Dynamic Integration Research Articles

TMODINET: A trustworthy multi-omics dynamic learning integration network for cancer diagnostic

Multiple types of omics data contain a wealth of biomedical information which reflect different aspects of clinical samples. Multi-omics integrated analysis is more likely to lead to more accurate clinical decisions. Existing cancer diagnostic methods based on multi-omics data integration mainly focus on the classification accuracy of the model, while neglecting the interpretability of the internal mechanism and the reliability of the results, which are crucial in specific domains such as precision medicine and the life sciences. To overcome this limitation, we propose a trustworthy multi-omics dynamic learning framework (TMODINET) for cancer diagnostic. The framework employs multi-omics adaptive dynamic learning to process each sample to provide patient-centered personality diagnosis by using self-attentional learning of features and modalities. To characterize the correlation between samples well, we introduce a graph dynamic learning method which can adaptively adjust the graph structure according to the specific classification results for specific graph convolutional networks (GCN) learning. Moreover, we utilize an uncertainty mechanism by employing Dirichlet distribution and Dempster-Shafer theory to obtain uncertainty and integrate multi-omics data at the decision level, ensuring trustworthy for cancer diagnosis. Extensive experiments on four real-world multimodal medical datasets are conducted. Compared to state-of-the-art methods, the superior performance and trustworthiness of our proposed algorithm are clearly validated. Our model has great potential for clinical diagnosis.

Tutorial on Molecular Latent Space Simulators (LSSs): Spatially and Temporally Continuous Data-Driven Surrogate Dynamical Models of Molecular Systems.

The inherently serial nature and requirement for short integration time steps in the numerical integration of molecular dynamics (MD) calculations place strong limitations on the accessible simulation time scales and statistical uncertainties in sampling slowly relaxing dynamical modes and rare events. Molecular latent space simulators (LSSs) are a data-driven approach to learning a surrogate dynamical model of the molecular system from modest MD training trajectories that can generate synthetic trajectories at a fraction of the computational cost. The training data may comprise single long trajectories or multiple short, discontinuous trajectories collected over, for example, distributed computing resources. Provided the training data provide sufficient sampling of the relevant thermodynamic states and dynamical transitions to robustly learn the underlying microscopic propagator, an LSS furnishes a global model of the dynamics capable of producing temporally and spatially continuous molecular trajectories. Trained LSS models have produced simulation trajectories at up to 6 orders of magnitude lower cost than standard MD to enable dense sampling of molecular phase space and large reduction of the statistical errors in structural, thermodynamic, and kinetic observables. The LSS employs three deep learning architectures to solve three independent learning problems over the training data: (i) an encoding of the high-dimensional MD into a low-dimensional slow latent space using state-free reversible VAMPnets (SRVs), (ii) a propagator of the microscopic dynamics within the low-dimensional latent space using mixture density networks (MDNs), and (iii) a generative decoding of the low-dimensional latent coordinates back to the original high-dimensional molecular configuration space using conditional Wasserstein generative adversarial networks (cWGANs) or denoising diffusion probability models (DDPMs). In this software tutorial, we introduce the mathematical and numerical background and theory of LSS and present example applications of a user-friendly Python package software implementation to alanine dipeptide and a 28-residue beta-beta-alpha (BBA) protein within simple Google Colab notebooks.

Dissociable effects of urgency and evidence accumulation during reaching revealed by dynamic multisensory integration.

When making perceptual decisions, humans combine information across sensory modalities dependent on their respective uncertainties. However, it remains unknown how the brain integrates multisensory feedback during movement, and which factors besides sensory uncertainty influence sensory contributions. We performed two reaching experiments on healthy adults to investigate whether movement corrections to combined visual and mechanical perturbations scale with visual uncertainty. To describe the dynamics of multimodal feedback responses, we further varied movement time and visual feedback duration during the movement. The results of our first experiment show that the contribution of visual feedback decreased with uncertainty. Additionally, we observed a transient phase during which visual feedback responses were stronger during faster movements. In a follow-up experiment, we found that the contribution of vision increased more quickly during slow movements when we presented the visual feedback for a longer time. Muscle activity corresponding to these visual responses exhibited modulations with sensory uncertainty and movement speed ca. 100ms following the onset of the visual feedback. Using an optimal feedback control model, we show that the increased response to visual feedback during fast movements can be explained by an urgency-dependent increase in control gains. Further, the fact that a longer viewing duration increased the visual contributions suggests that the brain accumulates sensory information over time to estimate the state of the arm during reaching. Our results provide additional evidence concerning the link between reaching control and decision-making, both of which appear to be influenced by sensory evidence accumulation and response urgency.Significance statement The time-course of multisensory integration during movement, along with the factors influencing this process, still requires further investigation. Here, we tested how visual uncertainty, movement speed, and visual feedback duration influence reach corrections to combined visual and mechanical perturbations. Using an optimal feedback control model, we illustrate that the time-course of multimodal corrections follows the predictions of a Kalman filter which continuously weighs sensory feedback and internal predictions according to their reliability. Importantly, we further show that changes in movement speed led to urgency-dependent modulations of control gains. Our results corroborate previous research linking motor control and decision-making by highlighting that multisensory feedback responses depend on sensory evidence accumulation and response urgency in a similar way as decision-making processes.

