Aiming at the shortcomings of low-quality face recognition algorithms based on unified feature space, such as poor robustness to low-quality faces and limited feature representation capability, a low-quality face image recognition algorithm based on knowledge distillation was proposed. First, the ResNeXt network was used as the backbone feature extraction network, and the two-channel attention module was introduced to construct a teacher-student knowledge distillation framework with an attention mechanism. Secondly, the output features of the teacher network were adopted as labeled knowledge, and the effective recognition features were passed to the student network. And the attention graph features were adopted as the intermediate layer knowledge to solve the lack of single knowledge information in the output layer, and the feature knowledge was enriched by combining two kinds of knowledge distillation to ensure the diversity of knowledge information in the teacher network model. Then, the weighted average of labeled knowledge distillation loss, attention graph distillation loss, and recognition loss were fused as the total network loss function to ensure that the student network model has a better learning ability. Finally, tested under different quality images in AgeDB-30 and CPLFW test sets, the results of the ablation experiments show that compared to the generic face recognition model without distillation, the model with two types of knowledge distillation gains 2.25%, 11.33%, 24.64% and 2.8%, 10.58%, 27.85% improvement in recognition accuracy, respectively. Comparative experiments show that the algorithm proposed in this paper also obtains different degrees of improvement in accuracy compared to other mainstream algorithms.

In order to address issues such as high noise, low brightness, and blurred details in low-light conditions, a new algorithm named UMDCEAD-NET, integrating zero-reference depth curves for low-light image enhancement and denoising, was developed. The algorithm's design was initially centered around a feature extraction network, employing a U-Net architecture as the backbone network. To enhance the feature extraction capabilities and preserve more detailed image information, Mobile-Net was integrated into the downsampling phase of the U-Net backbone. Subsequently, to address the issue of inadequate lighting and pixel-level image degradation, the extracted features underwent iteration using depth curve estimation (LE-curve), in conjunction with depth separable convolution, which served to reduce the network model's parameter count. Furthermore, five non-reference loss functions were engineered to bolster the algorithm's generalization capabilities and its retention of detail under varying lighting conditions. The enhanced image was then subjected to noise reduction in tandem with AD-NET(attentional denoising network), thereby diminishing the noise and aligning the image more closely with human visual perception. Experimental outcomes demonstrated that the proposed algorithm achieved an average PSNR (peak signal-to-noise ratio) of 22.29 on the public dataset Zero-DCE, which exceeded the performance of the Zero-DCE++ algorithm by 32%. Additionally, on the public dataset LOL, the algorithm attained an average PSNR of 21.15, outperforming the SGZ algorithm by 3%. These results indicate that the algorithm effectively mitigates noise in enhanced images, enriching the detail information in both dark and bright regions, and significantly improving image quality compared to other mainstream algorithms.

In order to deeply understand the spatio-temporal distribution patterns of forest fires and reduce the adverse effects of forest fires on the ecological environment and human activities, the key parameters and dynamics of multi-source satellite for fire point identification was established using data from 8 domestic and foreign meteorological satellites and based on the classic context method. The satellite monitoring buffer zone radius verification method was used to verify the authenticity of forest hot spots retrieved from multi-source satellites, and the real forest hot spot data during the fire prevention period in 2021—2022 were used to analyze the spatio-temporal characteristics of forest fires. The results show as follows. The accuracy of satellite fire spot monitoring is 84.42%, and the fire point classification accuracy is 89.90%. The established inversion method is reasonable and reliable. The spatial distribution of forest fires in Hunan is “more in the southwest and less in the northeast”. At the same time the high-incidence areas are mainly distributed in southern Hunan, and the second-highest-incidence area is western Hunan. In summary, the risk of forest fires during the autumn prevention period is much greater than that during the spring prevention period. During the extreme high temperature and drought in 2022, forest fires were mainly distributed in the southern Hunan region and the Hengshao Basin. From the perspective of process distribution, the distribution of forest fires can be divided into four stages. The number of forest fires in the first three stages showed a significant increase trend, and in the third stage, the number of forest fires increased significantly. Fire risk is the most serious. In the fourth stage, due to the dual impact of precipitation and the province's fire ban, the risk of forest hot spots was significantly reduced.

In order to simulate the seismic damage process of RC (reinforced concrete) columns precisely and effectively, the method of seismic damage analysis of reinforced concrete members was established by combining the uniaxial damage constitutive model of concrete and steel bar with the flexibility-based fiber beam-column element model. The quasi-static test of reinforced concrete column was simulated and analyzed. It is found that the simulation model can simulate the degradation process of stiffness and bearing capacity of component with good accuracy. The shaking table test of reinforced concrete column under biaxial loading was simulated and analyzed. The results show that the simulation model can simulate the nonlinear dynamic behavior and damage distribution of component with good accuracy. Furthermore, the established model can effectively simulate the evolution process of seismic damage of components, and describe weak parts of components, and has high computational efficiency and solution accuracy, which can be used to analyze the collapse process of buildings and bridge structures under earthquake action.

In order to study the effect of temperature on the shear strength of fully weathered mudstone with different saturations,fully weathered mudstone specimens from a typical seasonal frozen region were used as the object of study.The GDS dynamic triaxial test system and the GDS unsaturated test system were used to conduct indoor triaxial tests on specimens with different saturations to investigate the trends of the shear strength parameters of fully weathered mudstone in the seasonal frozen region under different temperatures and surrounding pressures, and to compare and analyse the shear strength parameters of the two different saturations of the fully weathered mudstone were compared and analysed. The results show that the fully weathered mudstone is strongly influenced by saturation and temperature.The cohesion of both soil samples increases with decreasing temperature, and the internal friction angle increases and then decreases. The cohesion of the unsaturated mudstone specimens is consistently greater than that of the saturated mudstone specimens when the temperature and test system conditions are consistent. Saturation has a low effect on the internal friction angle of the specimens, which peaks at -5 ℃ and 0 ℃for the two soil samples respectively.

