. Therefore, this paper proposed a GPU optimization scheme for half-precision batch GEMM (HGEMM). In terms of blocking strategy, it allocates equal memory access and computational loads to threads based on input matrix sizes, while enabling each thread to compute multiple matrix multiplications to improve arithmetic unit utilization. For memory access optimization, it trades redundant data reads for uniform memory access patterns per thread to facilitate compiler optimization, ensuring overlapping of memory and computation time. For extremely small-batch HGEMM with matrix dimensions smaller than 16, the proposed method employs a 4 × 4 × 4 Matrix Core and its corresponding tiling scheme to enhance memory performance while reducing computational resource wastage, and provides the option of whether to use shared memory to achieve the highest performance. This paper compares the performance of this scheme with two operators of rocBLAS on the AMD GPU MI210 platform. The results show that the ave-rage performance of this scheme on AMD GPU MI210 is 4.14 times that of rocBLASHGEMMBatched and 4.96 times that of rocBLASGEMMExBatched. For extremely small-batch HGEMM, the average performance is 18.60 times that of rocBLASHGEMMBatched and 14.02 times that of rocBLASGEMMExBatched.

Information Retrieval Re-Ranking Method Based on Bidirectional Text Expansion



TU Xinhui, GUO Cong, ZONG Yuhang

2025, 53(9):  59-67.  doi:10.12141/j.issn.1000-565X.240499

Asbtract ( 56 )   HTML ( 0)   PDF (1332KB) ( 1 )  

Figures and Tables | References | Related Articles | Metrics

With the rapid development of large language models (LLMs), remarkable progress has been made in both text matching and text expansion technologies in information retrieval. As two important methods for enhancing text representation, query expansion and document expansion have been widely applied in modern information retrieval systems. Currently, mainstream text expansion methods primarily rely on large language models. However, there are obvious differences in language diversity and style between the text generated by these models and the text created manually. These differences may affect the accuracy of calculating the query-document relevance, ultimately leading to a decline in the performance of the entire information retrieval system. To address this issue, this paper proposed an information retrieval re-ranking method based on bidirectional text expansion (BTE-IRRM). First, zero-shot prompting was used to enable the large language model to generate pseudo-queries for documents and pseudo-documents for queries. Then, the semantic similarity between these pseudo-queries and pseudo-documents was calculated. Finally, the similarity scores of the original query-document and the semantic similarity scores of the pseudo-query-pseudo-document were weighted and fused to obtain the final document ranking result. To validate the effectiveness of the proposed method, experiments were conducted on two public datasets (DL19 and DL20). Experimental results demonstrate that compared with the existing baseline methods, the BTE-IRRM method has achieved significant improvements in multiple evaluation indicators. Therefore, the bidirectional text expansion method proposed in this paper can further enhance the relevance matching between queries and documents, thereby improving the performance of the entire information retrieval system.

Mechanical Engineering

Research on Intelligent Ballistic Trajectory Simulation Football Auxiliary Training Robot



WEI Zhengjun, LIANG Zijian, ZHENG Kun, CHEN Liang

2025, 53(9):  68-75.  doi:10.12141/j.issn.1000-565X.240589

Asbtract ( 227 )   HTML ( 0)   PDF (2933KB) ( 0 )  

Figures and Tables | References | Related Articles | Metrics

With the increasing awareness of health and the popularization of competitive sports, the technological advancement and specialization of ball sports training have become a growing trend. In football training, the precise simulation of ball trajectories and the design of personalized training programs have emerged as key issues that need to be addressed urgently. To enhance the scientific rigor and effectiveness of football training and to promote its intelligent development, this study proposed an omnidirectional mobile intelligent ballistics trajectory simulation football training assistant robot by integrating technologies such as ball launching mechanisms, visual acquisition, data analysis, and motion control. Firstly, a forward dynamics model of football was constructed. Subsequently, considering complex physical factors such as air resistance and the Magnus force, this study designed an inverse kinematics solution model based on the RMSProp algorithm to solve the initial parameters for ball shooting, enabling precise adjustments of the yaw and pitch angles according to the target position, thereby achieving high-precision hits on the target point. Finally, a three-axis gimbal shooting robot capable of adjusting the shooting angle and position was developed and tested experimentally. Experimental results indicate that the training robot achieves a goal entry error of less than 0.45 m under various training conditions. The root mean square error between the theoretical and actual trajectories is less than 7.5 cm. These findings validate the robustness and precision of the previously described inverse kinematics solution model for ball launching. Additionally, this study established a detailed ball launching dataset, which can serve as an important resource for subsequent research in data science and artificial intelligence.

