Search

In order to solve the problem that the feature distribution is not compact enough due to insufficient processing of feature distribution in open intent detection task, an open intent detection method integrating BERT, distance loss and decision boundary was proposed. Firstly, the context features between texts were captured by BERT model. Then, the learning of sample features was made more compact by distance loss. Finally, decision boundary learning was carried out to achieve the task of open intent detection. The results show that the proposed method has high performance on the public dataset StackOverflow, with the best performance under two different known intent ratio settings, achieving the accuracy of 88.28% and 84.43%, and the F1 values of 87.51% and 87.40%, respectively. The research results complement the future representation reprocessing method for boundary detection,and can provide reference for open intent detection.

To support the "people-centered" chronic disease health management work, Shanghai has developed a community chronic disease health management information system. This system provides an efficient tool for achieving integrated medical care and disease prevention, as well as collaborative management between doctors and patients, through functions such as information registration and management, risk assessment, chronic disease screening, patient evaluation, and categorized follow-up. It delivers comprehensive, precise, and high-quality health management services for various chronic diseases, effectively promoting the implementation of comprehensive chronic disease prevention and control strategies and the transformative development of chronic disease health management services. However, the system currently requires the inclusion of a broader range of chronic diseases and the enhancement of interactive integration between clinic-based doctor follow-ups and home-based self-management. Future improvements may include the integration of additional disease types and the fusion of complex interaction scenarios to continuously adapt to the new situations and challenges in chronic disease prevention and control.

In order to reveal the evolution law of coal and gas outburst gas-solid particle flow, an analysis method combining field statistical data with numerical simulation data was proposed based on the discrete element numerical simulation method. The development process of gas-solid two-phase flow of coal and gas outburst was studied, and the movement track of coal and gas was revealed. Based on principal component analysis (PCA) and fuzzy comprehensive evaluation (FCE), 27 outburst mines were classified. The research results show that coal and gas outburst can be divided into four stages: coal and gas breaking through crosscut stage, coal and gas mixed impact stage, coal and gas two-phase flow diversion stage and coal and gas outburst attenuation stage; with the development of the outburst, the two-phase flow of coal and gas is divided. Compared with coal particles, the gas movement speed is higher, mainly above the roadway, most coal particles settle to the bottom of the roadway, and a small number of coal particles and gas continue to move forward in coordination, the coal particle movement tracks are more complex than gas; through numerical simulation, the coal throw volume of three outburst mines are 121.5 t, 42.7 t and 320.32 t; according to PCA-FCE comprehensive evaluation, 27 coal and gas outburst mines include 16 general outburst mines and 11 serious outburst mines.

Reports on the utilization of nitrogen compounds to trace oil and gas migration in oil-scarce regions are limited. This scarcity can be attributed to the lack of clarity regarding the effectiveness of nitrogen compound indices in oil-bearing sandstone extracts. To address this knowledge gap, two sets of experiments were designed, encompassing five extraction methods, to evaluate the validity of nitrogen compound indices. The results showed that as the extraction intensity increased, the ratios of 1-/4-methylcarbazole and 1, 8-/2, 7-dimethylcarbazole in the extracts gradually decreased but remained consistently elevated compared to the control oils. The benzocarbazole [a]/[c] ratio remained relatively stable and closely aligned with the control oils. Upon analysis, it was concluded that the observed variations could be attributed to the differing adsorption capacities of carbazole isomers. These differences lead to the fractionation of carbazole isomers between free and adsorbed hydrocarbons under surface conditions, as well as between the extracts and rocks during the extraction process. In comparison to 1-/4-methylcarbazole and 1, 8-/2, 7-dimethylcarbazole, the benzocarbazole [a]/[c] ratio is less susceptible to these two fractionations, with its variations primarily governed by migration fractionation. Additionally, the indicated direction of oil and gas migration aligns with geological realities. Therefore, the benzocarbazole [a]/[c] ratio can be considered an effective nitrogen compound index in oil-bearing sandstone extracts, suitable for oil migration tracing studies in oil-scarce regions.

