In order to solve the problem that traditional Top-k pattern mining results failed to meet the users' practical needs, a graph data Top-k frequent pattern mining approach that integrates subjective and objective evaluations was proposed. A pattern representation technique based on minimum DFS coding was introduced to encode patterns. The graph patterns evaluation model(GPEM)was built based on a siamese neural network, which learned the biased order relationships between pattern pairs and predicted subjective preference of patterns. A pattern interestingness evaluation function that combined subjective and objective factors was designed to guide Top-k pattern mining. Experiments on six real graph datasets demonstrated that GPEM outperformed other models on various metrics, with up to 93% accuracy.

To enhance road safety and prevent traffic accidents, a distracted driving behavior detection algorithm based on local and global knowledge distillation(LGD)was proposed. Built upon a knowledge distillation framework, the method introduced a distillation loss function that integrated both local and global features, guiding the student network to effectively learn the discriminative capabilities of the teacher model. While maintaining a lightweight network structure, the approach significantly improved the recognition accuracy of distracted driving behaviors. By effectively facilitating the learning process of the student network, the method enabled it to achieve detection accuracy comparable to that of the teacher model, despite having fewer parameters. Experimental results demonstrated that the proposed method achieved an accuracy of 91.79% with only 31.85 M parameters, highlighting its effectiveness in addressing the problem of distracted driving detection.

To solve the problem of the curse of dimensionality in multi-label learning and the problem that filter feature selection methods were prone to fall into local optima, a fast multi-label feature selection method based on global redundancy minimization was proposed. Candidate labels and candidate feature subsets were selected from the original label space and feature space through K-means clustering and mutual information calculation; the local optima problem was solved by minimizing global redundancy, and the feature weight with minimum feature redundancy was obtained to ensure outputting the best feature subset; an ensemble learning strategy was used to enhance the stability of feature selection. The experimental results on 14 multi-label datasets showed that the proposed method had better performance than other methods in all classification indicators.

In order to effectively solve the entity attribution and entity nesting in the academic domain of universities, a Chinese named entity recognition model Multi-feature BiGRU-MHSA-CRF(MBMC)was proposed based on multi-feature fusion. The text semantic features of the model were represented from three aspects of character, word and position to enrich the multi-dimensional semantic features of academic text. The feature vectors were fed into BiGRU(Bi-directional Recurrent Neural Network)to capture global semantic features. In order to solve the problem of entity boundary delimitation in the academic domain of higher education, the attention mechanism was improved by introducing a multi-head self-attention mechanism with Q, K, and V weight matrices and increasing the learning parameters to improve the recognition accuracy. All possible label sequences were output to the CRF, and the entity label sequence was generated by CRF decoding. The experimental results showned that the F1 value of the model reached 89.57% and 86.14% in the public dataset CoNLL2003 and the college academic domain dataset, respectively. It was better than other traditional models.

In view of the inefficiency of non-incremental attribute reduction in interval-valued decision-making information systems, the concept of knowledge granularity was extended to interval-valued decision-making information systems, and incremental attribute reduction in interval-valued decision-making information systems was systematically investigated from the viewpoint of knowledge granularity. The concept of tolerance degree was introduced into interval-valued decision-making information systems and the measurement method of interval-valued tolerance degree was improved; the tolerance relation was determined according to the tolerance degree and the corresponding tolerance relation matrix was constructed. This led to an calculation method of the knowledge granularity in the interval-valued decision-making information system based on the matrix; The updating mechanism of knowledge granularity under the condition of change in the object set was explored, based on which the attribute importance was represented by knowledge granularity, and an incremental attribute reduction algorithm was constructed with the attribute importance as the heuristic information. The experiments of the incremental attribute reduction algorithm were implemented on 6 selected UCI datasets. The experimental results demonstrated that the incremental attribute reduction method consumed less time than the non-incremental method without compromising the accuracy of the reduction results, indicating that the incremental method was more efficient.

