Abstract:【Object】In order to explore the vibration characteristics of high-speed train gearbox housing under the coupling excitation of wheel polygon and rail corrugation.【Method】The rigid-flexible coupling dynamic model of wheelset, gearbox housing and track was established. Three vibration acceleration sensors were arranged in the gearbox housing, and dynamic simulation was carried out under different working conditions to analyze the vibration acceleration of each measuring point of gearbox housing.【Result】Under the wheel-rail coupling excitation, at the same speed, when the wheel polygon is 23 order and the amplitude is 0.01 mm, the root mean square value of the vibration acceleration of each measuring point is the largest, and the wheel polygon has the greatest influence on the measuring point B in the three measuring points. Under the excitation of rail corrugation, the root mean square value of vibration acceleration of measuring point B is the largest among the three measuring points under the combined action of harmonic torque of traction motor and gear meshing. The root mean square value of vibration acceleration of measuring points A and C increases with the increase of amplitude. Compared with the wheel-rail coupling excitation with broad spectrum, the vibration frequency caused by wheel-rail excitation is close to the fifth order natural frequency of the gearbox housing, which induces resonance.【Conclusion】The resonance can be avoided by changing the speed of the train or changing the structure of the gearbox. This study provides a reference for the design of the gearbox structure.

Abstract:【Objective】To investigate the impact of surface surcharge on the settlement of adjacent tunnels, 【Method】a two-stage analysis method was adopted. In the first stage, the Boussinesq solution calculates the vertical additional force exerted on nearby subway tunnels due to surface surcharge. In the second stage, the subway tunnel is modeled as an Euler beam supported on the Pasternak foundation, simplifying the structural representation. Utilizing the Rayleigh Ritz method, functional expressions for each system component are formulated, leading to the establishment of the total energy equation. The control equation is subsequently solved through the application of the variational principle. 【Result】The obtained results are compared to measured data to validate the effectiveness of the proposed method. The research explores the impact of offset distance (d), tunnel burial depth (z), surcharge (p), and surcharge range (B and L) on settlement. 【Conclusion】The research findings indicate a significant impact on settlement within the range of (0.5-1) L directly below and on both sides of the surcharge; As d increases, the maximum settlement of the tunnel gradually decreases; There is a linear relationship between p and wmax and the larger the pile surcharge, the greater the maximum settlement; Increasing L and B initially leads to gradual settlement increase, followed by stabilization. L exerts a greater influence on wmax compared to B.

Abstract:Remote sensing image object detection has a wide range of applications in intelligent transport, such as dynamic monitoring of road network operation status, intelligent law enforcement on roads, and intelligent monitoring of road disasters. Due to the characteristics of small and dense targets, large scale changes, and arbitrary direction distribution in remote sensing images, general object detectors have poor detection performance when directly applied to remote sensing images. To address these challenges,this paper proposes a remote sensing image object detection algorithm based on improved Retinanet. First, the model introduces Improved Downsampling Module (IDM) on the base feature extraction network ResNet50, which performs multiple down-sampling processing on features, and then dynamically selects the spatial receptive field using the convolution kernel selection mechanism to model the multi-scale semantic information of the scene. Finally, the classification and regression results of the target object are obtained. Experimental results show that the proposed method improves the mAP by 3.2% on the large-scale remote sensing image object detection dataset DOTA compared to the original Retinanet network, enabling more accurate localization and identification of remote sensing targets.

Abstract:【Objective】The Clar covering polynomial of molecular graphs is a method to characterize the electronic structure of conjugated systems. By studying the Clar covering polynomials of plane bipartite graphs, the resonance theory of related molecular graphs and their related properties can be well studied.【Methods】Based on the theorem related to Clar covering polynomials of plane bipartite graphs, the method of generating functions is utilized to compute Clar covering polynomials of plane bipartite graphs.【Results】Recurrence relations for Clar covering polynomials of a special class of graphs are derived. In turn, explicit expressions for the Clar covering polynomials of two classes of catacondensed plane bipartite graphs are computed using the generating function method.【Conclusion】On the Clar covering polynomials of plane bipartite graphs, it is possible to understand the electronic structure of chemical molecules, predict their chemical properties and reaction behavior, and design new molecular structures.

