To supplement missing logging information without increasing economic cost, a machine learning method to generate synthetic well logs from the existing log data was presented, and experimental verification application effect analysis were carried out. Since traditional Fully Connected Neural Network (FCNN) is incapable of preserving spatial dependency, Long Short-Term Memory (LSTM) network, which kind Recurrent (RNN), utilized establish for reconstruction. By this method, can be generated...

Knowledge constitutes the accumulated understanding and experience that humans use to gain insight into world. In deep learning, prior knowledge is essential for mitigating shortcomings of data-driven models, such as data dependence, generalization ability, compliance with constraints. Here, we present a framework enable efficient evaluation worth by derived rule importance. Through quantitative experiments, assess influence volume estimation range on knowledge. Our findings elucidate...

In chemistry, empirical paradigms prevail, especially within the realm of chromatography, where selection separation conditions frequently relies on chemist's experience. However, underlying rationale for such experiential knowledge has not been established or analysed. This study explicitly elucidates how chemists use thin-layer chromatography (TLC) to determine column (CC) conditions, employing statistical analysis and machine learning techniques. An experimental dataset CC is generated...

Machine learning models have been successfully used in many scientific and engineering fields. However, it remains difficult for a model to simultaneously utilize domain knowledge experimental observation data. The application of knowledge-based symbolic AI represented by an expert system is limited the expressive ability model, data-driven connectionism neural networks prone produce predictions that violate physical mechanisms. In order fully integrate with observations, make full use prior...

Development of shale gas resources is expected to play an important role in China's projected transition a low-carbon energy future. The question arises whether the availability water could limit this development. paper considers range scenarios define demand for needed accommodate production through 2020. Based on data from field at Fuling, first large-scale China, it concluded that intensity development China (water per unit lateral length) likely exceed US by about 50%. Fuling would...

Electricity constitutes an indispensable source of secondary energy in modern society. Accurate and robust short-term electrical load forecasting is essential for more effective scheduling generation, minimizing the gap between generation demand, reducing electricity losses. This study proposes theory-guided deep-learning (TgDLF), which a gradient-free model that fully combines domain knowledge machine learning algorithms. TgDLF predicts future through ratio decomposition, dimensionless...

Geomechanical logs are of ultimate importance for subsurface description and evaluation, as well the exploration underground resources, such oil gas, groundwater, minerals, geothermal energy. Together with geological hydrological properties, low-cost high-accuracy models can be generated based on geomechanical parameters. However, it is challenging to directly measure parameters, they usually estimated other measured quantities. For example, may obtained certain empirical from sonic together...

Partial differential equations (PDEs) are concise and understandable representations of domain knowledge, which essential for deepening our understanding physical processes predicting future responses. However, the PDEs many real-world problems uncertain, calls PDE discovery. We propose symbolic genetic algorithm (SGA-PDE) to discover open-form directly from data without prior knowledge about equation structure. SGA-PDE focuses on representation optimization PDE. Firstly, uses mathematics...

Electrical energy is essential in today's society. Accurate electrical load forecasting beneficial for better scheduling of electricity generation and saving energy. In this paper, we propose an adaptive deep-learning framework by integrating Transformer domain knowledge (Adaptive-TgDLF). Adaptive-TgDLF introduces the model learning methods (including transfer different locations online time periods), which captures long-term dependency series, more appropriate realistic scenarios with...

The working mechanisms of complex natural systems tend to abide by concise partial differential equations (PDEs). Methods that directly mine from data are called PDE discovery, which reveals consistent physical laws and facilitates our interactions with the world. In this paper, an enhanced deep reinforcement-learning framework is proposed uncover symbolically open-form PDEs little prior knowledge. Particularly, based on a symbol library basic operators operands, can be represented tree...

Marine renewable energy is gaining prominence as a crucial component of the supply in coastal cities due to proximity and minimal land requirements. The synergistic potential integrating floating photovoltaics with offshore wind turbines presents an encouraging avenue for boosting power production, amplifying spatial generation density, mitigating seasonal output fluctuations. While global promise wind-photovoltaic hybrid systems evident, definitive understanding their remains elusive. Here,...

Well placement optimization is important in reservoir management, but it challenging to implement due the high-dimensional solution space and large number of simulations required. Surrogate models may assist alleviate computational burden by efficiently approximating full-order models. Although deep learning has been proven be effective for surrogate modeling, most surrogates are purely data-driven, underlying physical principles or theories subsurface flows not considered. In this work, a...

In this study, an efficient stochastic gradient-free method, the ensemble neural networks (ENN), is developed. ENN, optimization process relies on covariance matrices rather than derivatives. The are calculated by randomized maximum likelihood algorithm (EnRML), which inverse modeling method. ENN able to simultaneously provide estimations and perform uncertainty quantification since it built under Bayesian framework. also robust small training data size because of realizations essentially...

Scientific research's mandate is to comprehend and explore the world, as well improve it based on experience knowledge. Knowledge embedding knowledge discovery are two significant methods of integrating data. Through embedding, barriers between data can be eliminated, machine learning models with physical common sense established. Meanwhile, humans' understanding world always limited, takes advantage extract new from observations. not only assist researchers better grasp nature physics, but...

Unveiling the underlying governing equations of nonlinear dynamic systems remains a significant challenge. Insufficient prior knowledge hinders determination an accurate candidate library, while noisy observations lead to imprecise evaluations, which in turn result redundant function terms or erroneous equations. This study proposes framework robustly uncover open-form partial differential (PDEs) from limited and data. The operates through two alternating update processes: discovering...