Purposes Bearing is one of the core components in the manufacturing industry. Its health status determines the safety of the host. Short-term failure prediction can effectively ensure the smooth progress of the industrial production process. Methods In order to solve the end-to-end problem, a temporal convolutional network (TCN) based short-term fault prediction strategy was proposed. The network could directly output the types of failure that would eventually occur in the bearing and the degradation stage that would be in the next moment through the data monitored at the current moment. In addition, soft threshold with attention mechanism is proposed to solve the problem of background noise in the working environment of bearings or noise interference in the process of data acquisition. During the short-term fault prediction process, the attention mechanism adaptively generates a soft threshold according to the prediction target of the TCN network, and the soft threshold acts on the spatiotemporal features extracted by the TCN to achieve the purpose of reducing noise impact. Findings The experimental results show that the proposed algorithm has high accuracy, which verifies the effectiveness and high practical engineering application value of the proposed algorithm.

In view of the new service scenarios and the demand for high intelligence network in computing and network convergence (CNC), the concept of autonomous CNC (Auto-CNC) is elaborated, where intelligence was introduced into all the aspects of CNC, including resource integration, process automation, and system intelligence.The current research directions and remaining challenges of CNC were introduced, and three key features, i.e., intent-driven computing network, the autonomous system operation and the adaptive co-evolution of communication, computing intelligence, were summarized from the proposed Auto-CNC.Meanwhile, the reference architecture and key technologies of Auto-CNC were described, which were followed by several preliminary exploration cases.Finally, future research trends and technical advice were discussed and recommended.

Roof caving has been the main threat to the safety of underground mining, in which the caving of roof rock blocks is particularly concerning. The secondary structural planes of surrounding rocks around underground excavations, such as roadways and stopes, are developed. The rock mass is prone to break into several independent blocks, and these rock blocks may slide and fall under the action of static in-situ stress or external dynamic disturbances. Under quasistatic stress conditions, the instability and collapse of roof rock blocks are mainly caused by structural plane extensions and changes in the stress balance condition of the rock block system. However, under external dynamic disturbances, the sliding process and instability of roof rock blocks are associated with the development of fractures triggered by stress waves. Further, they are affected by the variation of stress balance conditions of the rock block system and the transmission of stress waves in the block system. This paper summarized the existing studies on roof rock block stability under quasistatic and dynamic disturbance conditions. Previous studies have proposed relatively mature theoretical systems for the stability analysis of roof rock blocks under static or quasistatic situations. However, a majority of the studies on rock block stability under an external dynamic disturbance condition examine stress wave propagation in a rock block system while overlooking the analysis of the crack development mechanism and dynamic variation in stress balance conditions. Therefore, further research is necessary to reveal the caving mechanism of roof rock blocks triggered by a dynamic disturbance. By summarizing relevant work, the difficulties encountered in the study of roof rock block caving under a dynamic disturbance are discussed. The key scientific problem is to uncover the fundamental mechanism behind roof rock block instability induced by a dynamic disturbance. Finally, a series of experimental and numerical simulations conducted on the sliding and instability of roof rock blocks under a dynamic disturbance revealed clear differences in the mechanism of the sliding instability of roof blocks due to dynamic disturbances in different directions. The reduction in friction between the blocks is the fundamental cause of the key block sliding under a lateral disturbance, while the sliding of the rock blocks under a vertical disturbance is mainly driven by the dynamic load. This finding can provide a theoretical reference for preventing and controlling roof caving in underground mines.

