Aiming at the multi-source noise of inertial navigation in the attitude estimation of coal mine bolting jumbo，a noise reduction method of inertial navigation heterogeneous signal is proposed based on noise sensitive prior and improved variational mode decomposition（VMD），which avoids the over-decomposition and under-decomposition problems caused by the parameter fixation. Firstly，the noise sensitivity difference of the heterogeneous signals （acceleration and angular velocity） of coal mine bolting jumbo is investigated by using the variation of the signal characteristics in the time and frequency domains. Secondly，according to the noise-sensitive characteristics，the dual decomposition layer and energy fluctuation model are constructed，so that the decomposition parameters have the ability of adaptive optimization and the synchronous optimal decomposition of the inertial-guide heterogeneous signals is realized. Based on the Pearson correlation coefficient （PCC），the modal component screening parameter，correlation coefficient P，is designed to consider the noise sensitivity difference，to achieve screening practical modal components and simultaneous noise reduction of heterogeneous signals. Finally，the proposed method is compared with the noise reduction results of complementary ensemble empirical mode decomposition （CEEMD） and improved complete ensemble empirical mode decomposition with adaptive noise （ICEEMDAN）. The results show that the method proposed in this paper considers the noise sensitivity differences of heterogeneous signals，thereby improving the signal-to-noise ratio of inertial measurement and enhancing the attitude initialization accuracy of bolting jumbo. The pitch error is reduced by 81.818 %，and the yaw error is reduced by 87.958 %，which lays a good foundation for accurate roadway support.

To address the issue of metallic objects around a balise affecting its electromagnetic transmission performance，a balise antenna model is established in electromagnetic simulation software. The simulation experiments are carried out by varying three parameters：the metal surface area，the vertical distance between the metal surface and the balise，and the metal surface thickness. The uplink signal amplitude curves of the balise under different parameters are obtained，and the transmission performance metrics are calculated to analyze the impact. Results show that： a larger metal surface area leads to lower performance metrics，such as a reduced number of safety message frames received by the balise transmission module （BTM），with a more significant degradation and greater interference from the sidelobe region. When the metal area is greater than 320 mm×320 mm，the uplink field strength consistency requirement can no longer be met. A greater absolute distance between the metal surface and the balise results in less interference from the sidelobe region，and the distance must be greater than 123 mm to satisfy the field strength consistency requirement. Increased metal surface thickness causes greater interference from the sidelobe region，and the thickness should not exceed 1 mm.

A novel approach based on electromechanical impedance is proposed to evaluate the corrosion degree for grounding conductors，which are difficult to simply detect and evaluate for traditional methods. Firstly，according to the electrochemical corrosion model of metals，the grounding conductor parametric corrosion model is established by importing the change of grounding conductor radius as corrosion parameter. Secondly，the corrosion model between an electromechanical impedance resonance frequency and corrosion parameter is established with the analysis of system electromechanical impedance and conductor mechanical admittance. Then，the resonance frequency and corrosion parameter signal of grounding conductors are obtained via finite element simulation method，while the coefficient of the corrosion model is obtained by the least square method. The results show that the electromechanical impedance detected by asymmetrical sensor layout can reflect variations of the corrosion parameter more clearly than that by symmetrical sensor layout. Linear model by finite element simulation and the least square method can predict the corrosion parameter. Moreover，the high consistency between the datum predicted by the linear model and the experimental datum，indicates the linear model is very accurate and can be used to detect the conductor corrosion in field applications.

A parameter identification method is proposed to accurately capture the nonlinear dynamic characteristics of dielectric elastomer actuators （DEAs）. First，the response signal of the actuator is acquired under swept-frequency excitation using the time-frequency analysis capability of the transient-extracting transform （TET）. Then，the harmonic and fundamental frequency components are separated and extracted，and the transfer functions for each component are computed to derive the overall transfer function of the DEA. Finally，the results are compared with experimental data. The proposed method achieves a fitting accuracy of 92.11% for the fundamental frequency transfer function and 90.35% for the second harmonic transfer function. This approach does not require prior knowledge of material properties or free energy density functions，and incorporates the influence of high-order harmonic components，offering a novel solution for parameter identification in electroactive material structures.

Wind tunnel pressure tests are conducted on a high-speed railway station roof to study the non-Gaussian characteristics and extreme wind pressure distribution on the long-span roof surface. First，the surface wind pressure is classified into Gaussian and non-Gaussian distributions. Then，the fitting effects of three different single probability distributions （Gumbel，Lognormal，and Weibull） and their corresponding combined distributions （double Gumbel，double Lognormal，and double Weibull） on the wind pressure time history of the roof surface are compared. The extreme wind pressures obtained from the combined probability distributions are compared with the estimates from the modified Hermite method. Finally，the extreme wind pressure distribution on the roof under all wind directions is presented. The results show that the combined probability distributions provide a better fit to the wind pressure time history than the single distributions. The extreme value estimates from each combined distribution at the same guarantee rate are more accurate than those from the single distribution. The combined distributions generally yield better estimates at the 99.90% guarantee rate compared to the modified Hermite method. The extreme wind pressure varies significantly with the wind direction，and under all wind directions，the minimum pressure coefficient reaches its lowest value at the middle of the roof edge side，reaching -5.9.