Spatiotemporal integration of contextual and sensory information within the cortical hierarchy in human pain experience.

Pain is not a mere reflection of noxious input. Rather, it is constructed through the dynamic integration of current predictions with incoming sensory input. However, the temporal dynamics of the behavioral and neural processes underpinning this integration remain elusive. In the current study involving 59 human participants, we identified a series of brain mediators that integrated cue-induced expectations with noxious inputs into ongoing pain predictions using a semicircular scale designed to capture rating trajectories. Temporal mediation analysis revealed that during the early-to-mid stages of integration, the frontoparietal and dorsal attention network regions, such as the lateral prefrontal, premotor, and parietal cortex, mediated the cue effects. Conversely, during the mid-to-late stages of integration, the somatomotor network regions mediated the effects of stimulus intensity, suggesting that the integration occurs along the cortical hierarchy from the association to sensorimotor brain systems. Our findings advance the understanding of how the brain integrates contextual and sensory information into pain experience over time.

The Effect of Bioalcohol Additives on Biofuel Diesel Engines

This study experimentally investigated a water-cooled four-cylinder turbocharged diesel engine (DE) under different loads and fuel blend ratios. The integration of Computational Fluid Dynamics (CFD) simulations enables a deeper analysis of the combustion process. Through an in-depth analysis of the combustion process, the focus was placed on investigating the specific impacts of ethanol and n-butanol additives on diesel engine performance. Research shows that a fuel mixture consisting of 70% diesel, 10% biodiesel, and 20% ethanol reduced NOx emissions by 5.56% compared to pure diesel at 75% load. Furthermore, this study explores the combustion performance of diesel/biodiesel blended with butanol/ethanol. The findings indicate that n-butanol improves thermal efficiency, particularly at 100% load, with the D70B10E20 and D70B10BU20 blends demonstrating thermal efficiencies of 9.94%and 8.72% higher than that of diesel alone, respectively. All mixed fuels exhibited reduced hydrocarbon and CO emissions under different loads, with a notable reduction in hydrocarbon emissions of 34.4% to 46.1% at 75% load.

Comparative Analysis of Chatbots Using Large Language Models for Web Development Tasks

In this study, we compare the performance of five chatbots using large language models (LLMs) in handling web development tasks. Three human testers asked each chatbot nine predefined questions related to creating a simple website with a dynamic form and database integration. The questions covered tasks such as generating a web document structure, designing a layout, creating a form, and implementing database queries. The chatbots’ outputs were ranked based on accuracy, completeness, creativity, and security. The experiment reveals that conversational chatbots are adept at managing complex tasks, while programming assistants require more precisely formulated tasks or the ability to generate new responses to address irrelevant outputs. The findings suggest that conversational chatbots are more capable of handling a broader range of web development tasks with minimal supervision, whereas programming assistants need more precise task definitions to achieve comparable results. This study contributes to understanding the strengths and limitations of various LLM-based chatbots in practical coding scenarios, offering insights for their application in web development.