Visual and Haptic Interaction Algorithm for Orthopedic Virtual Surgery



WANG Qinghui, FANG Daoxin, CHI Zipeng, NI Jianlong, XIE Hailong, LI Jingrong, LI Chunhai

2025, 53(9):  76-85.  doi:10.12141/j.issn.1000-565X.250025

Asbtract ( 149 )   HTML ( 0)   PDF (5069KB) ( 1 )  

Figures and Tables | References | Related Articles | Metrics

Real-time visual and precise haptic interaction algorithms are critical for achieving accurate “tactile sensation” in virtual surgical training. In order to reduce storage space, improve computational efficiency, accurately calculate cutting forces during bone milling, and balance the visual and haptic interaction effect, this paper proposed a visual and haptic interaction algorithm based on the Tri-dexel model. Firstly, the Tri-dexel model was employed to represent the bone and the surgical milling tool. Real-time geometric deformation during the virtual bone milling was achieved through boolean operations and rapid surface reconstruction algorithms. Secondly, by integrating the geometric parameters of the surgical milling tool, a haptic interaction model based on the micro-element cutting force was proposed. This model utilizes the boolean operation results between the bone and surgical milling tool to quickly and accurately solve the instantaneous undeformed chip thickness. Thirdly, the cutting force coefficients were identified and the haptic interaction model was validated through milling experiments to achieve haptic rendering. Finally, an orthopedic virtual surgical training system was built based on the above-mentioned algorithms, and the interaction algorithm was tested and evaluated experimentally. The results show that the predicted forces align with experimental measurements, with an average force error of less than 7%. The visual and haptic interactive algorithm satisfies a visual refresh rate of 30 Hz and a haptic refresh rate of 1 kHz. The developed orthopedic virtual surgical training system provides users with a highly immersive virtual bone milling training experience that can effectively improve users’ hand-eye coordination.

Modeling and Verification of the Magnetic Wheel Adsorption Force in Multiple Working Conditions for Underwater Welding Robots



WANG Zhenmin, ZHU Bin, CHI Peng, LUO Bende

2025, 53(9):  86-97.  doi:10.12141/j.issn.1000-565X.240500

Asbtract ( 138 )   HTML ( 0)   PDF (6897KB) ( 1 )  

Figures and Tables | References | Related Articles | Metrics

To address the issue of adsorption instability encountered by magnetic wheel adsorption welding robots during underwater operation, this paper proposed a critical adsorption force calculation method for magnetic wheels based on centroid offset and vector superposition. This method comprehensively considers multiple failure modes, including traditional sliding failure, detachment failure, overturning failure, and the rarely studied skidding failure, effectively addressing the issue of adsorption instability caused by low accuracy in traditional adsorption force calculations. Firstly, based on the robot chassis structure, static models corresponding to four non-instability adsorption states were established, and a vector superposition method was proposed based on static coupling relationships. This method fully accounts for the influence of centroid offset on adsorption stability during actuator motion, providing a theoretical basis for the accurate calculation of the critical adsorption force of magnetic wheels. Then, a case study was conducted based on the permanent magnetic adsorption chassis of the existing underwater welding robots. The static analysis results were solved using Matlab and the variation law of the critical adsorption force of the chassis with maximum centroid offset at different spatial angles was summarized. Finally, an experimental setup was constructed to test the adsorption stability of the robot under various operational conditions. The experimental results demonstrate that the vector superposition method based on centroid offset can effectively improve the adsorption stability of underwater welding robots, providing novel theoretical support for the design and magnetic force optimization of subsequent magnetic adsorption chassis.

