As an underwater thruster, the pump-jet propulsor (PJP) exhibits low radiation noise but generates significant line spectral in low-frequency band. In this paper, we equipped PJP hub with two types of propeller boss cap fins (PBCF): one fixed and other rotating rotor. The hybrid large eddy simulation Reynolds-averaged Navier–Stokes method, along Ffowcs Williams-Hawkings (FW-H) equation, are employed to systematically analyze hydrodynamics, exciting force, flow noise, field PJPs. results...

The concept of the exposome encompasses totality exposures from a variety external and internal sources across an individual's life course. wealth existing spatial contextual data makes it appealing to characterize individuals' advance our understanding environmental determinants health. However, is very different other factors measured at individual-level as are more heterogenous with unique correlation structures various spatiotemporal scales. These distinctive characteristics lead...

Comprehensively grasping the wake dynamics of pump-jet propulsor (PJP) lies at core developing and fine-tuning future PJP design, particularly exciting forces suppression noise reduction. In this work, a pre-swirl stator is considered to investigate its evolution mechanics. The stress-blended eddy simulation (SBES) implemented for obtaining turbulent flow, dynamic mode decomposition (DMD) method utilized analyze flow evolution. numerical results align with experimental data within an...

When an ice-class propeller is operating in ice-covered environment, as some ice blocks slide along the ship hull front of blades, inflow ahead will become non-uniform. Consequently, excitation force applied to blades increase and massive cavitation bubbles be generated. In this paper, a hybrid Reynolds-Averaged Navier–Stokes/Large Eddy Simulation method Schnerr–Sauer model are used investigate hydrodynamics, force, evolution flow field characteristics blockage conditions. The results show...

During the submarine's surfacing process, rotor of pump-jet propulsor (PJP) is subjected to nonuniform hydrodynamic loads and cavitation, inducing cavitation-induced noise. In this paper, hybrid Reynolds Averaged Navier–Stokes/Large Eddy Simulation method, Schnerr–Sauer model Ffowcs Williams–Hawkings equations are adopted explore hydrodynamics, excitation force, radiation noise, cavitation evolution PJP with different numbers σn oblique angles θ. The results show that when decreases from 1.5...

Shared gradients are widely used to protect the private information of training data in distributed machine learning systems. However, Deep Leakage from Gradients (DLG) research has found that can be recovered shared gradients. The DLG method still some issues such as “Exploding Gradient,” low attack success rate, and fidelity data. In this study, a Wasserstein method, named WDLG, is proposed; theoretical analysis shows under premise output layer model “bias” term, predicting “label” by...

This study provides the framework for a variational Bayesian convolutional neural network (VB-CNN) to quickly predict wake velocity field of pump-jet propulsor and quantify uncertainty. For engineering application experiments, can be obtained by using discrete pressure points when model is trained. The weight distribution altered from point probability method, which also takes into account prior knowledge datasets. VB-CNN produces superior results method in small datasets investigates...

Federated learning protects the privacy information in data set by sharing average gradient. However, "Deep Leakage from Gradient" (DLG) algorithm as a gradient-based feature reconstruction attack can recover training using gradients shared federated learning, resulting private leakage. has disadvantages of slow model convergence and poor inverse generated images accuracy. To address these issues, Wasserstein distance-based DLG method is proposed, named WDLG. The WDLG uses distance loss...

A numerical analysis based on stress-blended eddy simulation was conducted to investigate the pressure fluctuation of bow a submarine at various velocities (5.93 kn, 10 and 12 kn). The results were compared with experimental data demonstrate validity method. Self-power spectrum wave-number frequency discussed from perspective energy. show that increasing velocity, in axial direction increases, transition point moves forward, Tollmien–Schlichting wave raises.

Abstract This study investigates the impact of clinical trial eligibility criteria on patient survival and serious adverse events (SAEs) in colorectal cancer (CRC) drug trials using real-world data. We utilized OneFlorida+ network’s data repository, conducting a retrospective analysis CRC patients receiving FDA-approved first-line metastatic treatments. Propensity score matching created balanced case-control groups, which were evaluated machine learning algorithms to assess effects criteria....

Adversarial example generation techniques for neural network models have exploded in recent years. In the adversarial attack scheme image recognition models, it is challenging to achieve a high success rate with very few pixel modifications. To address this issue, paper proposes an method based on adaptive parameter adjustable differential evolution. The realizes dynamic adjustment of algorithm performance by adjusting control parameters and operation strategies evolution algorithm, while...

The paper investigates on the reachable domain of generated trjectories after an impulse with fixed magnitude and arbitrary direction applied at any point initial satellite orbit. By coalescing simulation analysis geometric analysis, a method is given to determine both circular elliptic For trajectories, present general feasible investigate reach single impulse.

Abstract This paper presents a numerical simulation of the steady propulsion state manta rays and investigates influence single motion parameters addition perturbation signals on hydrodynamic characteristics vortex evolution rays. A model equations ray were established by observing living organisms, then computational method combining immersed boundary (IBM) Sphere function-based Gas Kinetic Scheme (SGKS) was used to simulate active ray. The results show that in parameter, as frequency...

Deep learning is one of the most exciting and promising techniques in field artificial intelligence (AI), which drives AI applications to be more intelligent comprehensive. However, existing deep usually require a large amount expensive labeled data, limit application development techniques, thus it imperative study unsupervised machine learning. The representations by mutual information estimation maximization (Deep InfoMax or DIM) method has achieved unprecedented results DIM method,...

In deep learning, supervised learning techniques usually require a large amount of expensive labeled data to train the network, and feature representations extracted by model mix multiple attributes, resulting in that are difficult decouple non-interpretable, which restricts application development techniques, for this reason, it is particularly important study decoupled representation methods unsupervised learning. Although Learning mutual information estimation maximization (DIM) method...

Ambiguity and misunderstanding of free-text clinical trial eligibility can affect the accuracy translating investigators' mental model study population to correct cohort identification queries. In this pilot study, eliminate ambiguity when parsing criteria, we built ontology-based representations standardize criteria. We analyzed 10 Alzheimer's disease (AD) trials' criteria categorized them into general query patterns using an annotation schema borrowed from literature on constructing...