A5025865498- Reservoir Engineering and Simulation Methods
- Hydraulic Fracturing and Reservoir Analysis
- Enhanced Oil Recovery Techniques
- Drilling and Well Engineering
- Oil and Gas Production Techniques
- Meteorological Phenomena and Simulations
- Vibration and Dynamic Analysis
- Manufacturing Process and Optimization
- Hydraulic and Pneumatic Systems
- Hydrocarbon exploration and reservoir analysis
- Speech and Audio Processing
- Laser and Thermal Forming Techniques
- Mining Techniques and Economics
- Bayesian Methods and Mixture Models
- Imbalanced Data Classification Techniques
- Infrastructure Maintenance and Monitoring
- Speech Recognition and Synthesis
- Mineral Processing and Grinding
- Advanced Data Processing Techniques
- Fluid Dynamics and Turbulent Flows
- Model Reduction and Neural Networks
- Artificial Intelligence in Healthcare
- Electric Power System Optimization
- CO2 Sequestration and Geologic Interactions
- Advanced Adaptive Filtering Techniques
State Key Laboratory of Oil and Gas Reservoir Geology and Exploitation
2022-2024
China University of Petroleum, East China
2024
Motorola (United States)
2002
Interface (United States)
2002
Data imbalance is a common problem in most practical classification applications of machine learning, and it may lead to results that are biased towards the majority class if not dealt with properly. An effective means solving this undersampling borderline area; however, difficult find area fits boundary. In paper, we present novel framework, whereby clustering samples conducted segmentation then performed boundary according clusters obtained; enables better shape be obtained via performance...
Summary History matching is a crucial process that enables the calibration of uncertain parameters numerical model to obtain an acceptable match between simulated and observed historical data. However, implementation history-matching algorithm usually based on iteration, which computationally expensive due numerous runs simulation. To address this challenge, we propose surrogate for simulation autoregressive combined with convolutional gated recurrent unit (ConvGRU). The proposed...
Due to the influence of long-term waterflooding, reservoir physical properties and percolation characteristics tend change greatly in offshore unconsolidated sandstone reservoirs at ultrahigh water-cut stage, which can affect remaining oil distribution. Remaining characterization proper development strategy-making are vital importance achieve high-efficiency mature reservoirs. The present numerical simulation method is difficult apply due problems noncontinuous low computational efficiency....
Summary Production optimization is a crucial component of closed-loop reservoir management, which typically aims to search for the best development scheme maximum economic benefit. Over decades, large body algorithms have been proposed address production problems, among surrogate-assisted evolutionary algorithm (SAEA) gained much research popularity due its problem information-agnostic implementation and strong global capability. However, existing methods often optimize individual tasks from...
The rapid and accurate forecasting of two-phase flow in porous media is a critical challenge oil field development, exerting substantial impact on optimization decision-making processes. Although the Convolutional Long Short-Term Memory (ConvLSTM) network effectively captures spatiotemporal dynamics, its generalization predicting complex engineering problems remains limited. Similarly, although Fourier Neural Operator (FNO) demonstrates adeptness at learning operators for solving partial...
Summary The wellsite serves as the fundamental unit in development of oil and gas fields, functioning a hub for production activities, with workover operations being critical means to ensure continuity. In addition, it plays crucial role environmental protection, preventing leakage pollution. Various pieces mechanical equipment deployed at are essential tasks such extraction well repair operations, holding pivotal position oil- gasfield development. Consequently, intelligent implementation...
Summary Rod pump systems are the primary production tools in oilfield development. Analyzing their working conditions provides a theoretical foundation for formulating optimization plans and adjusting equipment parameters. Existing machine learning–based condition analysis methods rely on dynamometer cards cannot capture actual operational trajectory of pumping unit. To address this issue, paper proposes keypoint detection method units based pose estimation heatmap-free joint from video...
The development of offshore oil fields confronts challenges associated with high water cut and low displacement efficiency. Reservoir injection-production optimization stands out as an effective means to reduce costs enhance efficiency in oilfield development. process optimizing injection production reservoirs involves designing strategies for a large number wells time steps, constituting large-scale, complex, costly computation problem. In recent years, the rapid advancements big data...
Summary In this study, we propose an inverse design approach based on a neural network for novel multiport power divider (MP‐PD) with complex geometry. The is obtaining geometry from the desired physical performance to address challenge of conventional methods. We develop hybrid model design. backbone architecture incorporates bidirectional long short‐term memory module, multihead self‐attention and convolutional modules. This employed capture feature learn relationship between geometric...
To assess whether a development strategy will be profitable enough, production forecasting is crucial and difficult step in the process. The history of other reservoirs same class tends to studied make predictions accurate. However, permeability field, well patterns, regime must all similar for two considered class. This results very few available experiences from even though there lot historical information on numerous because it find such reservoirs. paper proposes learn-to-learn method,...
In the wheat nutrient deficiencies classification challenge, we present DividE and EnseMble (DEEM) method for progressive test data predictions. We find that (1) images are provided in challenge; (2) samples equipped with their collection dates; (3) of different dates show notable discrepancies. Based on findings, partition dataset into discrete groups by train models each divided group. then adopt pseudo-labeling approach to label incorporate those high confidence training set....
Prior information about the operating environment of a speech recognizer is often general and abstract. Frequently such as number speakers with foreign accents or callers using cellular phones readily available. Incorporating this during model training difficult. This paper generalizes popular MAP algorithm derived by Gauvain Lee (1994) so that more prior can be utilized training. The priors are in form mixture distributions each component representing unique property data mixing weights...