During the geothermal development of dry hot rock, the high temperature rock mass is subjected to repeated cold and thermal cycles. It leads to the rupture of thermal reservoirs and the change of physical and mechanical properties. In order to further explore the mechanism of the influence of temperature and cooling-heating cycle on rock characteristics, the granite specimens subjected to different high temperature nodal heat treatment were treated with natural cooling, fresh water cooling and seawater cooling respectively. The physical and mechanical indexes and microstructure were studied. The damage constitutive equations of granite under uniaxial compression with three cooling cycles were established. The results show as follows. With the increase of temperature and cycle times, the mass loss rate is in the order of freshwater cooling > natural cooling > seawater cooling, but at 600 ℃, serious particle breakup and shedding cause the mass loss of seawater cooling rock sample to exceed that of natural cooling. The elastic modulus, compressive strength and tensile strength are decreasing. The damage of water cooling to high temperature rock is greater than that of natural cooling. The damage effect of high temperature is more obvious than that of cycle times. The micro-cracks of seawater cooling rock sample are more developed. The damage variables consider the effects of temperature and cycle times, and add the damage coefficient to consider the damage effects of freshwater cooling and seawater cooling. The uniaxial compressive stress-strain curves combined with damage analysis under load are compared with the experimental results in a high degree of fitting, which reflects the rationality of the model.

With the development of the concept of composite materials, in order to further explore the mechanical properties of composite modified recycled concrete under the coupling effect of NS (nano-SiO₂) modified recycled coarse aggregate and PVA(polyvinyl alcohol) fibers, slump, cubic compression, axial compression, splitting tensile and flexural tests were carried out to study the working performance and mechanical performance changes of modified recycled coarse aggregate concrete with increasing PVA fiber content under different substitution rates. The results show that the slump of concrete increases with the increase of fiber volume. The damage of concrete is brittle, and the damage pattern of recycled concrete mixed with fiber is better. When the fiber volume content is 0.05 vol% and 0.10 vol%, the cubic compressive strength, ultimate bearing capacity, splitting tensile strength, folding strength and static elastic modulus will decrease under different regeneration and replacement rates, but all the strengths will exceed and increase when the fiber volume content is 0.15 vol%. PVA fiber will reduce the ultimate compressive bearing capacity and have different positive and negative effects on the peak strain. It is recommended to add PVA fiber with a volume content of 0.1 vol%. If PVA fiber is needed, it is recommended to use it when the regeneration and replacement rate is less than 30 wt%. In addition, it is found that the modified reclaimed coarse aggregate has good performance and can effectively replace natural aggregate or be mixed with natural aggregate in practical engineering.

In order to study the effect of basalt fiber on the durability of recycled concrete under the erosion of salt solution, the durability of recycled concrete specimens with different basalt fiber contents after salt-dry-wet cycle coupling erosion was studied. A comprehensive durability index D value was established to evaluate the durability of recycled concrete based on the entropy weight method. The effects of dry-wet cycle period of salt solution and basalt fiber content on D value were analyzed. A GM (1,1) mean model was constructed to reveal the time-varying law of the D-value of recycled concrete, and the predicted life of recycled concrete under different conditions was obtained. The results show that the D value can reflect the influence of different salt solution dry-wet cycle cycles and basalt fiber content on the durability of recycled concrete. As the salt solution's dry-wet cycle increased, the D value of the specimen gradually decreased, indicating a severe change. However, adding basalt fiber to the recycled concrete can effectively enhance its D value and durability. When the content of basalt fiber is 1.0%, the durability of recycled concrete is the best. The GM(1,1) model can more accurately predict the time-varying pattern of D values of recycled concrete under coupled salt-dry-wet cycle erosion when the amount of data is small.

To efficiently identify the opening and closing status of train doors and control the synchronous opening and closing of platform doors, a lightweight MobileNet network and machine vision based image recognition method was proposed to achieve linkage control between high-speed railway platform doors and train doors. A large dataset of train door images was collected from Beijing South Station and preprocessed to serve as the training and testing dataset for the model. The constructed network was trained and optimized using a binary cross-entropy loss function and the Adam optimization algorithm to achieve efficient and accurate recognition of door status. Validation results demonstrate an accuracy rate of over 95% in recognizing train door actions, with recognition time kept within 400 milliseconds. These results meet the current industry application requirements and greatly enhance the automation and intelligence level of the platform door system.

Aiming at the problem of coaxial wheel partial wear of HX high-power electric locomotive in China, the locomotive dynamics model was established based on the dynamics software SIMPACK, and the damage function prediction method based on wear number was used to analyze the influence of different coaxial sequence, wheel diameter difference, curve radius and other conditions on wheel tread damage. The results show that when there is wheel diameter difference in one axle or multiple axles, the influence on tread damage of 2-axle and 3-axle wheels is greater, and the influence on 1-axle wheels is less. For different values of wheel diameter difference, with the increase of wheel diameter difference, the impact of rolling contact fatigue degree on the right wheel tread of 1-axle is small, and the wear degree of the left wheel tread increases. The damage degree of wheel on both sides of 2-axle is reduced; the cracks of 3-axle left wheel are accentuated. Under left curve condition and R400 condition, wheel diameter difference has greater influence on the wear and rolling contact fatigue of each axle. Under right curve condition and R800 condition, wheel diameter difference has more influence on the wear and rolling contact fatigue of each axle. Compared with right curve condition, wheel diameter difference has more influence on wheel tread damage under left curve condition. Compared with curve radius, wheel diameter difference has less influence on wheel tread damage.