Analysis of Bidirectional Fluid-Structure Interaction Characteristics of Gas Bearing-Rotor Systems and the Establishment and Verification of Equivalent Gas Film Thickness



MA Wenqi, MA Hailong, QIN Yubin, HUANG Dali

2025, 53(9):  98-105.  doi:10.12141/j.issn.1000-565X.240305

Asbtract ( 91 )   HTML ( 0)   PDF (2511KB) ( 1 )  

Figures and Tables | References | Related Articles | Metrics

When simulating the flow field characteristics of the gas bearing-rotor system using the computational fluid dynamics (CFD) method, the gas film thickness is one of the crucial structural parameters. However, shape and dimensional errors arising during component machining, as well as the deviations caused by the system assembly, can lead to certain discrepancies between the actual gas film and the ideal designed gas film in terms of the spatial morphology and scale. This further affects the reliability and accuracy of the numerical calculation results. Therefore, this paper first proposed the concept of effective gas film thickness for the flow field. Through the comparative analysis and correction of the bidirectional fluid-structure interaction numerical simulation and experimental results, the reasonable equivalent gas film thickness was finally determined. The research results show that the adoption of the bidirectional fluid-structure interaction numerical simulation method can reveal the transient characteristics of the gas film flow field and the variation law of the rotor attitude, and predict and evaluate whether the gas bearing-rotor system can operate safely, saving the cost of experimental testing. The rotor inclination angle was adopted as the comparative analysis feature, providing an intuitive reference basis for the system performance deviation analysis between numerical simulations and experimental tests. The established equivalent gas film thickness maximally simplifies the numerical simulation model while enhancing computational efficiency, and simultaneously maintains reasonable result reliability. Taking a supply pressure of 0.6 MPa and a unilateral steady-state force of 80 N as an example, through error analysis and approximation, the estimated equivalent gas film thickness in the fluid-structure interaction simulation model was cyclically established and corrected. Eventually, the relative error of the system inclination angle was controlled within 5%, which greatly improved the consistency between the numerical simulation results and the performance of the actual engineering system. Furthermore, this approach provides a reliable method and basis for the application of the gas bearing-rotor simulation system in structural design, performance prediction and evaluation.

Active Tooth Surface Design and Performance Optimization of High Reduction Ratio Hypoid Gears



JI Shuting, LI Jiahao, ZHANG Yueming

2025, 53(9):  106-116.  doi:10.12141/j.issn.1000-565X.240586

Asbtract ( 202 )   HTML ( 0)   PDF (4605KB) ( 0 )  

Figures and Tables | References | Related Articles | Metrics

To enhance the comprehensive transmission performance of hypoid gears with high reduction ratios, this paper proposed a design method for significantly inclined contact lines based on active tooth surface design techno-logy. Firstly, multiple tooth surface imprints with varying degrees of contact line inclination were preset, with specified values for the semi-major axis of the contact ellipse and the length of the contact trace. The pinion conjugate tooth surface was then modified with a parabolic shape to achieve a tooth surface that meets the preset parameters. Subsequently, by integrating Tooth Contact Analysis (TCA) and Load Tooth Contact Analysis (LTCA) techniques, the amplitude of transmission error (ATE), amplitude of loaded transmission error (ALTE), tooth surface load distribution, root bending stress amplitude, and tooth surface flash temperature amplitude were obtained for each tooth surface. The influence of variations in contact trace length on these performance parameters was then analyzed. Finally, a target modified tooth surface was selected, and its comprehensive performance was analyzed and compared with that of the original tooth surface. A case study demonstrates that for a hypoid gear pair with a gear ratio of 5∶75, under conditions of highly inclined contact trace on the tooth surface, a longer contact trace length leads to lower contact stress, as well as reduced root bending stress and flash temperature on the tooth surface. The target tooth surface exhibits weakened edge contact, a 12.0% reduction in maximum root bending stress, more uniform contact stress distribution, and a 6.3% decrease in peak flash temperature. As a result, the scuffing load-carrying capacity is enhanced. Overall, the modified target tooth surface exhibits superior contact performance, better load-carrying capacity, and significantly enhanced comprehensive transmission performance.

