The internal damage of the steel rail is serious, but the non-destructive testing B-display detection image has a lot of noise and noise, and the spatiotemporal distribution characteristics of different damages are not obvious, making it difficult to effectively identify. In response to this situation, a rail screw hole crack B-image recognition algorithm based on improved YOLOv8 was studied to improve the accuracy of intelligent identification of rail damage. Firstly, to reduce the missed detection of small damage targets, RepHGNetv2 network was used to optimize the YOLOv8 backbone network and improve the detection recall rate. Then, in order to improve the adaptability of the model to different types of damage detection, the detection head of YOLOv8 was replaced with Effientnet to improve the detection accuracy of the model. Finally, the LSKA attention mechanism was introduced into the SPPF module to enhance the model’s anti-interference ability against noise signals and improve its accuracy. The actual line detection results have verified that the detection accuracy of the above model reaches 95.1%, the recall rate reaches 93.8%, and the average accuracy reaches 97.6%, which is improved compared to other commonly used algorithms.

Traditional whole-brain dynamical modeling techniques are typically constrained by static single features, neglecting dynamic fluctuations in brain networks and lacking qualitative analysis of corresponding indicators, which limits modeling accuracy and comprehensibility. In order to address this issue, a multi-objective expectation maximization algorithm based on bifurcation analysis was proposed. This approach integrates a dynamic mean-field model with brain structural-functional features extracted from multi-mode imaging data for modeling purposes. Bifurcation theory was employed to qualitatively analyze multiple constraint indicators of the model, including functional connectivity, dynamic functional connectivity, and metastability for model inversion. Initial parameter values were determined through bifurcation analysis, and parameter combinations were iteratively refined using an expectation maximization algorithm. Quantitative analysis validates the accuracy and stability of this method.

In traditional blind deconvolution algorithms, recalculating the gradient or redesigning the optimization approach for filter coefficients becomes necessary when altering the characterization index. This requirement can render the development process of new blind deconvolution algorithms inflexible. To address these issues, a blind deconvolution algorithm that employs NRO(Newton-Raphson optimizer) to seek out the optimal filter coefficients was proposed. Initially, generalized spherical coordinate transformation was used to define the search range for the filter coefficients. Subsequently, the generalized l_p/l_q norm of the envelope spectrum was adopted as the characterization index. The proposed blind deconvolution algorithm is then utilized for the early detection of minor faults in rolling bearings. Both simulation and experimental results confirm the efficacy of the proposed algorithm, demonstrating its faster convergence rate compared to classical PSO(particle swarm optimization).

To explore the application regularity and mechanism of formulas containing Huang Qin (Scutellariae radix)-Bai Zhu (Atractylodis macrocephalae Rhizoma) (HQ-BZ) herb pair in the treatment of ICP (intrahepatic cholestasis of pregnancy). All literature on prescriptions containing HQ-BZ herb pairs against ICP was screened from the VIP, Wanfang, and CNKI databases. Subsequently, the R language was employed to analyze and summarize its medication rules and core prescriptions. Network pharmacology was used to predict the mechanism of core prescriptions against ICP, followed by molecular docking and experimental verification to confirm the potential mechanism. A total of 68 prescriptions were included, involving 67 herbs characterized mainly by cold, bitter, and spleen meridian. The core prescription “Artemisiae scopariae Herba-Rhei radix et Rhizoma-Gardeniae fructus-Scutellariae radix-Atractylodis macrocephalae Rhizoma-Poriacocos” was obtained based on the comprehensive analysis of traditional Chinese medicine data, among which quercetin, apigenin, and other key active components may act on core targets such as AKT1 (serine/threonine kinase B1), BAX (BCL2-associated X protein), and participate in PI3K-AKT (phosphatidylinositol 3-kinase/protein kinase B), apoptosis, and other multiple targets and pathways to play the role of ICP therapy. The molecular docking results showed that apigenin demonstrated superior binding affinity with the top 11 core targets compared to quercetin and beta-sitosterol. HTR-8/SVneo cell experiments proved that apigenin significantly reduced the apoptosis rate induced by TCA (taurocholic acid) and elevated the protein expression levels of Bax/Bcl-2 (P<0.01), as well as p-PI3K/PI3K, and p-AKT/AKT (P<0.01). Pre-treatment with LY294002 could reverse the anti-apoptosis effects and the expression levels of the aforementioned proteins induced by apigenin. In summary, the core prescription that includes the HQ-BZ can provide references for the clinical prescription of ICP. Apigenin, a key component of core prescription, can inhibit the apoptosis in HTR-8/SVneo cells induced by TCA and has the potential to treat ICP, and its mechanism may be related to the regulation of the PI3K-AKT signaling pathway.

To address the reduction in support bearing capacity caused by concrete shrinkage and to improve the damping performance of supports under mining-induced seismic conditions, this study focuses on the energy-absorbing and shock-resisting performance of single arch sand-filled steel tubular frames and the damping performance of multi-arch combined support systems. Finite element software was employed to establish models of surrounding rock and sand-filled steel tubular frames, as well as multi-arch support systems connected with flexible cables and dampers. The performance of sand-filled steel tubular frames under static and dynamic loading, as well as their seismic resistance under mining-induced tremors, was investigated. The results indicate that the deformation of the tunnel under static loading remains stable, while the support effectiveness is satisfactory under impact loading except for relatively large deformations at the crown. Under static and dynamic loading, the equivalent plastic strain at the crown of the sand-filled steel tubular frames shows a significant increase, while changes in other areas remain minimal, demonstrating good load-bearing capacity. In the three-arch support system, the third arch experiences reduced vibration amplitude due to the dual energy dissipation effects of flexible cables and dampers. Calculations of the safety factor at the maximum shear stress of the tunnel reveal a significant improvement in the seismic performance of adjacent supports, providing insights for further studies on support damping mechanisms.