To investigate the influence of pore density and grooved dimension on the operating mechanism of metallic nickel foam evaporators at the microscale, flow boiling heat transfer experiments were conducted using twelve samples with three pore densities(100, 500, and 1 000 mesh)and four grooved dimensions(0, 0.4, 0.7, and 1.0 mm). The heat flux density, convective heat transfer coefficient and pressure drop were calculated and analysed. The results showed that, for samples with the same pore density, the grooved dimension of 0.7 mm yielded the highest convective heat transfer coefficient. Under identical grooved dimension, the optimal performance was achieved by the sample with the pore density of 500 mesh and the groove dimension of 0.7 mm, which convective heat transfer coefficient increased by 3.02 times compared with the 100 mesh sample, reaching a maximum critical heat flux of 152.40 W/cm² and a peak convective heat transfer coefficient of 27 630.50 W/(m²·K). Surface modification of metallic nickel foam with silica sol was further carried out to examine the effect of wettability on heat transfer. The results indicated that the combination of grooving and hydrophilic modification did not exhibit a synergistic effect. Without grooves, hydrophilic modification enhanced heat transfer performance. However, it also significantly weakened the improvement brought by grooving, resulting in lower performance compared with untreated metallic nickel foam samples.

Cryogenic loop heat pipes(CLHPs), as efficient two-phase heat transfer systems for the cryogenic temperature range, have been widely used in many engineering applications both on the ground and in space. The operational stability of CLHPs directly influenced the performance of the thermal control system. In this study, a system-level transient model was employed to conduct transient simulations on the operational stability of a neon-charged CLHP(Ne-CLHP). The phenomena and causes of pressure and operating temperature oscillations in the Ne-CLHP were simulated and investigated. The gas-liquid distribution characteristics inside the key components of the system during the instability process were analyzed in detail. Combined with the temperature-pressure drop phase diagram, the instability mechanism was examined, which revealed four distinct operating states—stable operation, oscillation attenuation, oscillation amplification, and periodic oscillation—that occurred under different operating conditions of the CLHP.

To enhance the comprehensive performance of the heat pipe heat exchanger, the finned tube structure in a thermosyphon evaporator for nuclear power plant passive cooling systems was optimized using the hybrid back propagation(BP)neural network prediction model and non-dominated sorting genetic algorithm Ⅱ(NSGA-Ⅱ). Six structural parameters, such as fin thickness, spacing, height, transverse tube pitch, longitudinal tube pitch, and aspect ratio, were selected as design variables. Predictive models for the Nusselt number, pressure drop, and minimum cross-sectional airflow velocity were developed, enabling multi-objective optimization targeting maximum heat transfer and minimum flow resistance. The optimal configuration(1 mm fin thickness, 6 mm spacing, 5 mm height, 70 mm transverse tube pitch, 75 mm longitudinal tube pitch, 1.4 aspect ratio)resulted in a 25.86% increase in the heat factor, a 17.96% reduction in the flow resistance, and a 35.24% improvement in the overall performance coefficient. These results validated the effectiveness of the combined BP neural network and genetic algorithm for heat pipe design, providing both critical parameters and theoretical guidance for engineering optimization of nuclear power plant passive cooling systems.

This study divided the ice making process into four stages based on the heat transfer form of the ice-making unit of the large direct-cooled ice maker and the change of internal phase state: the convective heat transfer cooling process of water, the conductive heat transfer cooling process of water, the freezing process of water, and the cooling process of ice. By using energy conservation equations and heat transfer equations, mathematical models of each unit in the whole ice-making process were established, and the time of each stage was solved by MATLAB software programming. In this study, the icing process in the ice making unit of a large direct-cooled ice maker was experimentally determined. It had been verified that the calculation errors of each stage of the heat transfer model were all within 8.00%, and the total error was less than 4.00%, which indicated the high accuracy of the model. This study provided technical guidance and theoretical support for the design optimization of a large direct-cooled ice maker.

