High Computational Efficiency Research Articles

Research on Lateral Stability Control of Four-Wheel Independent Drive Electric Vehicle Based on State Estimation

This paper proposes a hierarchical framework-based solution to address the challenges of vehicle state estimation and lateral stability control in four-wheel independent drive electric vehicles. First, based on a three-degrees-of-freedom four-wheel vehicle model combined with the Magic Formula Tire model (MF-T), a hierarchical estimation method is designed. The upper layer employs the Kalman Filter (KF) and Extended Kalman Filter (EKF) to estimate the vertical load of the wheels, while the lower layer utilizes EKF in conjunction with the upper-layer results to further estimate the lateral forces, longitudinal velocity, and lateral velocity, achieving accurate vehicle state estimation. On this basis, a hierarchical lateral stability control system is developed. The upper controller determines stability requirements based on driver inputs and vehicle states, switches between handling assistance mode and stability control mode, and generates yaw moment and speed control torques transmitted to the lower controller. The lower controller optimally distributes these torques to the four wheels. Through closed-loop Double Lane Change (DLC) tests under low-, medium-, and high-road-adhesion conditions, the results demonstrate that the proposed hierarchical estimation method offers high computational efficiency and superior estimation accuracy. The hierarchical control system significantly enhances vehicle handling and stability under low and medium road adhesion conditions.

Research on Small-Target Detection of Flax Pests and Diseases in Natural Environment by Integrating Similarity-Aware Activation Module and Bidirectional Feature Pyramid Network Module Features

In the detection of the pests and diseases of flax, early wilt disease is elusive, yellow leaf disease symptoms are easily confusing, and pest detection is hampered by issues such as diversity in species, difficulty in detection, and technological bottlenecks, posing significant challenges to detection efforts. To address these issues, this paper proposes a flax pest and disease detection method based on an improved YOLOv8n model. To enhance the detection accuracy and generalization capability of the model, this paper first employs the Albumentations library for data augmentation, which strengthens the model’s adaptability to complex environments by enriching the diversity of training samples. Secondly, in terms of model architecture, a Bidirectional Feature Pyramid Network (BiFPN) module is introduced to replace the original feature extraction network. Through bidirectional multi-scale feature fusion, the model’s ability to distinguish pests and diseases with similar features and large scale differences is effectively improved. Meanwhile, the integration of the SimAM attention mechanism enables the model to learn information from three-dimensional channels, enhancing its perception of pest and disease features. Additionally, this paper adopts the EIOU loss function to further optimize the model’s bounding box regression, reducing the distortion of bounding boxes caused by high sample variability. The experimental results demonstrate that the improved model achieves a significant detection performance on the flax pest and disease dataset, with notable improvements in the detection accuracy and mean average precision compared to the original YOLOv8n model. Finally, this paper proposes a YOLOv8n model with a four-headed detection design, which significantly enhances the detection capability for small targets such as pests and diseases with a size of 4 × 4 pixels or larger by introducing new detection heads and optimizing feature extraction. This method not only improves the detection accuracy for flax pests and diseases but also maintains a high computational efficiency, providing effective technical support for the rapid and precise detection of flax pests and diseases and possessing an important practical application value.

An analytical method for broadband acoustic analysis of 2D cavities containing or bounded by porous materials

Real-life vibro-acoustic systems often involve porous treatments, resulting in complex-valued and frequency-dependent models that are challenging to solve. Traditional prediction techniques like the finite element (FE) method requires huge computational cost, especially in the mid to high frequency ranges. This paper develops a novel spectral dynamic stiffness (SDS) formulation, using very few number of degrees of freedom but describing the broadband acoustic behaviour of acoustic cavities with porous materials highly accurately. The method employs frequency-dependent shape function that satisfies exactly the (damped) Helmholtz equation to describe the (equivalent) acoustic pressure field, and also features an innovative approach to use the fast-convergent Modified Fourier series to describe any arbitrary acoustic BCs. Finally, the SDS matrices for cavities containing or bounded by porous materials are formulated in an analytical manner. It is demonstrated that the method exhibits a much higher computational efficiency over the FE package COMSOL, at least 6 times faster than the COMSOL with a similar level of accuracy, and benchmark solutions are provided. This promising method can provide a powerful tool for systems with porous materials that have frequency-dependent characteristics, paving the way for efficient and accurate acoustic analysis in complex engineering applications.

