AI-Enabled Compact and Efficient Dynamic Light Scattering System for Precise Microparticle Sizing

Applied Sciences, Journal Year: 2025, Volume and Issue: 15(4), P. 1900 - 1900

Published: Feb. 12, 2025

A cost-effective and efficient AI-DLS framework integrating dynamic light scattering (DLS) with artificial intelligence (AI) enables precise microparticle size characterization. compact DLS system was developed using a helium–neon laser, high-frequency photodetectors, optimized optical components, achieving significant miniaturization—4.5% volume, 16.7% weight, 25% cost of conventional systems. Advanced signal processing, such as Kalman filtering, improved data quality, while deep learning models (deep neural network (DNN), one-dimensional convolutional (1D-CNN), 2D-CNN Gramian angular field transformations) enhanced prediction accuracy. The model achieved exceptional results, mean absolute error 10 nm 99% matched the stability accuracy traditional instruments, reducing test time by 80%. This scalable, low-cost solution overcomes limitations, offering broad accessibility for scientific industrial

Language: Английский

A Microseismic Phase Picking and Polarity Determination Model Based on the Earthquake Transformer

Applied Sciences, Journal Year: 2025, Volume and Issue: 15(7), P. 3424 - 3424

Published: March 21, 2025

Phase arrival times and polarities provide essential kinematic constraints for dynamic insights into seismic sources, respectively. This information serves as fundamental data in seismological study. For microseismic events with smaller magnitudes, reliable phase picking polarity determination are even more challenging but play a crucial role source location focal mechanism inversion. study innovatively proposes deep learning model suitable simultaneous continuous waveforms. Building upon the Earthquake Transformer (EQT) model, we implemented structural improvements through four distinct decoders specifically designed three tasks of P-wave picking, S-wave first-motion named EQT-Plus (EQTP). Notably, task was decomposed two independent to enhance characteristics. Through training on northern California dataset testing (Md < 3) Geysers region, results demonstrate that EQTP achieves superior performance both compared PhaseNet+ model. It not only provides accurate also shows higher consistency manual determination. We further validated good generalization ability DiTing from China. advances adaptation seismology reliably delivers refined inversion, offering an alternative advanced tool community.

Language: Английский

Seismic response of deep soft soil sites with varying shear wave velocities

Frontiers in Earth Science, Journal Year: 2024, Volume and Issue: 12

Published: Oct. 1, 2024

The variations in seismic response between deep soft soil sites with different shear wave velocities were not fully understood. This study focuses on the of lower reaches Yangtze River, China. A nonlinear dynamic finite element model was developed for two representative borehole profiles and velocity tested by single method. Two cyclic constitutive models are used thus compared through site response. To accurately calibrate parameters, resonant column tests conducted 21 samples collected from boreholes. results show that peak ground acceleration ( PGA ) under low-frequency (Liuan) input motion higher to medium- high-frequency (Kobe Nahanni) motions. amplification factor motions can be approximated an exponential function. tends as equivalent V se decreases. shapes spectral similar sites, despite a substantial difference them. Additionally, crossover point observed sites. period corresponding this increased increasing intensity motions, indicating became softer generally exhibited longer characteristic acceleration. paper also highlights significance selecting precise calibration parameters analysis providing scientific basis future analyses.

Language: Английский