To overcome the limitations of loop heat pipes under multi-heat-sources conditions and enhance the circulation and heat exchange efficiency of the working fluid, a novel loop heat pipe was designed and fabricated based on the traditional structure to provide cooling for multi-heat-sources simultaneously. This loop heat pipe featured one evaporator and multiple heat sinks. By circulating the working fluid internally, it absorbed heat from the evaporator and heat sinks and released it at the condenser, achieving simultaneous cooling of multi-heat-sources. Through the optimization of the traditional loop heat pipe design, the novel structure was manufactured, and the effect of varying the number of heat sinks on its operational performance was further investigated. Experimental tests verified the feasibility of the new loop heat pipe, showing that the system generated a maximum circulation flow rate of 45 mL/min. Compared with the pre-optimized design, the overall heat dissipation capacity of the system increased by up to 90%. A comparison of the startup and stable operation performance parameters of different structural configurations showed that changes in the number of heat sinks affected the circulation resistance of the working fluid inside the system, thereby influencing the circulation flow rate and ultimately determining the cooling capacity of the evaporator and heat sinks.

To address the thermal path optimization for high heat dissipation and high heat flux concentration devices on scanning-mode spacecraft rotating platforms, a centrifugally-enhanced involute heat pipe(CEIhp)was designed. By optimizing the heat pipe configuration, the impact of platform rapid rotation on the startup of the heat pipe was reduced, and the rotational acceleration was leveraged to enhance its maximum heat transfer capacity. A centrifugal test platform was established to validate both thermal transfer and startup performance of the CEIhp. Experimental results demonstrated that the CEIhp could successfully initiate operation across angular velocity of 0-2.512 rad·s^-1. The heat transfer capacity of CEIhp increased from 170 W·m to 421 W·m as the angular velocity rose, with the startup power requirement escalating to 6.6 W. This achievement fulfilled the application requirements for efficient heat transfer on scanning-mode spacecraft rotating platforms and enabled full-range efficient thermal dissipation for compact rotating platforms.

Aiming at the problems of resource waste, environmental pollution caused by the ‘temporary’ nature of existing deep foundation pit support, the difficulty in controlling the flatness of the traditional circular support pile wall, and the obstruction of promoting permanent support, this study proposed an upper rectangular, medium-expanded, and lower circular variable section(RECV)pile as a permanent support structure for deep foundation pits. This study combined Rankine's earth pressure theory and elastic foundation beam method to derive the calculation formulas of bending moment, shear force and stress of the whole pile of RECV pile. The finite element method was used to establish the numerical model, and the reliability of the model was verified based on the engineering monitoring data. The horizontal displacement and bending moment characteristics of RECV pile and traditional circular pile as foundation pit supporting pile were simulated and compared. The numerical simulation results showed that under the condition of bolt support, the horizontal displacement of RECV pile was significantly smaller than that of traditional circular pile, which showed stronger anti-deformation ability and unit volume concrete support efficiency advantage. The theoretical calculation results were basically consistent with the simulation results, which could provide reference for the internal force calculation of RECV piles, lay a foundation for the technical theory and design method of RECV piles, and promote the development of permanent support technology.

In order to study the influence of steel slag fine aggregate volume fraction on the softening characteristics of concrete, the steel slag fine aggregate with stability meeting the specification limit requirements was selected to prepare concrete for wedge splitting test, and the inverse analysis program was established by ISIGHT integrated MATLAB and ABAQUS, and the trilinear softening curve of steel slag fine aggregate concrete was deduced, and the accuracy of finite element analysis of the softening curve was verified based on the wedge splitting tensile test results. The results showed that steel slag could significantly improve the tensile strength of concrete and accelerate the stress reduction of concrete after cracking. When the steel slag volume fraction was 20%, the slope of the second segment of the trilinear softening curve decreased the most. The cracking displacement of concrete at the end of the softening curve decreased with the increase of steel slag volume fraction, and the ductility of concrete deteriorated. The softening curve of steel slag fine aggregate concrete provided a theoretical basis for studying its softening performance.