Abstract:【Objective】Fully exploiting the spatial correlation of passenger flow between related stations in the subway network has a positive effect on the improvement of subway passenger flow prediction accuracy. Capturing and quantifying spatial patterns in passenger flow data is difficult due to the difficulty of learning and transferring spatial correlations between metro stations. 【Method】An improved graph-convolution gated recurrent neural network metro passenger flow prediction model is proposed to enhance the model"s ability to handle different data types by integrating multivariate spatio-temporal data. A spider wasp optimisation algorithm based on Tent chaotic mapping and Levi"s flight perturbation strategy is used to dynamically adjust the model structural parameters in order to optimise the hidden layer structure of the gated recurrent neural network.【Result】Prediction results on weekdays and weekends show that 20 iterations lead to optimal results, with higher prediction accuracy with larger training samples. 【Conclusion】Dynamic optimisation of the hidden structure of gated recurrent networks can lead to better convergence of the prediction model and higher prediction accuracy.

Abstract:【Objective】To grasp the deformation characteristics of Jiangxi red clay under the action of straight shear plus unloading and the law of energy dissipation. 【Methods】Direct shear tests under shear loading and unloading conditions were carried out on Jiangxi red clay, and its strength-deformation characteristics and energy dissipation laws were analyzed. 【Results】The results show that: 1) Overall, the shear elastic displacement increases with the increase of normal stress. The shear elastic displacement at different unloading positions under the same unloading condition increases with the increase of unloading times. In the case of low normal stress, the slope of the red clay hysteresis curve with the increase of the number of unloading shows a nonlinear decay change, and the decay rate under the low normal stress condition is from slow to fast; 2) the red clay in the unloading and loading of the straight shear test process as a whole shows a shear shrinkage deformation, the shear unloading stage of the normal displacement increases, and then loading stage of the normal displacement first increases and then reduces the deformation, resulting in shear expansion deformation, and shear expansion is over when shear displacement reaches the value of unloading before. When the shear displacement reaches the value before unloading, the shear expansion ends. Shear unloading will increase the normal displacement of red clay, increase the shear shrinkage, unloading effect on the normal displacement with the increase of normal stress gradually decreases; 3) the same loading and unloading conditions of red clay shear strength with the increase of normal stress increases. The effect of loading and unloading on the shear strength of red clay is larger, and the effect of shear unloading under high normal stress is more obvious than that under low normal stress. 4) The dissipation energy of different unloading times under the same unloading condition increases with the increase of normal stress and the number of unloading and loading times, and the normal stress is between 100kPa and 300kPa, the dissipation energy of the unloading two conditions is larger than that of the unloading one condition. The dissipation energy of unloading twice is larger than that of unloading once at 100kPa to 300kPa, while that of unloading three times and four times is smaller than that of unloading twice, and that of unloading more times is larger than that of unloading once at 400kPa normal stress. 【Conclusion】: The dissipation energy and deformation of red clay under different loading and unloading conditions have certain regularity.

Abstract:【Objective】Aiming at the characteristics of various target scales, complex background and dense small targets in aerial images of unmanned aerial vehicles,a small target detection algorithm LM-YOLO based on YOLOV5 is proposed.【Method】Firstly,increase the small target detection head and K-DBSCAN clustering algorithm is used to optimize the anchor frame ,so as to generate an anchor frame more suitable for small target detection and improve the detection accuracy of the algorithm.Then,a more efficient MobileNetV3-CBAM is designed as a feature extraction network to reduce the size of the network model.Finally,the large kernel selective attention mechanism LSK is introduced into the feature fusion network to increase the resolution of the model to similar targets.【Result】The experimental results on the public data set VisDrone2019 show that the average detection accuracy of LM-YOLO for all targets is improved by 7.6% and the model size is reduced by 45% compared with the benchmark model YOLOV5.【Conclusion】Experiments show that the proposed algorithm can reduce the model size while maintaining good detection accuracy, and is suitable for the target detection task of aerial images.