Hybrid Laplace-time domain model for coupled dynamic analysis of multibody mating system in topside float-over installations

The mating operation is the most critical phase during the topside float-over installation for jacket platforms, involving complex load transfer process and nonlinear couplings among the topside, float-over barge, jacket, and other buffering devices. It is very important and yet challenging to accurately and efficiently capture the coupled dynamics of the multibody mating system, as the time domain (TD) model requires a sufficiently small time interval to clearly capture the strong impacts on the buffering devices, resulting in a worse computational efficiency. This paper innovatively proposes a hybrid Laplace-time domain (HLTD) model to improve the computational efficiency in the coupled dynamic analysis while the computational accuracy is remained. The HLTD model computes the coupled dynamics by the time-domain iterative operations, mixing the Laplace-domain pole-residue operations for computing the motion responses in each iteration. Since the tedious differential and integral solutions in the TD model are replaced by the simple algebraic pole-residue calculations, the HLTD model has much higher computational efficiency with the same computational accuracy as the TD model. By comparing with the commercial software OrcaFlex, the proposed HLTD model shows satisfactory accuracy and superior efficiency in the heave-only mating analysis during the topside float-over installation of a jacket platform.

Resolving severely overlapping ion mobility peaks using enhanced Fourier self-deconvolution.

Fourier self-deconvolution is an effective method for resolving overlapping spectra. However, the selection of the half-width for the deconvolving function is often subjective, which can lead to either excessive convolution or insufficient resolution enhancement. Additionally, ion mobility peaks exhibit tailing effects, which may be misinterpreted as new peaks when the deconvolving function is modelled with a Gaussian function. This paper proposes an improved Fourier self-deconvolution method based on continuous wavelet transform. The proposed method determines the signal half-width by calculating the horizontal distance between the peaks and troughs of the wavelet coefficients, offering a more accurate estimation. Furthermore, an asymmetric function is employed to optimize the peak shape of the deconvolving function, significantly reducing the likelihood of peak misidentification. The effectiveness of the proposed method is validated using both simulated and experimental ion mobility spectrometry data. Experimental results demonstrate that the proposed method effectively enhances peak resolution and resolves overlapping peaks. Moreover, compared with other peak segmentation algorithms based on Fourier self-deconvolution, the proposed method demonstrates lower parameter estimation error and higher computational efficiency, particularly for severely overlapping peaks.

Data-driven parametrization of molecular mechanics force fields for expansive chemical space coverage.

A force field is a critical component in molecular dynamics simulations for computational drug discovery. It must achieve high accuracy within the constraints of molecular mechanics' (MM) limited functional forms, which offers high computational efficiency. With the rapid expansion of synthetically accessible chemical space, traditional look-up table approaches face significant challenges. In this study, we address this issue using a modern data-driven approach, developing ByteFF, an Amber-compatible force field for drug-like molecules. To create ByteFF, we generated an expansive and highly diverse molecular dataset at the B3LYP-D3(BJ)/DZVP level of theory. This dataset includes 2.4 million optimized molecular fragment geometries with analytical Hessian matrices, along with 3.2 million torsion profiles. We then trained an edge-augmented, symmetry-preserving molecular graph neural network (GNN) on this dataset, employing a carefully optimized training strategy. Our model predicts all bonded and non-bonded MM force field parameters for drug-like molecules simultaneously across a broad chemical space. ByteFF demonstrates state-of-the-art performance on various benchmark datasets, excelling in predicting relaxed geometries, torsional energy profiles, and conformational energies and forces. Its exceptional accuracy and expansive chemical space coverage make ByteFF a valuable tool for multiple stages of computational drug discovery.

