Search

The automatic identification of foreign accents can play a crucial role in various speech systems, including speaker identification, e-learning, telephone banking, and more. Additionally, it can greatly enhance the robustness of Automatic Speech Recognition (ASR) systems. Non-native accents in speech signals are characterized by distinct pronunciations, prosody, and voice characteristics of the speaker. However, automatically identifying foreign accents poses significant challenges, particularly in the context of multi-class modeling. Multi-classification models face difficulties in achieving high performance and dealing with computational challenges when confronted with multi-dimensional and unbalanced datasets, such as those with more than two accents. Furthermore, the choice of features remains a bottleneck problem for Foreign Accent Identification (FAID), further hindering performance in these tasks. Consequently, the accuracy of current systems is typically low. To address these challenges, this paper proposes a framework based on the Multi-Kernel Extreme Learning Machine (MKELM) model for the multi-classification of FAID. The MKELM model utilizes a novel weighted scheme to classify various non-native English accents, including Arabic, Chinese, Korean, French, and Spanish. The model first combines Mel-frequency cepstral coefficients (MFCCs) and prosodic features as input, trains pairwise binary classifiers independently, and subsequently employs a weighting scheme to distinguish between classes and identify accents. Through experiments, the proposed model achieves an accuracy rate of 84.72% using a paired weighting scheme. In contrast, the accuracy rate drops to 66.5% when employing the traditional non-weighted multi-classification scheme. A comparison with other models demonstrates the significant advantages of the proposed model in FAID multi-class classification, showcasing improved accuracy, reduced computational complexity (requiring fewer computations, faster learning rates, and shorter training time), and enhanced stability compared to state-of-the-art classification methods.

The geological storage of carbon dioxide (CO2) is a crucial technology for mitigating global temperature rise. Near-depleted edge–bottom water reservoirs are attractive targets for CO2 storage, as they can not only enhance oil recovery (EOR) but also provide important potential candidates for geological storage. This study investigated CO2-enhanced oil recovery and storage for a typical near-depleted edge–bottom water reservoir that had been developed for a long time with a recovery factor of 51.93%. To improve the oil recovery and CO2 storage, new production scenarios were explored. At the near-depleted stage, by comparing the four different scenarios of water injection, gas injection, water-alternating-gas injection, and bi-directional injection, the highest additional recovery of 3.62% was achieved via the bi-directional injection scenario. Increasing the injection pressure led to a higher gas–oil ratio and liquid production rate. After shifting from the near-depleted to the depleted stage, the most effective approach to improving CO2 storage capacity was to increase reservoir pressure. At 1.4 times the initial reservoir pressure, the maximum storage capacity was 6.52 × 108 m3. However, excessive pressure boosting posed potential storage and leakage risks. Therefore, lower injection rates and longer intermittent injections were expected to achieve a larger amount of long-term CO2 storage. Through the numerical simulation study, a gas injection rate of 80,000 m3/day and a schedule of 4–6 years injection with 1 year shut-in were shown to be effective for the case considered. During 31 years of CO2 injection, the percentage of dissolved CO2 increased from 5.46% to 6.23% during the near-depleted period, and to 7.76% during the depleted period. This study acts as a guide for the CO2 geological storage of typical near-depleted edge–bottom water reservoirs.