Experimental study on the dynamic integration of PCM in building walls for enhancing thermal performance in summer condition

Passively integrating phase change material into the walls to enhance the thermal performance of the building has been a promising solution in recent years. As the PCM has a high latent heat capacity, it leads to damping the high variations of temperature and provides an obvious benefit in energy saving and indoor comfort. However, the traditional passive integration method of the PCM limited the utilization of the PCM. The thermal resistance between the PCM and the indoors restrains the thermal response of the PCM, and it between the PCM and the outdoors reduces the impact of outdoor heating or cooling on the PCM. In this study, a dynamic PCM integration in the building envelope method was proposed and experimentally investigated. A PCM layer and an air layer were combined to be integrated into the wall assembly, which allows the position of the PCM layer and the air layer could be changed to adjust the thermal resistance (air layer) between the PCM and indoors and outdoors. The results showed that this dynamic method can dramatically reduce the indoor temperature and the heat flux across the interior surface of the wall. Compared to the envelope with only static PCM layer configurations, the dynamic PCM provided a reduction of 9.1% in the indoor average temperature and a reduction of 116.0% in the peak heat flux during the experiment's three days, as well as the dynamic PCM, exploited more latent heat than the other static configurations. Considering the energy performance, the dynamic integration of PCM showed a 100% reduction in heat gain through the interior surface compared to the envelope with only a static PCM layer under summer conditions.

Dynamic responses of monopile offshore wind turbines under different wind-wave misalignment angles

Wind-wave misalignment is a common occurrence for monopile offshore wind turbines during extreme typhoon conditions. This phenomenon has a significant impact on the structural dynamic response of turbine structures and poses a threat to their structural integrity. In this study, finite element analysis is conducted on 75 sets of working conditions of monopile offshore wind turbines to investigate their dynamic response under different wind turbine operating states, wind-wave misalignment angles, and randomness of wind and wave load conditions. The results of the analysis indicate that the impact of wind-wave misalignment angles on the dynamic response of the monopile offshore wind turbine varies depending on the operational conditions. Furthermore, the study reveals that there are variations in the dynamic response of the monopile offshore wind turbine under different combinations of wind and wave loads. During the parked state and yaw system fault of the monopile offshore wind turbine, the most adverse working conditions need to consider wind-wave misalignment angles of 0° and 180°. In contrast, the most adverse working conditions for operational states emerge when the wind-wave misalignment angle is 90°. These findings highlight the importance of considering wind-wave misalignment angles and the randomness of wind and wave load conditions when assessing the dynamic response and structural integrity of monopile offshore wind turbines.

Integrating active demand into the distribution system using metaheuristic techniques

AbstractIntegrating non‐conventional renewable energy sources into distribution systems, alongside data science and enabling technological infrastructures, presents significant challenges, particularly in managing active demand. The rapid evolution of the electric energy system and increasing electricity demand highlight the need for reliable tracking and predictive methods to manage Distributed Energy Resources and digital infrastructure. These methods are essential for advancing carbon neutrality, democratizing environmental sustainability, and improving energy efficiency. Effective active demand monitoring requires understanding the transactional system concept, including digital infrastructure and decentralized demand. Although metaheuristic techniques are increasingly important in demand response integration, much research focuses on specific techniques rather than providing a comprehensive view of dynamic transaction integration for active demand. Technological advancements, like smart meters and communication systems, are shifting from basic consumption measurement to active customer participation. This article reviews key concepts in electrical distribution systems, such as active demand, DERs, and transactive systems. It examines prevalent metaheuristic techniques, emphasizing their role in integrating and predicting active demand and DER behaviors. Additionally, the study presents a methodology serving as a roadmap for efficient DER integration and the transition to active demand and transactive electricity systems, addressing gaps in the current literature.

Decoding aging and cognitive functioning through spatiotemporal EEG patterns: Introducing spatiotemporal information-based similarity analysis.

Exploring spatiotemporal patterns of high-dimensional electroencephalography (EEG) time series generated from complex brain system is crucial for deciphering aging and cognitive functioning. Analyzing high-dimensional EEG series poses challenges, particularly when employing distance-based methods for spatiotemporal dynamics. Therefore, we proposed an innovative methodology for multi-channel EEG data, termed as Spatiotemporal Information-based Similarity (STIBS) analysis. The core of this method is to first perform state space compression of multi-channel EEG time series using global field power, which can provide insight into the dynamic integration of spatiotemporal patterns between the steady states and non-steady states of brain. Subsequently, we quantify the pairwise differences and non-randomness of spatiotemporal patterns using an information-based similarity analysis. Results demonstrated that this method holds the potential to serve as a distinguishing marker between young and elderly on both pairwise differences and non-randomness indices. Young individuals and those with higher cognitive abilities exhibit more complex macrostructure and non-random spatiotemporal patterns, whereas both aging and cognitive decline lead to more randomized spatiotemporal patterns. We further extended the proposed analytics to brain regions adversarial STIBS (bra-STIBS), highlighting differences between young and elderly, as well as high and low cognitive groups. Furthermore, utilizing the STIBS-based XGBoost model yields superior recognition accuracy in aging (93.05%) and cognitive functioning (74.29%, 64.19%, and 80.28%, respectively, for attention, memory, and compatibility performance recognition). STIBS-based methodology not only contributes to the ongoing exploration of neurobiological changes in aging but also provides a powerful tool for characterizing the spatiotemporal nonlinear dynamics of the brain and their implications for cognitive functioning.