Abstract:To address the challenge of identifying traveling waves with different components, including fault point incident waves, reflected waves, and waves reflected by AT, which makes fault location on all parallel AT traction networks difficult, this paper proposes a wave similarity-based fault location method using Adaptive Projection Intrinsically Transformed Multivariate Empirical Mode Decomposition (APIT-MEMD). The APIT-MEMD algorithm is applied to adaptively decompose the fault signals of multi-conductor lines in both directions to extract transient high-frequency characteristics representing different components of fault traveling waves. By constructing the cross-correlation function matrix of different wave mode components to identify traveling waves along different paths and calculating the corresponding maximum time delay of the cross-correlation function, the fault location of the traction network"s traveling waves is achieved. Experimental results demonstrate that the proposed method, based on time-frequency mode feature extraction, achieves fault location accuracy within 102 m with an average absolute error of 49 m.Compared with the results of different projection parameters in the multi-dimensional empirical mode decomposition algorithm, the effectiveness of fault location accuracy is optimized. The optimized fault location algorithm meets the requirements for high-precision fault location.

Abstract:【AIMS】 In view of the lack of comprehensive evaluation system for high-speed train wheel and rail optimization scheme, this study aims to develop an evaluation method based on improved fuzzy hierarchy analysis method to determine the most appropriate equivalent conicity scheme and ensure the smooth operation of the train.【Methods】Firstly, in the light of the main kinetic factors affecting the selection of equivalent conicity, 14 evaluation indexes were selected to construct the index decomposition model, and the improved three-scale hierarchy analysis method was applied to determine the weight of each index. Subsequently, three different working conditions with equivalent conicity were established in the dynamics software Simpack, and the underlying index values were calculated and dimensionless to determine the fuzzy evaluation membership matrix, and the fuzzy theory system evaluation model was constructed to obtain the comprehensive evaluation results of the three schemes.【Results】The comprehensive evaluation results show that the scores of the three equivalent conicity schemes are: M1:26.2%, M2:17.9%, and M3:16.2%, from which M1 was determined as the optimal choice. By calculating the kinetic simulation of the two groups and observing the intuitive simulation results of the post-processing module, the results are consistent with the conclusion of the comprehensive evaluation system, and then the reliability of the evaluation method is verified.【Conclusion】The method of equivalent conicity evaluation of high-speed train wheel and rail based on improved fuzzy hierarchy analysis proposed in this study is effective, which can provide scientific basis for the optimization of equivalent conicity scheme of high-speed trains and ensure the stability and safety of trains.

Abstract:Density Peak Clustering (DPC) has been widely used as an efficient and non-iterative clustering algorithm. However, studies have found that DPC struggles to select correct cluster centers, especially in datasets with non-spherical clusters and non-uniform density. Moreover, the algorithm is heavily influenced by the truncation distance parameter. 【Objective】In order to address the issue of poor performance of DPC on datasets with uneven density distributions, 【Method】we propose a density peak clustering algorithm based on natural and weighted shared nearest neighbors. This algorithm first introduces natural nearest neighbor computations to calculate weights, then redefines the similarity between data objects based on the definitions of first-order and second-order shared nearest neighbors. Subsequently, by fusing the definitions of shared nearest neighbor similarity and natural nearest neighbor weights, relative density and relative distance are calculated. Finally, a novel strategy for distributing cluster centers is designed. 【Result】Experimental results on sixs different types of datasets demonstrate that the proposed algorithm outperforms four other comparative algorithms significantly in terms of clustering performance. 【Conclusion】The method achieves better cluster center identification on datasets with non-uniform density, effectively addressing the aforementioned issues.