Energy，Power & Electrical Engineering

Multi-Working Condition Energy Consumption Anomaly Detection Method for Office Building Lighting Sockets Based on LSTM-AE



CHEN Cheng, WANG Miao, WANG Xinyao, GAO Zhiming, ZHOU Xuan, YAN Junwei

2025, 53(9):  117-126.  doi:10.12141/j.issn.1000-565X.240575

Asbtract ( 113 )   HTML ( 0)   PDF (3612KB) ( 2 )  

Figures and Tables | References | Related Articles | Metrics

Anomaly detection of energy consumption in building lighting and socket systems can effectively improve energy efficiency. It holds significant importance for the implementation of building energy optimization measures and the realization of energy-saving management and control. Since the energy consumption of building lighting and plug load systems is heavily influenced by the random behavior of building occupants, and given the challenges posed by noisy time-series data and difficulty in feature extraction, this study proposed an unsupervised anomaly detection method that integrates operating condition classification with deep learning, aiming to enhance the accuracy and robustness of energy consumption anomaly identification. First, the decision tree algorithm was employed to classify the energy data based on attributes such as working days vs. non-working days and working hours vs. non-working hours. Then, for each identified condition, a long short-term memory autoencoder (LSTM-AE) model was constructed to detect anomalies. This model learns to reconstruct normal data and calculates the reconstruction error. By setting differentiated thresholds, it enables energy consumption anomaly detection under unlabeled data conditions. Using 578 days of hourly lighting and socket energy consumption data from an office building located in a hot-summer and warm-winter region, the study conducted model training and hyperparameter optimization experiments. Results indicate that the number of iterations, the number of neurons, and the activation function have significant effects on the model’s performance. Energy data during working days demonstrate greater stability than those on non-working days, resulting in higher detection accuracy. The proposed method achieves average precision, recall, and F₁ of 91.23%, 90.87%, and 90.80%, respectively, across four typical operating conditions, demonstrating its effectiveness in detecting energy anomalies in building lighting and socket systems.

Auxiliary Frequency Regulation Control Strategy of District Cooling System Based on Model Predictive Control with Terminal Constraints



LIU Mingbo, LAO Ziqing, DONG Ping

2025, 53(9):  127-137.  doi:10.12141/j.issn.1000-565X.250017

Asbtract ( 101 )   HTML ( 0)   PDF (3340KB) ( 0 )  

Figures and Tables | References | Related Articles | Metrics

The district cooling system (DCS) belongs to a class of centralized air-conditioning loads and has frequency regulation potential. This paper proposed an auxiliary frequency regulation control strategy of DCS based on model predictive control (MPC) with terminal constraints, which controls the power consumption of the DCS by adjusting the chilled water flow rate and the number of chiller shutdowns. Firstly, the study established a dynamic model of DCS and traditional units considering the relationship between chilled water flow rate and chilled water outlet temperature, and constructed the state space expression of the system. Then, based on MPC with terminal constraints, it established a joint frequency regulation control model for DCSs and traditional units, with the objective function of minimizing frequency deviation, building temperature deviation from human comfort temperature, chilled water flow’s control instructions, and traditional unit’s control instructions. The terminal constraints include terminal cost function and terminal set. Moreover, it was proved that the MPC problem with terminal constraints is asymptotically stable by constructing the Lyapunov function of the system. Finally, simulations on a 10-unit 39-bus system and an actual power system were carried out. The results verify that adding terminal constraints can improve system stability, and the use of DCS to assist in grid frequency regulation can help the system to quickly restore the rated frequency and improve regulation performance. In addition, the participation of DCSs in grid frequency regulation have no significant impact on comfort.