In order to improve the fault detection performance of wheelset bearings under small sample image conditions，a machine vision inspection method based on a novel multi-resolution siamese neural network （MrSNN） is proposed for surface defect detection of wheelset bearings. First，the siamese neural network （SNN） is used as the basic model framework. A multi-resolution convolution fusion block （MrCFB） containing convolution kernels of different sizes and dilation factors is constructed to comprehensively extract the detailed features and contour features from images. Then，a dual attention mechanism combining channel and spatial information is adopted to recalibrate the multi-resolution feature weights，further enhancing the image feature extraction capability of the model. Finally，the algorithm is validated through the detection and analysis of four types of wheelset bearings images： normal，scratched，pitted and spalled. Experimental results show that the recognition rate for the three types of faulty images reaches 100%，the recognition rate for normal images is 95%，and the overall recognition accuracy is 98.75%. The recognition accuracy is superior to that of traditional SNN and YOLO-V5 models.

When early failures occur in planetary gearboxes，the weak fault features are difficult to extract and identify due to the interference of background noise in industrial environments and the attenuation of fault impacts in complex transmission paths. To address this issue，a sparse-guided improved empirical wavelet transform （IEWT） is proposed combined with multipoint optimal minimum entropy deconvolution adjusted （MOMEDA） method for weak fault feature extraction. Firstly，a new fault composite index （FCI） is introduced，and the original signal is adaptively decomposed into a set of IEWT components based on the amplitude envelope of the signal spectrum. Secondly，the sensitive components，selected through the sparse-guided method，are used as the sparse representation of the original weak fault signal. Finally，the MOMEDA technique is applied to the sensitive component signals to reduce signal noise and extract the weak fault feature frequencies for identification. The effectiveness of the proposed method is validated through simulations and experiments，successfully extracting and identifying the weak fault features of planetary gearboxes. This demonstrates that the method has good diagnostic performance for noisy，non-stationary，and non-linear fault signals in planetary gearboxes，providing a new approach for the diagnosis and identification of weak faults in engineering practice.

To study the influence of subway wheel polygons on low-frequency vibration of the car body，the polygon wear of wheels of a subway line is investigated，on the basis of grasping the distribution characteristics of wheel polygons of subway lines. A vertical dynamic model of elastic car body considering wheel polygons is established，by the time-domain integral solution method. The relationship between wheel polygon excitation frequency and low frequency vibration of vehicle body is studied. By comparing the low-frequency vibration of wheel polygons of different orders，the effect of changes in the operating speed of metro vehicles under service conditions and changes in the radius of wheel wear on the polygonal action of the wheel body is discussed separately. It is shown that a wheel polygon of order 1—3 at common operating speeds generate a low-frequency excitation frequency of 0—20 Hz，and when the excitation frequency is close to the first-order droop frequency of 10.2 Hz resonance will occur. At the beginning of service，the influence of the second order wheel polygon becomes severe on the low frequency vibration of the car body. With the wear of the wheel radius during service，the influence of the first three order wheel polygon changes on the vibration law of the car body. This paper provides a good reference value for service subway operation and wheel maintenance.

To investigate the fatigue strength of welded structures under high cyclic loads，a synchronized coupled testing method combining digital image correlation （DIC） and infrared thermography （IR） is employed to simultaneously acquire full-field strain and temperature data on the structural surface，explaining the evolution of structural damage. First，tensile testing is performed on 45# steel specimens，and the patterns of variations in surface temperature and strain during the tensile process are obtained. The temperature characteristics at various stages of the damage process agree well with the material load-time curve，reflecting the different stages of damage in 45# steel. Finally， ultra-low cycle fatigue testing is conducted on GH4061 welded structure used in engines，capturing the entire process of crack initiation，propagation，and fracture under cyclic loading，with synchronized full-field strain and temperature testing and analysis performed. The experimental results demonstrate the feasibility and effectiveness of this method in monitoring fatigue damage in welded structures.

The transient vibration of a vertical axis washing machine is strong in the dehydration process，so a new planar variable damping structure is proposed to reduce the transient vibration. Firstly，the kinetic energy and potential energy of various rigid bodies of the washer are deduced，the generalized forces of the suspension structure are described，the force of the liquid balancer is analyzed，and the vibration model of the vertical axis washing machine is established using Lagrange's equation. The working principle of the planar damping structure is explored，its damping force is described and its suppression effect on transient vibration of the washer is verified. Secondly，the influence of the planar damping structure on dynamic characteristics of the washer is evaluated，the bifurcation theory is employed to analyze stability of the system. Furthermore，the distributions of the stable regions of the system is analyzed，and the appropriate disengaging speed range of the damping structure is obtained. Finally，the effect of the damping structure for suppressing transient suppression of the washer is validated though experiments，and the appropriate disengaging speed of the structure is analyzed. The results show that the planar damping structure can suppress transient vibrations effectively with little influence on other dynamic characteristics of the washer.