A5044393709- Geophysical and Geoelectrical Methods
- Geophysical Methods and Applications
- Seismic Imaging and Inversion Techniques
- Seismic Waves and Analysis
- Solar Radiation and Photovoltaics
- Computational Physics and Python Applications
- Energy Load and Power Forecasting
- Aerospace and Aviation Technology
- Electromagnetic Simulation and Numerical Methods
- Underwater Acoustics Research
- Advanced Numerical Methods in Computational Mathematics
- Model Reduction and Neural Networks
- High-pressure geophysics and materials
- Electric Power System Optimization
- Anomaly Detection Techniques and Applications
- Earthquake Detection and Analysis
- Autonomous Vehicle Technology and Safety
Zhejiang University
2022-2024
Southern University of Science and Technology
2023-2024
University of Jinan
2023
The nonlinear conjugate gradient (NLCG) algorithm is one of the popular linearized methods used to solve frequency-domain electromagnetic (EM) geophysical inverse problem. During NLCG iterations, model guides searching direction while line-search determines step length each iteration. Normally, line search requires solving corresponding forward problem a few times. Since usually computationally inefficient, we introduce adaptive descent (AGD) accelerate EM within framework. AGD variant...
Artificial neural networks (ANN) have gained significant attention in magnetotelluric (MT) inversions due to their ability generate rapid inversion results compared traditional methods. While a well-trained ANN can deliver near-instantaneous results, offering substantial computational advantages, its practical application is often limited by difficulties accurately fitting observed data. To address this limitation, we introduce novel approach that customizes an auto-encoder (AE) whose...
The efficiency of solving geophysical inverse problem largely relies on the corresponding forward problem. As for electromagnetic (EM) data modeling in frequency domain, conventional numerical methods, e.g. finite difference method (FDM), discretize governing equations resulting a large linear system which is usually expensive to solve. Meanwhile, inversion iteration we normally do not need solve high precision. Thus rapid surrogate approach uses neural network promising replacing module...
SUMMARY Effective viscosity of the upper mantle is a critical parameter for comprehending dynamics lithosphere and plate tectonics. In recent years, magnetotelluric (MT) surveys have gained attention as potential tool determining viscosity. However, direct physical basis relationship between effective electrical resistivity still needs to be improved. To address this issue, we established that connects under different thermochemical conditions principle neutrality. The creep conductions...
Large-scale wind power forecasting with high accuracy and reliability can proficiently decrease the unfavorable impacts of uncertainty. However, existing performances are greatly influenced by intermittent changes environmental meteorological factors, which brings much challenge for integration. To solve above problem, based on a dynamic selection seasonal features, an improved Seasonal Feature Selection Temporal Convolutional Network (SFS-TCN) model is proposed to forecast ultra-short-term...
This research delves into the exploration of Guangdong's urban subsurface using magnetotelluric (MT) surveys, with a specific focus on uncovering hidden fault structures for planning land resources. Despite challenges posed by region's economic activities, meticulous data processing techniques were employed to refine raw subsequent 3D inversion. The successful application these methods resulted in reliable inverted resistivity models, effectively delineating at depth. Alignment prior...
Supervised learning has emerged as an effective approach to solving geophysical inversions. The quality of such inversion depends mainly on the training datasets (labeled samples), usually created by synthetic models and forward simulations. We propose a straightforward generating abundant representative labeled samples for 2D or 3D magnetotelluric (MT) data (or other similar exploration methods). requires very little prior information been demonstrated be quick recovery random blocky models.