Numerical investigation of angle parameters of tip sails for improvement of aircraft flight performance

Inspired by the tip feathers of soaring birds, the device of tip sails was designed and installed on the wing to achieve drag reduction. The influences of design parameters including cant angles, toe angles and twist angles on aerodynamic characteristics of tip sails were investigated through 77 cases of 11 configurations at 7 different angles of attack. According to the results grouped, the optimal cant angles ( 30 , 15 , 0 ) were determined. By adjusting the toe angles, drag reduction can be improved under different flight conditions. The decreasing twist angles can inhibit flow separation on sail surface. The optimal toe angles ( 18 , 15 , 12 ) and twist angles ( 12 , 9 , 6 ) resulted in the weakest flow separation. Ensuring that the twist angle of each sail is 6 ∘ less than that of the corresponding toe angle can effectively inhibit flow separation on the sail surface. The tip sails can also split and weaken the wingtip vortex. In addition, in the validation of the numerical method, we also investigate the influence of numerical schemes and turbulence models on the results of wingtip vortex flow. In the cases with unstructured meshes, the third-order finite volume scheme had higher computational efficiency than the traditional second-order finite volume scheme in obtaining grid independent solutions, and the turbulence model with rotation correction can improve the calculation accuracy of the wingtip vortex flow. The results of this paper provide a valuable reference for the design of tip sails.

A MODIFIED NUMERICAL METHOD OF DETERMINING HYDRAULIC SYSTEM PARAMETERS FOR THE DEVELOPMENT OF MATHEMATICAL MODELS AND INFORMATION COMPLEXES OF COMPUTER SIMULATION MODELS FOR INDUSTRIAL CHEMICAL PRODUCTION

The paper introduces a modified numerical method for determining the parameters of hydraulic systems used in mathematical modeling and simulation-based computer systems for industrial chemical production. It explores the dynamics of fluids, gases, and their mixtures as they traverse pipelines equipped with valves, compressors, and other complex components. Traditional methods for solving nonlinear equation systems often face limitations, including sensitivity to initial approximations and the need for relaxation coefficient adjustments, which complicate simulation processes. The proposed method converts nonlinear equations into an optimization problem, us- ing the Nelder-Mead algorithm to minimize deviations between initial estimates and comput- ed parameters. This derivative-free approach employs a simplex geometric transformation, overcoming issues associated with iterative methods and delivering high accuracy, stable convergence, and computational efficiency. Its adaptability makes it suitable for handling in- tricate hydraulic systems. To validate the method, a practical example involving a pipeline system with multiple restricting devices is analyzed. The mathematical model includes equations for pressures and flow rates, supplemented by material balance constraints at branching points. Numerical ex- periments evaluate how input pressure influences system parameters, demonstrating the mod- el's robustness and precision. The results confirm that the method accurately simulates system parameters under steady-state conditions and identifies critical optimization points. This innovative approach offers significant advantages for designing and managing industrial chemical facilities. By simplifying model construction, enhancing reliability, and accelerating simulation processes, the method provides a powerful tool for optimizing the op- eration and control of complex technological systems.

Modified Particle Swarm Optimization on Feature Selection for Palm Leaf Disease Classification

Palm oil plantations in Indonesia face challenges in enhancing productivity and profitability, notably due to pest attacks that reduce production. Early identification and classification of plant conditions, particularly palm oil leaves, are crucial for mitigating losses. This study explores the application of artificial intelligence, specifically computer vision and machine learning, for disease detection. Various machine learning techniques, including Local Binary Pattern (LBP), K-Nearest Neighbors (KNN), and Support Vector Machine (SVM), have been used in different studies with varying accuracy. This research focuses on modifying Particle Swarm Optimization (PSO) for feature selection in identifying diseases in palm oil leaves. The PSO modification combined with logistic regression and Bayesian Information Criterion (BIC) significantly enhances KNN performance. Accuracy improved from 95.75% to 97.85%, while precision, recall, and F1-score reached approximately 98.80%. Additionally, the modified KNN+PSO achieved the shortest computation time of 0.0872 seconds, indicating high computational efficiency. These results demonstrate that the PSO modification not only improves accuracy but also computational efficiency, making it an effective method for enhancing KNN performance in detecting palm oil leaf diseases.