To solve the problem that traditional named entity recognition methods cannot effectively utilize entity boundary information, a named entity recognition method based on joint entity boundary detection was proposed. The method took entity boundary detection as an auxiliary task, so that the model can enhance the ability of entity boundary recognition, and then improve the effect of entity recognition. Firstly, the Bert pretraining language model was used to embed the features of the original text to obtain word vectors, and the self-attention mechanism was introduced to enrich the context features of words. Secondly, on the basis of named entity recognition task, an auxiliary entity boundary detection task was added to enhance the recognition ability of the model to the entity boundaries. Thirdly, the effectiveness of the named entity recognition method and the baseline method was compared, and the test results were from ablation experiments. Finally, the influence of loss weight β on entity boundary detection was analyzed by examples. The experimental results show that on the English social media dataset Twitter-2015, the named entity recognition method combined with entity boundary detection achieves higher accuracy, recall rate and F1 value than the baseline model, of which the F1 value can reach 7357%. In addition, the boundary detection auxiliary task has a certain improvement effect on the baseline method. The proposed method can effectively utilize entity boundary information to obtain better entity recognition effect, and promote the development of human-computer interaction system, which is of great significance for downstream tasks of natural language processing.

Expansive soil is prone to develop a large number of cracks, which have a significant impact on its engineering properties. A prefabricated crack method of expansive soil samples is introduced to carry out the unconfined compression strength on the expansive soil samples under the specific crack morphology. The influence of the crack morphology on the strength characteristics of the expansive soil is examined, and the mechanism of crack action is revealed. The results show that: (1) The crack morphology has a significant effect on the failure mode of the samples. When the cracks become wider, which are perpendicular to the stress direction, the failure mode changes from shear-tensile failure to tensile failure. When the cracks parallel to the stress direction become more and more, the failure mode changes from shear failure to swelling failure. (2) The shape of the cracks perpendicular to the stress direction has a significant effect on the stress-strain relationship of the expansive soil samples. The wider the crack is, the greater the closed strain is. The deeper the crack is, the smaller the initial crack initiation stress is and the greater the strength increment is. The shape of cracks parallel to the stress direction has no obvious effect on the stress-strain relationship. Based on the results, four types of curves are divided, namely, the multi-level rise, fluctuation rise, multi-level decline and standard curve. (3) The direction of crack obviously affects the strength reduction of the cracked soil. The peak strength of the sample containing cracks parallel to the stress direction is close to that of the non-cracked sample. The peak strength and deformation modulus of the sample containing cracks perpendicular to the stress direction decrease with the increasing crack number, and have a good logarithmic function relationship. The peak strain is less affected by crack. This study can provide a theoretical basis for the shallow instability mechanism of the expansive soil slope, and a guidance for the design and construction of the corresponding cracked soil engineering.

Traditional radars can detect moving targets using the Doppler effect. However, traditional radars have shadow areas in detecting the angular motion of the rotating targets. The discovery of the rotational Doppler effect based on vortex electromagnetic waves helps solve the problem of detecting the angular motion of the rotating targets under direct vision, which has attracted considerable attention from domestic and foreign scholars. In this study, we discussed the recent research progress on the rotational Doppler effect of vortex electromagnetic waves, particularly for related results in the microwave band, including the rotational Doppler effects on the target under on-axis and off-axis cases; decoupling linear Doppler, micro-Doppler and rotational Doppler effects under complex motion cases; and rotational Doppler effects on the applications of radar imaging and velocity measurement. We summarized and analyzed the existing problems demanding prompt solutions in this field, and proposed future research directions and relative applications.

Vortex electromagnetic wave carries the Orbital Angular Momentum (OAM), and thus has a spiral wavefront structure, which contains the azimuthal information of the target in the echo signal when performing radar imaging compared with planar waves. Hence, this kind of electromagnetic wave shows great potential for various applications in the field of radar detection and imaging, and it is expected to become the development direction of new-system radars. This paper describes in detail the research progress in vortex radar imaging technology in recent years. First, the characteristics of the vortex electromagnetic wave and the principles of imaging using a uniform circular array are introduced. Then the paper reviews the developmental history and research status of vortex radar imaging technology, according to three research directions: vortex radar imaging model, vortex radar gaze imaging algorithm, and vortex radar motion imaging. Finally, the prospects for the development of vortex radar imaging have been presented, along with the key scientific issues and trends for the future developments of vortex radar imaging.