Artificial Neural Network-Based Data-Driven Parameter Estimation Approach: Applications in PMDC Motors

The optimal performance of direct current (DC) motors is intrinsically linked to their mathematical models’ precision and their controllers’ effectiveness. However, the limited availability of motor characteristic information poses significant challenges to achieving accurate modeling and robust control. This study introduces an approach employing artificial neural networks (ANNs) to estimate critical DC motor parameters by defining practical constraints that simplify the estimation process. A mathematical model was introduced for optimal parameter estimation, and two advanced learning algorithms were proposed to efficiently train the ANN. The performance of the algorithms was thoroughly analyzed using metrics such as the mean squared error, epoch count, and execution time to ensure the reliability of dynamic priority arbitration and data integrity. Dynamic priority arbitration involves automatically assigning tasks in real-time depending on their relevance for smooth operations, whereas data integrity ensures that information remains accurate, consistent, and reliable throughout the entire process. The ANN-based estimator successfully predicts electromechanical and electrical characteristics, such as back-EMF, moment of inertia, viscous friction coefficient, armature inductance, and armature resistance. Compared to conventional methods, which are often resource-intensive and time-consuming, the proposed solution offers superior accuracy, significantly reduced estimation time, and lower computational costs. The simulation results validated the effectiveness of the proposed ANN under diverse real-world operating conditions, making it a powerful tool for enhancing DC motor performance with practical applications in industrial automation and control systems.

Optimal Configuration Model for Large Capacity Synchronous Condenser Considering Transient Voltage Stability in Multiple UHV DC Receiving End Grids

In a multi-fed DC environment, the UHV DC recipient grid faces significant challenges related to DC phase shift failure and voltage instability due to the high AC/DC coupling strength and low system inertia level. While the new large-capacity synchronous condensers (SCs) can provide effective transient reactive power support, the associated investment and operation costs are high. Therefore, it is valuable to investigate the optimization of SC configuration at key nodes in the recipient grid in a scientific and rational manner. This study begins by qualitatively and quantitatively analyzing the dynamic characteristics of DC reactive power and induction motors under AC faults. The sub-transient and transient reactive power output model is established to describe the SC output characteristics, elucidating the coupling relationship between the SC’s reactive power output and the DC reactive power demand at different time scales. Subsequently, a critical stabilized voltage index for dynamic loads is defined, and the SC’s reactive power compensation target is quantitatively calculated across different time scales, revealing the impact of transient changes in DC reactive power on the transient voltage stability of the multi-fed DC environment with dynamic load integration. Finally, an optimal configuration model for the large-capacity SC is proposed under the critical stability constraint of dynamic loads to maximize the SC’s reactive power support capability at the lowest economic cost. The proposed model is validated in a multi-fed DC area, demonstrating that the optimal configuration scheme effectively addresses issues related to DC phase shift failures and voltage instability resulting from AC bus voltage drops.

Temporal regulation of gene expression through integration of p53 dynamics and modifications.

The master regulator of the DNA damage response, the transcription factor p53, orchestrates multiple downstream responses and coordinates repair processes. In response to double-strand DNA breaks, p53 exhibits pulses of expression, but how it achieves temporal coordination of downstream responses remains unclear. Here, we show that p53's posttranslational modification state is altered between its first and second pulses of expression. We show that acetylations at two sites, K373 and K382, were reduced in the second pulse, and these acetylations differentially affected p53 target genes, resulting in changes in gene expression programs over time. This interplay between dynamics and modification may offer a strategy for cellular hubs like p53 to temporally organize multiple processes in individual cells.

Spatial-temporal graph neural networks for groundwater data.