Strong pulsation is one of the most important causes of damage to hydraulic structures such as stilling basins, so it is crucial to clarify the characteristics of pulsation in hydraulic structures to ensure the safe operation of the project. In this study, a fine numerical model of the hydraulic model of a stilling basin was established based on DES (detached eddy simulation) model and VOF (volume of fluid) method, and the simulation results were in good agreement with the experimental results. Based on the numerical simulation results, the distribution of pressure fluctuation, pulsating velocity, vorticity, and turbulent kinetic energy in the stilling basin were analyzed. The results show that the pressure fluctuation in the stilling basin shows a bimodal distribution along the flow direction, and the pulsation accounts for more than 10% in the flow impact area. The spanwise and streamwise velocity pulsation dominate in the stilling basin, and the vertical pulsation is weaker, while the spanwise pulsation at the centerline of the bottom floor in the collision area decreases rapidly. The streamwise and vertical vorticity on the centerline of bottom floor and surface outlet are characterized by “small time-average value and large pulsation value”, and the transverse-axis vortex caused by the drop sill has a large influence on the distribution of the spanwise vortices. The results of turbulent kinetic energy spectrum analysis show that the pulsations in the stilling basin mainly consist of a large number of low-frequency pulsations below 1 Hz, indicating that the turbulent fluctuations in the stilling basin are mainly controlled by large-scale and low-frequency vortices. The analysis helps to deepen the knowledge of the flow characteristics of flood discharge and energy dissipation structures such as stilling basins, and provides a certain reference for design and safety assessment.

The hybrid DC transmission system has problems such as inconsistent boundary components, inconsistent fault response characteristics, difficult resolution of high resistance fault effects, and low accuracy in identifying near end faults, which reduces the reliability of protection schemes. Therefore, the phase characteristics of the regional refractive index of the hybrid DC transmission system were analyzed for the first time, and a single ended protection scheme suitable for hybrid boundaries was proposed based on this. Firstly, establish a hybrid DC transmission system model and analyze the traveling wave transmission characteristics of different fault types. Subsequently, the fault areas of the hybrid DC transmission system were divided, and the refractive index expressions and phase frequency characteristics of the areas were derived separately. Finally, a single ended protection scheme based on a specific frequency refractive index is proposed and its performance is tested. The test results show that the proposed protection scheme not only has the speed of traditional protection schemes, but also has better resistance to high impedance faults, noise interference, and other abilities.

Aiming at the problems of poor real-time detection, low accuracy, and false detection and omission of pavement disease detection including hole and crack, an improved algorithm based on YOLOv9 was proposed to resolve the problem. Firstly, AKConv (alterable kernel convolution) was introduced into the backbone network to replace the convolution module in RepNCSPELAN4, which improves the feature extraction ability of the network for different diseases and effectively solve the problem that road disease is difficult to distinguish from background environment features. Secondly, selective image attention mechanism (SimAM) and DySample sampling modules were introduced to focus on the key information in the detection head, and the capability to extract information features was enhanced more efficiently. Finally, the inner-IOU function was used to optimize the weight parameters of the model to improve the learning ability of mixed samples. The experimental comparison between YOLOv9-c and our model showed that the accuracy, recall rate and MAP of the improved model are increased by 40.17%, 15.99% and 20.95% respectively. The performance has been significantly improved, and the detection effect is more accurately and efficiently, and the accuracy and generalization ability of pavement disease detection algorithm are improved.

To address the limitations of traditional protocol recognition methods caused by the presence of numerous non-standard protocols in IC (industrial control) sector, a method based on edge-distributed deep learning was studied to enhance IC protocol recognition technology. A recognition method based on CNN (convolutional neural networks) was proposed: real IC protocol data from the network was collected and preprocessed, and an appropriate CNN model was selected according to protocol characteristics to implicitly extract the essential features of the protocols. This achieved classification and recognition of seven types of IC protocols with an accuracy of up to 99.92%. Furthermore, the IC protocol recognition model was deployed at the network edge, leveraging a data-parallel distributed strategy for collaborative training within an edge server computing cluster. This improved the training efficiency of the model by 1.87~2.81 times while maintaining high accuracy. The results show that this method significantly improves the accuracy of IC protocol recognition, greatly enhances model training efficiency, and is well-suited for deployment in edge computing environments. It is evident that this method has significant value in optimizing IC protocol recognition performance.

The cornering stiffness of automobile tires is closely linked to the vehicle’s handling characteristics, and accurately estimating the tire cornering stiffness in real time is of significant importance for enhancing the stability of vehicle handling. Addressing the challenge of direct measurement of cornering stiffness, a real-time identification method based on the estimation of tire lateral force and slip angle was proposed. Firstly, considering the influence of longitudinal force on lateral force, a tire lateral force estimator was designed based on the yaw dynamics model and sliding mode observer algorithm, followed by the design of a slip angle feedback estimator based on the estimation error of lateral force. Secondly, a nonlinear tire force model that describes the relationship among tire lateral force, slip angle, and cornering stiffness was established. Taking the real-time estimated lateral force and slip angle as inputs, a recursive least squares online identification algorithm with limited memory was designed to address the issue of estimation error due to “data saturation” and improve identification accuracy. Finally, joint simulation experiments using Simulink and CarSim were conducted. The experimental results indicate that the estimation error of tire lateral force is approximately 4.153 9% on average, while the estimation error of tire slip angle is 3.285 2% on average. The identification model based on the recursive least squares method is robust to changes in road conditions, demonstrating good tracking accuracy and stability under both high and low adhesion conditions, with an average estimation accuracy of tire cornering stiffness of approximately 98.379 3%.