Pavement runoff could enter the karst aquifer system through sinkholes, karst windows, karst pools in karst areas, which could influence the karst water environment quality. Consequently, it is necessary to study the hydrochemical environment characteristics of pavement runoff in karst areas. Pavement runoff of Yaji, Qingshuiqiao and Baizhujing were sampled, characteristics of hydrochemical compounds and their influencing factors, hydrochemical environment quality were analyzed using multiple statistical method, Nemerow index method and comprehensive pollution index method. External influencing factor has small influences on the common hydrochemical ions, while has great influences on these trace elements. The compounds influencing the water environment of pavement runoff were nutrient compounds (NH₃-N, TP, COD_Mn), landscape compound (suspended solids) and metal compounds (Mn, Hg and TFe) by analyzing the concentrations of hydrochemical compounds of pavement runoff. These compounds have close relationships with pavement behaviors, surrounding vegetations, traffic flow and came from fuel, lube, slop oil, gasoline, worn tyre and vegetations. Four main factors with the cumulative variance contribution rate of 97.99% were extracted from the monitoring dataset using the factor analysis method. It could be known from the four main factors that carbonates weathering was the main source of hydrochemical compounds of pavement runoff, the second was the particles of atmospheric and pavement influencing the SS of pavement runoff, the third was the human activities including pavement behaviors and protective measures of surrounding vegetations. Hydrochemical environment quality of Qingshuiqiao, Yaji and Baizhujing decreased in turn by using the Nemerow index method and comprehensive pollution index method. Hydrochemical environment quality of Baizhujing were poorest, which had potential risks for the water ecological environment, the pavement runoff could be reused for the surrounding vegetations through reasonable measurements. The results could not only provide scientific instructions for the treatment measures of pavement runoff, but also provide scientific evidences for the reasonable exploitation and utilization of karst water resources.

Bismuth ferrite has become an effective semiconductor photocatalyst for the degradation of various wastewaters due to its narrow band gap, high chemical stability, and good visible light response. Pure phase BiFeO₃ nanofibers were prepared by electrospinning method. The optimal degradation conditions for Congo Red were obtained through single factor experiments such as calcination temperature, PVP (polyvinyl pyrrolidone)concentration, collection distance, spinning voltage, and pushing speed. Four factors that significantly affect photocatalytic efficiency were selected for response surface analysis experiments with four factors and three levels. After optimization, the optimal PVP concentration was 12.17 wt%, collection distance was 14.07 cm, and spinning voltage was 12.03 kV The pushing speed is 0.74 μm/s, and under this condition, the efficiency of BiFeO₃ photocatalytic degradation of Congo red can reach 90.43%. The phase analysis and morphology characterization of bismuth ferrite nanofibers were carried out using X-ray diffraction, scanning electron microscopy, Raman spectroscopy, and Fourier transform infrared spectroscopy. The results show that the pure phase BFO nanofibers prepared by electrospinning has a rough surface and obvious particle sensation, presenting a one-dimensional rod-shaped structure with a size of about 300 nm. This nanorod-shaped structure has a larger specific surface area and more active sites, Can improve the photocatalytic degradation efficiency of BFO.

To address the issues of low accuracy in rolling bearing life prediction and the difficulty of constructing health indicators, a bearing remaining life prediction model based on ASFF (adaptively spatial feature fusion) and AAKR (auto associative kernel regression) combined with CNN (convolutional neural networks) and BILSTM (bi-directional long-short term memory networks) was proposed. Firstly, the multidimensional features were extracted in the time domain, frequency domain, and time-frequency domain, and the sensitive features were screened using monotonicity and trend. Secondly, the sensitive features were feature fused using ASFF-AAKR to construct the health indicators. Finally, the health indicators were inputted into CNN and BILSTM to realize the life prediction of rolling bearings. The results show that the constructed life prediction model is better than other models, and the method has lower error and higher life prediction accuracy.