This paper introduces a novel application of spatial-temporal graph neural networks (ST-GNNs) to predict groundwater levels. Groundwater level prediction is inherently complex, influenced by various hydrological, meteorological, and anthropogenic factors. Traditional prediction models often struggle with the nonlinearity and non-stationary characteristics of groundwater data. Our study leverages the capabilities of ST-GNNs to address these challenges in the Overbetuwe area, Netherlands. We utilize a comprehensive dataset encompassing 395 groundwater level time series and auxiliary data such as precipitation, evaporation, river stages, and pumping well data. The graph-based framework of our ST-GNN model facilitates the integration of spatial interconnectivity and temporal dynamics, capturing the complex interactions within the groundwater system. Our modified Multivariate Time Graph Neural Network model shows significant improvements over traditional methods, particularly in handling missing data and forecasting future groundwater levels with minimal bias. The model's performance is rigorously evaluated when trained and applied with both synthetic and measured data, demonstrating superior accuracy and robustness in comparison to traditional numerical models in long-term forecasting. The study's findings highlight the potential of ST-GNNs in environmental modeling, offering a significant step forward in predictive modeling of groundwater levels.

A Stock Prediction Method Based on Multidimensional and Multilevel Feature Dynamic Fusion

Stock price prediction has long been a topic of interest in academia and the financial industry. Numerous factors influence stock prices, such as a company’s performance, industry development, national policies, and other macroeconomic factors. These factors are challenging to quantify, making predicting stock price movements difficult. This paper presents a novel deep neural network framework that leverages the dynamic fusion of multi-dimensional and multi-level features for stock price prediction, which means we utilize fundamental trading data and technical indicators as multi-dimensional data and local and global multi-level information. Firstly, the model dynamically assigns weights to multi-dimensional features of stocks to capture the impact of each feature on stock prices. Next, it applies the Fourier transform to the global features to capture the long-term trends of the global environment and dynamically fuses these with local and global features of the stocks to capture the overall market environment’s impact on individual stocks. Finally, temporal features are captured using an attention layer and an RNN-based model, which incorporates historical price data to forecast future prices. Experiments on stocks from various industries within the Chinese CSI 300 index reveal that the proposed model outperforms traditional methods and other deep learning approaches in terms of stock price prediction. This paper proposes a method that facilitates the dynamic integration of multi-dimensional and multi-level features in an efficient manner and experimental results show that it improves the accuracy of stock price predictions.

The Escol@s Digitais Project. An experience of working with technologies in Primary Education in Portugal

The Escol@s Digitais Project is an action-research project that was born on the initiative of the Municipality of Amadora (Portugal) and has been under development by a research team of the Institute of Education of the University of Lisbon. As part of the Education XXI line of research, it aims to support the ongoing digital transformation in all public schools of Primary Education in the municipality to enhance the quality of processes, performances, and results of learning. Seeking to deepen and consolidate existing knowledge and practices, the Project defines four objectives to be pursued intentionally throughout its term: i) valuing the work carried out within schools in the digital transition, distinguishing and promoting successful practices; ii) reinforcing the sustained integration of teaching and learning dynamics with digital, in the classroom and in other local contexts; iii) recognizing and assessing students' digital skills, bearing in mind the Portuguese Profile of Students Leaving Compulsory Schooling; iv) monitoring project activities and understand their impact on teachers and students. In this article, we present the main starting questions and the conceptual foundations and premises that underpin the intervention plan's design. Although the plan involved other dimensions, particularly related to school management, here we will emphasize the assumption of digital technologies that we assumed and how we worked with the teachers. We conclude with the presentation of the Project's assumptions and work axes.

Artificial neural network-enhanced mathematical simulation based on Carrera unified formulation for dynamic analysis and structural integrity assessment of advanced nanocomposites reinforced tunnel structures

This article presents the application of the Carrera Unified Formulation (CUF) for dynamic analysis and structural integrity assessment of advanced nanocomposite-reinforced tunnel structures. CUF offers a generalized framework that simplifies the modeling of complex structures, providing an efficient approach to address the dynamic behavior of nanocomposites. Advanced nanocomposites, which enhance mechanical performance and durability, are increasingly used in critical infrastructure such as tunnel systems. Traditional methods often fall short in capturing the intricate interactions within these materials, particularly under dynamic loads. CUF overcomes these challenges by incorporating higher-order theories and multi-scale analysis, allowing for accurate prediction of displacements, stresses, and potential failure modes. To further validate the results, an Artificial Neural Network (ANN) model is trained using simulated data, ensuring robust predictions for various loading conditions. The ANN assists in approximating the dynamic response and integrity of the structure, enabling real-time assessments with minimal computational expense. The combined CUF-ANN approach demonstrates high accuracy and efficiency, making it a reliable tool for the structural integrity assessment of nanocomposite-reinforced tunnels. This study highlights the significant potential of integrating CUF and ANN for the analysis and design of advanced engineering structures subjected to dynamic environmental conditions.