Traceable Flue Gas Flow Rate Measurement Method for Large-Diameter Stacks of Thermal Power Plant



LU Zhimin, XIE Zili, LU Weiye, CHEN Xiaoxuan, HUANG Yongru, LIU Zeming, TIAN Xuejun, YAO Shunchun

2025, 53(9):  138-148.  doi:10.12141/j.issn.1000-565X.240571

Asbtract ( 144 )   HTML ( 0)   PDF (3796KB) ( 0 )  

Figures and Tables | References | Related Articles | Metrics

The tracer gas dilution method can address the issue of significant measurement errors in flue gas flow caused by the complex flow field in large-diameter stacks of power plants. The method is traceable and operates on a measurement principle different from the conventional velocity-area method, making it a promising candidate for on-site calibration of flow measurements. To this end, this paper employs numerical simulation to analyze the feasibility and accuracy of the tracer gas dilution method for measuring flue gas flow in power plant stacks. On this basis, it studies the influence of the tracer gas dilution ratio and injection cross-section on measurement results. In addition, different tracer gas sampling schemes were designed to evaluate the stability of the measurements. The results demonstrate that, at a height of approximately 9D (where D is the stack diameter), the tracer gas achieves full mixing with the flue gas; both excessively high and low tracer gas dilution ratios can negatively affect the mixing efficiency; injecting the tracer gas at the flue section can effectively reduce flow measurement errors. Under 80% load rate, when the tracer gas is injected into the stack, the measurement errors vary considerably across different sampling schemes. However, the three-point sampling method demonstrates a stable and accurate performance, with measurement errors of only -3.59%, -0.69%, and -1.05% at the 3D, 8D, and 12D cross-sections, respectively. When the tracer gas is injected into the horizontal flue, the flow measurement errors for all sampling schemes remain within ±10%. Specifically, with three-point sampling, the errors at the 3D, 8D, and 12D cross-sections are 0.98%, -0.52%, and 0.21%, respectively—all within ±1%. These results demonstrate the feasibility and accuracy of the tracer gas dilution method for flue gas flow measurement in large-diameter stacks.

Experimental Study on Thermal Performance of Phase Change Heat Transfer Module with Roll Bond Aluminum Vapor Chambers



GAN Yunhua, XIE Yuheng, LIU Fengming, LIAO Yuepeng, LI Yong

2025, 53(9):  149-162.  doi:10.12141/j.issn.1000-565X.240534

Asbtract ( 107 )   HTML ( 0)   PDF (7071KB) ( 6 )  

Figures and Tables | References | Related Articles | Metrics

For the thermal management of high power consumption modules in 5G communication base stations, this study proposed a phase change heat transfer module with roll bond aluminum vapor chamber, in which the evaporation chamber of the module is interconnected with the flow channels of all vapor chambers. By constructing a performance testing platform, experimental studies were conducted to investigate the heat transfer performance of the module under different filling ratios. The impacts of the boiling state and the flow distribution of the working fluid on both temperature uniformity and heat dissipation efficiency of the module were analyzed. Additionally, the variation in surface temperature distribution of the heat source under different lateral inclination angles was also explored. The research results indicate that under the condition of an input power not exceeding 400 W, as the filling ratio increases, the total thermal resistance of the phase change heat transfer module initially decreases and then increases, reaching its minimum at a filling ratio of 15.0% with the lowest total thermal resistance being 0.211 6 ℃/W. Appropriately reducing the filling ratio induces boiling of the liquid working fluid at the bottom of the vapor chambers, thereby promoting the balanced distribution of gaseous working fluid among different vapor chambers and enhancing both the heat dissipation efficiency and temperature uniformity of the phase change heat transfer module. At input powers of 350 W and 400 W respectively, reducing the filling ratio from 30.0% to 15.0% leads to a decrease in the standard deviation of temperatures among the vapor chambers by 40.92% and 34.04%, resulting in a significant improvement in temperature uniformity. When the tilt angle of the phase change heat transfer module changes, the liquid level in the evaporation chamber shifts, leading to uneven temperature distribution on the heat source surface. This adverse effect becomes more pronounced with increasing inclination. At a tilt angle of 10.0° (under the same power conditions), the maximum temperature difference on the heat source surface increases to more than 11.7 times that under horizontal placement.