Extreme learning machine for identifying soil-dwelling microorganisms cultivated on agar media

The aim of this research is to create an automated system for identifying soil microorganisms at the genera level based on raw microscopic images of monocultural colonies grown in laboratory environment. The examined genera are: Fusarium, Trichoderma, Verticillium, Purpureolicillium and Phytophthora. The proposed pipeline deals with unprocessed microscopic images, avoiding additional sample marking or coloration. The methodology includes several stages: image preprocessing, segmenting images to isolate microorganisms from the background, calculating features related to image color and texture for classification. Using an extensive dataset of 2866 images from the National Institute of Horticultural Research in Skierniewice the Extreme Learning Machine model was trained and validated. The model showcases high accuracy and computational efficiency compared to other Machine Learning state-of-the art methods e.g. CatBoost, Random Forest or Convolutional Neural Networks. Statistical techniques, including Multivariate Analysis of Variance were employed to confirm significant differences among the datasets, enhancing the model’s robustness. Nevertheless, Shapley Additive Explanations values provided transparency into the model’s decision-making process. This approach has the potential to improve early detection and management of soil pathogens, promoting sustainable agriculture and demonstrating machine learning’s potential in environmental monitoring, microbial ecology or industrial microbiology.

An Improved Single‐Layer Smoothed Particle Hydrodynamics Model for Water–Soil Two‐Phase Flow

ABSTRACTIn coastal and offshore engineering, the intense water–soil motion poses significant challenges to the safety of buildings and structures. The smoothed particle hydrodynamics (SPH) method, as a mesh‐free Lagrangian solver, has considerable advantages in the numerical resolution of such problems. SPH models for the water–soil two‐phase flow can be categorized into the multilayer type and the single‐layer type. Although the single‐layer model envisions a simpler algorithm and higher computational efficiency, its accuracy, stability, and recovery of interfacial details are far from satisfactory. In the present work, an improved single‐layer model is established to alleviate these limitations. First, the soakage function, which takes effect near the phase interface, is introduced to characterize the two‐phase coupling status. Additionally, the stress diffusion term and a modified density diffusion term applicable in density discontinuity scenario are introduced to ease the numerical oscillation. Finally, to remove the unphysical voids in the interfacial region, the particle shifting technique with special treatment tailored for free‐surface particles is implemented. Validations of the proposed model are carried out by a number of numerical tests, including the erodible dam‐break problem, the wall‐jet scouring, the flushing case, and the water jet excavation. Appealing agreements with either experimental data or published numerical results have been achieved, which verifies the accuracy, stability, and robustness of the proposed model for water–soil two‐phase flows.

Cooperative Control of Distributed Drive Electric Vehicles for Handling, Stability, and Energy Efficiency, via ARS and DYC

Distributed drive electric vehicles (DDEV), characterized by their independently drivable wheels, offer significant advantages in terms of vehicle handling, stability, and energy efficiency. These attributes collectively contribute to enhancing driving safety and extending the all-electric range for sustainable transportation. Nonetheless, the challenge persists in designing a control strategy that effectively coordinates the objectives of handling, stability, and energy efficiency under both lateral and longitudinal driving conditions. To this end, this paper proposes a cooperative control strategy for DDEVs, incorporating active rear steering (ARS) and direct yaw moment control (DYC) to enhance handling capabilities, stability, and energy efficiency. A stability boundary is delineated using an analytical expression that correlates with the front wheel steering angle, and an adjustment factor is introduced to quantify vehicle stability based on this input parameter. This factor aids in establishing a coordinated control reference for handling and stability. At the upper-level motion control layer, a model predictive control method is developed to track this reference and implement ARS and DYC for superior performance. Specifically, the rear lateral force serves as the control command for ARS, which is converted into a rear wheel steering angle using a tire inverse model. Meanwhile, the front lateral force is modeled as linear-time-varying to simplify calculations. At the lower-level torque allocation layer, the adjustment factor is utilized to balance tire workload rate and in-wheel motors’ (IWM) energy consumption, enabling efficient switching between energy consumption and driving stability targets, and the torque allocation is conducted to acquire the expected IWMs’ command. Both the upper and lower-level optimization problems are formulated as convex problems, ensuring efficient and effective solutions. Simulations verify the effectiveness of this strategy in improving handling, stability, and energy economy under DLC cases, while maintaining high computational efficiency.