PB-Trajectron: Physics bounded neural network for generalized trajectory prediction

Vehicle trajectory prediction is a critical technology in autonomous driving systems, as the quality of prediction directly affects downstream path planning and vehicle control. Although recent studies have shown that models combining kinematic rules with data-driven methods exhibit better interpretability in trajectory prediction, these models often adopt fixed constraint strengths, limiting their generalization ability in diverse traffic scenarios. This fixed constraint strength design restricts the model’s adaptability to different traffic environments.To address this issue, we propose PB-Trajectron, a dynamic integration mechanism that enhances the model’s prediction performance and generalization capability. PB-Trajectron is a dynamic integration mechanism that combines vehicle kinematic rules with deep learning prediction models for generalized vehicle trajectory prediction. The model integrates physics-based motion models and Deep Neural Networks (DNNs) to provide reasonable physical explanations for predicting vehicle trajectories at different speeds. First, we establish two vehicle kinematic constraints applicable to different prediction scenarios, enabling the agent to switch between these constraints based on the causal relationships between vehicle motion states to avoid generating non-physical trajectory results. Second, we propose a kinematic constraint decision framework based on velocity thresholds, which demonstrates adaptive adjustment, allowing the agent to switch tasks according to real-time state conditions and adaptively adjust the contribution of different physical constraints based on actual vehicle speeds. Finally, we investigate the advantages of PB-Trajectron in Out-of-Domain(OOD) generalization.Cross-scenario experiments on the nuScenes dataset show that, compared to previous methods, PB-Trajectron reduces the final displacement error by 7.69% when the prediction step length is 4. Furthermore, out-of-domain generalization tests on the INTERACTION dataset demonstrate that PB-Trajectron achieves a 12.23% reduction in average prediction error on datasets from different countries and scenarios compared to previous methods. The proposed mechanism can better adapt to complex and diverse traffic scenarios, laying the foundation for explainable and robust autonomous driving systems.

Neurophysiological principles underlying predictive coding during dynamic perception-action integration

A major concept in cognitive neuroscience is that brains are “prediction machines”. Yet, conceptual frameworks on how perception and action become integrated still lack the concept of predictability and it is unclear how neural processes may implement predictive coding during dynamic perception-action integration. We show that distinct neurophysiological mechanisms of nonlinearly directed connectivities in the theta and alpha band between cortical structures underlie these processes. During the integration of perception and motor codes, especially theta band activity in the insular cortex and temporo-hippocampal structures is modulated by the predictability of upcoming information. Here, the insular cortex seems to guide processes. Conversely, the retrieval of such integrated perception-action codes during actions heavily relies on alpha band activity. Here, directed top-down influence of alpha band activity from inferior frontal structures on insular and temporo-hippocampal structures is key. This suggests that these top-down effects reflect attentional shielding of retrieval processes operating in the same neuroanatomical structures previously involved in the integration of perceptual and motor codes. Through neurophysiology, the present study connects predictive coding mechanisms with frameworks specifying the dynamic integration of perception and action.

Integration of deep learning and computational fluid dynamics for rapid aerodynamic force prediction of compressor blades

The distribution of flow fields around compressor blades is crucial for the performance and reliability of aircraft engines. To effectively obtain aerodynamic loads, this study combines deep learning with computational fluid dynamics (CFD) to develop an efficient aerodynamic prediction model. Initially, CFD is used to acquire detailed flow field data for the blade surface and its surrounding environment. Subsequently, a distance field parameterization method is applied to process the blade geometry, and deep learning models are used to capture the complex relationship between blade geometry and aerodynamic parameters with high precision. The results indicate that the proposed model can predict aerodynamic loads within seconds with a mean squared error of less than 2%. Compared to traditional parameterization methods and other deep learning approaches, this model exhibits higher accuracy. The findings highlight the effectiveness of integrating deep learning with CFD to enhance aerodynamic predictions and provide a promising approach for future aerodynamic modeling research.