The physical properties lower limits of oil-gas charging in tight sandstone reservoirs are identified through a large number of core test and analysis data. The Chang8 reservoir types, pore-throat structure, and physical properties were clarified. Methods such as oil-gas occurrence, displacement pressure, physical property statistics, oil recovery index, and minimum pore-throat radius were employed to determine the current reservoir physical properties lower limit. By integrating the period of hydrocarbon accumulation and pore evolution, the critical physical properties during oil-gas charging were inverted. The results show that the reservoir types of Chang 8 are mainly feldspar sandstone and lithic feldspar sandstone in Fuxian area, with complex pore-throat relationship. These reservoirs are classified as tight reservoirs with low to extra-low porosity and extra-low to ultra-low permeability. It is preferred that the physical properties lower limits of the current reservoir are 7.0% and 0.15 mD, respectively. For inversion of oil-gas charging in Phase I (192.5~152.0 Ma), the lower limits of physical properties are 20.8% and 7.37 mD, respectively, for oil-gas charging in Phase II (152.0~126.0 Ma), the lower limits of physical properties are 8.2% and 0.22 mD. For oil-gas charging in Phase III (65.0~36.5 Ma), the lower limits of physical properties are basically consistent with the current lower limits of physical properties. The research findings provide an significant geological basis for the evaluation of reservoir and the prediction of favorable in the study area.

In order to improve the accuracy and efficiency of inventory counting in the process of monitoring and auditing biological assets, a biological asset detection model YOLOSC incorporating the attention mechanism and loss function optimization was proposed. Firstly, the SENet attention mechanism was introduced into the backbone network of the YOLOv5s model to enhance the ability of extracting the key features in the pictures of the biological assets. Secondly, the CIoU was adopted as the regression of the detection frames with the loss function to enhance the regression speed and localization accuracy of the detection frame during the training process. Finally, a biological asset datasets was constructed for targeted training of the proposed model to enhance the model detection effect. The experimental results show that compared with the YOLOv5model, the precision, recall, F₁ value and AP of YOLOSC are improved by 2.3%, 2.1%, 2.7% and 1.6%, respectively, which proves the effectiveness of the proposed biological asset detection model YOLOSC.

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.

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.

Ship targets in remote sensing images have multi-scale characteristics, changeable backgrounds, and complex meteorological characteristics, which lead to low accuracy, false detection, and missed detection of small target ships. In response to the above situation, an improved small-target ship detection model based on YOLOv5s was proposed. First, in order to solve the problems of scale changes and background variability in ship detection, the ASFF(adaptive spatial feature fusion) module was introduced. Secondly, in order to reduce the calculation amount and parameter amount of the detection network, the BoTNet attention mechanism was introduced, and then in order to improve the overall network to improve the detection accuracy, the EIoU border loss function was used, and finally the Slim-neck network was introduced to ensure the overall lightweight of the network. Experiments show that on the main data set LEVIR-Ship, compared with the benchmark YOLOv5s, mAP@0.5 increased by 7.1% to 81.3%, the number of parameters is reduced by 0.44 M, the calculation amount is reduced by 0.6GFLOPs, and the weight was reduced by 0.9 M. The proposed method performs better in various key indicators and achieves high-precision small target ship detection in complex environments. Comparative experiments are conducted on the verification data set McShips. The experiments show that the proposed method still performs better, verifying the universal applicability of the proposed method.

With the increasing penetration of distributed power sources in the distribution network, active distribution has become the mainstream direction of the power grid in the future. Microgrids and active stations that are slightly smaller than their scale will gradually increase. Whether these subsystems can be used as energy to schedule becomes the key to improving the economy and stability of power grid operation. Therefore, aiming at the overall optimal scheduling of microgrid and active station area with distributed power supply, a scheduling model solution method based on improved target cascade method was proposed, which mainly takes the optimal benefits generated by different stakeholders as the final scheduling target, and adopts the opportunity constraint description for the processing of uncertain factors of wind and solar. The upper layer is the distribution network and the optimal target of the distribution network, and the lower layer is the microgrid and active station area with the ability to participate in the scheduling. Based on the modeling of distribution network and microgrid, the improved target cascade method was introduced, and the interactive power was used as a shared variable to equivalent the generator and virtual load, so as to realize the decoupling and independent optimization of the upper and lower layers. The comparison of the experimental results shows that the target cascade method with the balance coefficient can obtain better results in the number of iterations, convergence performance, anti-interference performance and overall economic evaluation.

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.