Parking Trajectory Planning for Autonomous Vehicles Under Narrow Terminal Constraints

Trajectory planning in tight spaces presents a significant challenge due to the complex maneuvering required under kinematic and obstacle avoidance constraints. When obstacles are densely distributed near the target state, the limited connectivity between the feasible states and terminal state can further decrease the efficiency and success rate of trajectory planning. To address this challenge, we propose a novel Dual-Stage Motion Pattern Tree (DS-MPT) algorithm. DS-MPT decomposes the trajectory generation process into two stages: merging and posture adjustment. Each stage utilizes specific heuristic information to guide the construction of the trajectory tree. Our experimental results demonstrate the high robustness and computational efficiency of the proposed method in various parallel parking scenarios. Additionally, we introduce an enhanced driving corridor generation strategy for trajectory optimization, reducing computation time by 54% to 84% compared to traditional methods. Further experiments validate the improved stability and success rate of our approach.

Classification of diabetic retinopathy grades using CNN feature extraction to segment the lesion

Diabetes's microvascular aftereffect, diabetic retinopathy (DR), is the primary cause of eyesight loss in the globe. In order to prevent vision impairment and to intervene promptly, early detection and precise classification of DR severity are essential. Using standard methods for diagnosing DR requires ophthalmologists to grade cases by hand, a process that can be laborious, subjective, and subject to observer error. In supervised learning task of classification, data instances are classified into predefined classes based on features. The relation between the traits and the classes can be found from the labelled data. After the training is completed, the classes of the unseen data. The frequent reason found for the loss of vision in diabetic retinopathy (DR) is found to be diabetes. Visual damage can be prevented by identifying the degree of DR at right time. For the grading of the DR, deep learning techniques are found to be very effective with maximum possible accuracy. The proposed model is useful in accurately classifying the DR images using the feature extraction with lesion segmentation, by implementing the patterns in the DR images. ReLU activation function is used in the proposed model. CNN feature extraction is used for the important feature extraction by applying the Convolution layers, and edges, textures, and forms are identified. As the model proceeds layer by layer, complicated patterns in the photos can be learned by the model, and can be analysed better. The features of the photos were extracted and found useful in segmentation and classification. ReLU is helpful in improving the convergence and also found useful in learning the patterns. Among the other activation functions, ReLU has higher computational efficiency and therefore is used in the model, which suits well for the DR application. A strong framework is proposed for the classification of the DR grade, for the lesion segmentation and CNN feature extraction. DR categorization using the proposed model is evaluated by data visualization of the important calculated metrics and found to be very effective.

A Review of the Current Status and Prospects of Improving Indoor Environment for Lightweight Buildings in High-Altitude Cold Regions

Lightweight structures, characterized by rapid assembly, are vital for creating habitats in outdoor environments, but their implementation in high-plateau cold regions encounters significant challenges in heating and ventilation. This paper systematically introduces the environmental characteristics and reviews the demands and primary influencing factors of indoor environments in these regions. The advantages and limitations of underground lightweight construction are also discussed. Current research indicates that evaluation methods for air quality in high-altitude cold regions require further development. Reducing building heat loss and minimizing cold air infiltration can enhance indoor environments and lower energy consumption. However, it is essential to establish effective ventilation strategies to prevent the accumulation of air pollutants. Then, potential passive ventilation improvement measures suitable for the environmental characteristics of high-cold plateaus are outlined. The application potential and possible limitations of these measures are summarized, providing references for future research. Finally, the main research methods for ventilation and heating within building interiors are organized and discussed. Findings indicate that computational fluid dynamics models are predominantly used, but they demonstrate low efficiency and high resource consumption for medium- to large-scale applications. Integrating these models with network models can achieve a balance of high computational accuracy and efficiency.