There are abundant bottom water reservoirs in China. Unlike conventional oil reservoirs, bottom water reservoirs have various problems, such as early water breakthrough, short water-free oil recovery period, and rapid water cut increase. For example, during water flooding, the injected water easily breaks into the bottom water and does not effectively displace the upper crude oil. The recovery rate is generally low. Based on this phenomenon, an experimental study of nitrogen foam flooding in bottom water reservoirs is conducted in this paper. The seepage characteristics of nitrogen foam in oil and water layers are studied through one-dimensional core tube experiments. Through two-dimensional plate oil displacement experiments, we have revealed the fluid migration and distribution characteristics in the plane and vertical directions during nitrogen foam flooding in bottom water reservoirs; additionally, we have summarized the mechanisms of nitrogen foam in bottom water reservoirs involved in improving oil recovery characteristics. The research results show that the seepage resistance of foam in the water layer is much greater than that in the oil layer, effectively increasing the displacement strength of the oil layer. During the development stage of bottom water flooding in bottom water reservoirs, the water cut increases rapidly, the bottom water coning is obvious, and the residual oil is mainly distributed between the oil wells and the upper part of the oil layer near the wellbore. During nitrogen foam flooding, the foam enters the water layer to form an effective plug so that the subsequent foam is diverted into the oil layer; additionally, the oil is displaced laterally to the production well for production. When the foam enters the oil layer, it defoams and floats to form a secondary gas cap; this effect causes displacement of the residual oil at the top and effectively improves the displacement efficiency by weeping volume of the injected fluid”

Data-driven machine learning technology can learn and extract features, a factor which is well recognized to be powerful in the warning and prediction of severe weather. With the large-scale deployment of the radar wind profile (RWP) observational network in China, dynamical variables with higher temporal and spatial resolution in the vertical become strong supports for machine-learning-based severe convection prediction. Based on the RWP mesonet that has been deployed in Beijing, this study uses the measurements from four triangles composed of six RWP stations to determine the profiles of divergence, vorticity, and vertical velocity before rainfall onsets. These dynamic feature variables, combined with cloud properties from Himawari-8 and ERA-5 reanalysis, serve as key input parameters for two rainfall forecast models based on the random forest (RF) classification algorithm. One is for the rainfall/non-rainfall forecast and another for the rainfall grade forecast. The roles of dynamic features such as divergence, vorticity, and vertical velocity are examined from ERA-5 reanalysis data and RWP measurements. The contribution of each feature variable to the performance of the RF model in independent tests is also discussed here. The results show that the usage of RWP observational data as the RF model input tends to result in better performance in rainfall/non-rainfall forecast 30 min in advance of rainfall onset than using the ERA-5 data as inputs. For the rainfall grade forecast, the divergence and vorticity that were estimated from the RWP measurements at 800 hPa show importance in improving the model performance in heavy and moderate rain forecasts. This indicates that the atmospheric dynamic variable measurements from RWP have great potential to improve the prediction skill of convection with the aid of a machine learning model.

Abstract Many factors will affect the result of nitrogen-assisted gravity drainage (NAGD). In this work, sandpack and 2-dimensional (2D) visual plate model were employed to investigate the influence factors of NAGD, such as gas cap volume, formation dip angle, and back pressure. Nitrogen foam was also studied to improve the gas channeling and breakthrough. The experimental results indicated that the gas cap volume was the most significant influence factor on the oil recovery. With the increase of the gas cap volume, the oil recovery increased first and then decreased, and finally increased. The NAGD had a better performance in big dip angle of formation. There was a critical value of back pressure to obtain the maximum oil recovery. However, the gas channeling was intensely observed by experiments of both sandpack and 2D visual plate model. Fortunately, nitrogen foam could effectively inhibit the gas channeling and improve the result of NAGD.

Under acute perturbations from outer environment, a normal cell can trigger cellular self-defense mechanism in response to genome stress. To investigate the kinetics of cellular self-repair process at single cell level further, a model of DNA damage generating and repair is proposed under acute Ion Radiation (IR) by using mathematical framework of kinetic theory of active particles (KTAP). Firstly, we focus on illustrating the profile of Cellular Repair System (CRS) instituted by two sub-populations, each of which is made up of the active particles with different discrete states. Then, we implement the mathematical framework of cellular self-repair mechanism, and illustrate the dynamic processes of Double Strand Breaks (DSBs) and Repair Protein (RP) generating, DSB-protein complexes (DSBCs) synthesizing, and toxins accumulating. Finally, we roughly analyze the capability of cellular self-repair mechanism, cellular activity of transferring DNA damage, and genome stability, especially the different fates of a certain cell before and after the time thresholds of IR perturbations that a cell can tolerate maximally under different IR perturbation circumstances.