In order to investigate the plastic hinge performance of precast segmental piers with two different connection methods based on the "cast-in-place" principle, a total of 66 concrete piers with rectangular or square cross-sections at home and abroad were collected and assembled based on the experimental studies of four precast piers. The effects of pier height, section height in the loading direction, material properties, reinforcement ratio, and axial compression ratio on the plastic hinge length were analyzed and the existing plastic hinge length formulae were evaluated. The formula of plastic hinge length suitable for precast segmental assembled piers was proposed, and its feasibility was verified by analytical method. The results indicated that the plastic hinge length increases with the increase of the pier height, section height, yield strength, and diameter of longitudinal bars and reinforcement ratio, and decreases with the increase of the axial compression ratio; and the formula proposed in this paper was suitable for the calculation of the plastic hinge length of precast segmental assembled pier.

In order to solve the engineering problems such as insufficient lateral restraining capacity of the soil body and the susceptibility to spreading and bulging damage at the top of the pile when the rubble type bulk piles were reinforcing the soft ground foundation, the composite foundation technology of waste tyre grillage ring-restrained piles was proposed. By considering the changes of reinforcement type, pile diameter and bedding form, single-pile and composite foundation bearing tests were carried out, and the load-displacement curves were tested; based on different damage modes, the calculation method of single-pile bearing capacity of bulk piles was proposed and verified for accuracy. The results showed that: compared with ordinary bulk piles that reach the ultimate damage state under small displacement conditions, the ultimate bearing capacity of ring-restrained piles was obviously improved; compared with the geogrid ring-restrained piles, the load-displacement curves of tyre grid-restrained piles did not show any obvious damage points, which indicated that their radial restraining capacity was stronger; comparing with the load-displacement curves of ring-restrained piles under different pile radius conditions, it indicated that increasing the pile diameter was very effective in improving the bearing capacity of the piles. Comparing the load-displacement curves of piles with different pile diameters, it was indicated that increasing the pile diameter appropriately had a significant effect on increasing the pile bearing capacity, and the setting of gravel bedding further improved the pile bearing capacity. The study results had important engineering application value for the promotion of this type of pile foundation and the improvement of recycling of solid waste resources.

The optimal mass fraction of fly ash was determined based on the principles of maximum strength and appropriate stiffness, with cement and recycled brick aggregate mass fractions and curing age set as control parameters. Mechanical properties of 15 recycled brick mix water stabilizing materials(recycled brick mix water stabilizing material, RBSM)under different mix designs were investigated. The comprehensive performance of RBSM was evaluated using the entropy-weighted TOPSIS(technique for order preference by similarity to ideal solution)model, yielding optimal values for each parameter. Results indicate,(1)Early-age splitting tensile strength and unconfined compressive strength first increase then decrease with rising fly ash content, peaking at 10% mass fraction. Compressive rebound modulus increases with fly ash content, confirming 10% as the optimal fly ash content in cement.(2)The optimum moisture content of RBSM gradually increases with the mass fraction of recycled brick aggregate, while the maximum dry density decreases accordingly.(3)During the early curing stage, the splitting strength of RBSM gradually decreases with increasing recycled brick aggregate mass fraction, while the unconfined compressive strength first increases and then decreases, peaking at a mass fraction of 50%. As the curing age increases, the mechanical strength of RBSM improves.(4)Through entropy-weighted TOPSIS theory, optimal mass fractions for cement and recycled brick aggregate were determined. The final recommended optimal mix design for recycled brick-based water-stabilized materials was 10% fly ash mass fraction in cement, 7% cement mass fraction in recycled brick-based water-stabilized material, and 50% recycled brick aggregate mass fraction in recycled brick aggregate.