Predicting House Prices with a Linear Regression Model

House price forecasting is an important area of economic and social research. Among the many house price forecasting methods, the linear regression model is widely used because of its simplicity, easy interpretation, and high computational efficiency. This paper aims to investigate the effectiveness of linear regression models in house price forecasting. This paper will first introduce the basic theory of linear regression model, and discuss the factors that affect the housing price, then build and evaluate the housing price prediction model, and then verify the constructed data model through real data. Finally, we discuss the accuracy of the prediction, analyze the results of the passing model, and find that housing prices can be predicted more accurately for cases where the variables are relatively simple and differentiated, such as the ownership of specific facilities. However, linear regression prediction still has some defects, and there are more effective and general methods for housing price prediction to solve the problems that linear regression method cannot pay attention to, which will be the subject of future research.

Solution of Non-Autonomous Bloch Equation via Multistage Differential Transformation Method

This study explores the multistage differential transformation method (MsDTM) as an efficient approach for solving non-autonomous differential equations. The proposed method demonstrates wide-ranging applicability in fields such as computer graphics, quantum optics, biomathematics, and image processing. Specifically, the Bloch equations, which describe the interaction of a spin-1/2 system (or two-level atom) with a mono- or bichromatic laser field in the presence of an off-resonant broad-band squeezed vacuum (SV), are examined. As a system of non-autonomous ordinary differential equations, the Bloch model captures the quantum dynamics between matter and electromagnetic fields, offering insights into more complex and experimentally relevant models. The MsDTM is employed to obtain numerical solutions with high precision and computational efficiency, outperforming the classical 4th-order Runge-Kutta (RK4) method. A key advantage of the MsDTM is its adaptability; its accuracy can be further enhanced by either increasing the number of iterations or refining the time-step in the numerical scheme. Consequently, MsDTM emerges as a robust tool for computing solutions to a broad class of non-autonomous equations.

A Multi-Objective Optimization Design Method for High-Aspect-Ratio Wing Structures Based on Mind Evolution Algorithm Backpropagation Surrogate Model

The design of high-aspect-ratio wings enhances the flight efficiency of UAVs but also introduces significant aeroelasticity challenges. The efficient optimization of wing structures in complex environments has become critical. To address the current challenges in balancing wing strength with lightweight structural designs, this study proposed an intelligent solution method for optimizing wing dimensions and structural layout. Driven by mechanical simulation data, the method established a mapping relationship between the structural layout and dimensions of the wing and its bending stiffness. This approach was further enhanced by the mind evolution algorithm (MEA) to optimize the solution performance of the surrogate model. The wing structure optimization model was established using the multi-objective grey wolf optimizer (MOGWO) based on the surrogate model for search and optimization. This study focused on the composite material wing of a long-endurance unmanned aerial vehicle (UAV). The established MEA-BP surrogate model demonstrated high computational efficiency, with the prediction error standard deviation (STD) of wing deflection not exceeding 0.495 mm. The optimization model required 175 s to calculate the Pareto front solutions. The optimized structure resulted in a 28.32% increase in wing equivalent stiffness, and weight only increased by 6.67% compared to the original structure. These results showcased the effectiveness of the proposed method and validated the feasibility of integrating intelligent optimization algorithms and machine learning in the field of aircraft design.

A dual-decoder banded convolutional attention network for bone segmentation in ultrasound images.

Ultrasound (US) has great potential for application in computer-assisted orthopedic surgery (CAOS) due to its non-radiative, cost-effective, and portable traits. However, bone segmentation from low-quality US images has been challenging. Traditional segmentation methods cannot achieve satisfactory results due to their high customization and dependence on bone morphology. Existing deep learning-based methods make it difficult to ensure efficient and accurate segmentation due to the ignorance of prior knowledge of bone features during feature learning. This paper aims to systematically investigate feature extraction and segmentation methodologies of bone US images and then proposes an innovative convolutional neural network to address the need for precise and efficient bone structure extraction in CAOS. This paper has proposed a dual-decoder banded convolutional attention network (BCA-Net), which takes the raw US image as input and simplified U-Net as the baseline network. Multiscale banded convolution kernels are employed internally in the BCA-Net model, leveraging the prior knowledge that bone surfaces in US images are exhibited as bright bands of a few millimeters in width. Additionally, a shared encoder to extract input features and two independent decoders to generate outputs for the bone surface and bone shadow mask are integrated into the BCA-Net model, leveraging the prior knowledge that US bone surfaces manifest low-intensity hollow shadows below. Then, a new task consistency loss is introduced that can utilize inter-task dependency fully and enhance the performance of our model. In the network construction process, a dataset containing 1623 sets of US images was adopted, and a five-fold cross-validation strategy was divided into the training and validation sets for the model's training and validation. Many vital metrics were introduced to comprehensively evaluate the model performance, including overlap, edge distance, area under curve, and model efficiency. Finally, the model performance was subjected to a confidence interval, Tukey's honest significant difference, and Cohen's d statistics at a significance level (5%) to ensure the accuracy and reliability of the obtained findings. The experimental results show that the BCA-Net model performs well in the bone surface segmentation task. Its average Dice coefficient reaches 87.51%, 4.04% higher than U-Net's, proving its superior bone surface segmentation accuracy. Meanwhile, the average distance error is 0.2mm, 0.33mm lower than U-Net's, highlighting its accuracy in detail capture and boundary recognition. Using a confidence distance threshold of 1.02mm, the Dice coefficient of the BCA-Net model exceeds 98%, an improvement of 1.87% over U-Net's, which is highly consistent with manual labeling. The BCA-Net model achieves a statistical significance of p-values<0.05 in the above accuracy comparisons. In addition, the BCA-Net model has a small parameter count (5.58M) and high computational efficiency (35.85 frames per second), further validating its excellent potential in bone surface segmentation tasks. The proposed method achieves excellent performance with high accuracy and efficiency, aligning well with clinical requirements and holding excellent potential for advancing the utilization of US images in CAOS.

A hybrid statistical–dynamical framework for compound coastal flooding analysis

Abstract Compound coastal flooding due to astronomic, atmospheric, oceanographic, and hydrologic forcings poses severe threats to coastal communities. While physics-driven approaches are able to dynamically simulate temporally and spatially varying compound flooding generated by multiple partially correlated drivers, computational burdens limit their capability to explore the full range of conditions that contribute to compound coastal hazards. Data-driven statistical approaches address some of these computational challenges, however they are also unable to explore all possible forcing combinations due to short observational records, and projections are typically limited to a few locations. This study proposes a hybrid statistical-dynamical framework for compound coastal flooding analysis that integrates a stochastic generator of compound flooding drivers, a hydrodynamic model, and a machine learning-based surrogate model. The framework is demonstrated in San Francisco (SF) Bay over the past 500 years with high accuracy and computational efficiency. The stochastic generator of compound flooding drivers is developed by coupling a sea surface temperature (SST) reconstruction model with a climate emulator, weather generator, and hydrologic model. Using reconstructed SSTs as input, the generator of compound flooding drivers is employed to simulate time series of the forcing factors contributing to compound flooding in SF Bay. A process-based hydrodynamic model is built to predict total water levels (TWLs) varying in time and space throughout SF Bay based on the stochastically generated drivers. A machine learning-based surrogate model is then developed from a relatively small library of hydrodynamic model simulations to efficiently predict water levels (WLs) for compound flooding analysis under the full range of stochastic drivers. This study contributes a hybrid statistical-dynamical framework to better understand the spatial distribution and temporal evolution of compound coastal flooding, along with the relative contributions of the forcing drivers in complex nearshore, estuarine, and river environments for centennial timescales under past, present, and future climates.