For diesel transported by long-distance pipeline, its quality decreases with the increase of transportation distance. Here, the factors affecting the diesel transport quality of the Western Products Pipeline were studied through batch tracking and sampling test. The results show that the average flash point attenuation of 0# and －35# diesel at the terminal station are 5.75 ℃ and 3 ℃, respectively. Thus, the recommended minimum flash point of 0# and －35# diesel filled into the pipeline should be 66 ℃ and 48 ℃, respectively. In the case of batch transportation, the quality attenuation of diesel is relatively serious due to the great difference of the quality between adjacent oil products, and the quality of diesel attenuates more seriously with the increasing shutdown time of pipeline. Hence, the shutdown of pipeline should be minimized. The impurities in the pipeline can also result in the quality decrease of diesel transported. So, regular pipeline pigging is conductive to improving the quality of pipeline diesel. Additionally, change of pipeline flow, which is greater than the critical flow, has little influence on the quality of diesel transported. Conclusively, the research results could provide theoretical reference for economic and efficient operation of long-distance products pipelines.

Objective To investigate temporal variation in the abundance of total Vibrio parahaemolyticus in shellfish farm in Qingdao. Methods The abundance of total V. parahaemolyticus in shellfish and farm environment was detected using the most probable number real-time polymerase chain reaction(MPN real-time PCR) method during June 2013 and May 2014, and the correlation between V. parahaemolyticus abundance and the season was analyzed. Results The lowest level of MPN in shellfish and seawater presented from January to March, increased slowly before June, then rapidly rised up and reached the highest level until Augest. After that, it deceased rapidly to the lowest level in January. The lowest level of MPN in sediment was in March and the highest level kept from May to December(18 067.5-24 000.0 MPN/g), then dropped rapidly in January, and reached the lowest level in March(88.0 MPN/g). Conclusion Compared with the traditional methods, MPN-PCR has better sensitivity in reflecting V. parahaemolyticus growing trend with the the season and tempreture change. The level of bactiera abundence in oyster and seawater can be accurately assessed, which might play a role of early warning. The rapid growth time of V. parahaemolyticus in seawater and oyster is from June to Augest, which should be taken notice in the process of handling and eating to avoid foodborn disease.

In order to solve the problem of data sparseness in the task of point-of-interest recommendation and make full use of the diverse information in the location-based social network to further improve the quality of personalized recommendation, a point-of-interest recommendation algorithm integrating multiple impact factors was proposed. Geographic influence modeling and social influence modeling were performed on geographic information and social information, and temporal information and geographic information were combined to model temporal and spatial influence, and the three influence scores were integrated in a weighted summation manner to obtain user preference score. According to the user preference score, each user was provided with a recommendation list containing Top-N points of interest. The experimental results show that on the two public datasets, the point-of-interest recommendation model that integrates multiple impact factors performs better than the baselines. In addition to the user check-in frequency, the geographic-social-spatial-temporal influence is also a key part of the point-of-interest recommendation task, and the modeling of these three influences is of great significance, which provides certain reference value for the research of point-of-interest recommendation integrating key information.

With the development of orbital angular momentum (OAM) based optical communication technology,OAM router played an important role in improving the flexibility and practicability of OAM communication system.In recent years,the research on multifunctional OAM router had also been developed rapidly.Based on this,the research results on the OAM router had been summarized and reviewed,in which the principles,research methods,advantages and disadvantages of six types of OAM router,with the functions of switching,multicast,filtering,adding/extraction,and jumping and mode dividing respectively,had been discussed and summarized in details.Meanwhile,the future developing trend of the OAM router has also been prospected and discussed in the end.

In order to use lexical features (including n-gram and part of speech information) to identify more key vocabulary content in the summarization generation process to further improve the quality of the summarization, an algorithm based on sequence-to-sequence (Seq2Seq) structure and attention mechanism and combining lexical features is proposed. The input layer of the algorithm combines the part of speech vector with the word vector, which is the input of the encoder layer. The encoder layer is composed of bi-directional LSTM, and the context vector is composed of the output of the encoder and the lexical feature vector extracted from the convolution neural network. The convolutional neural network layer in the model controls the lexical information, the bi-directional LSTM controls the sentence information, and the decoder layer uses unidirectional LSTM to decode the context vector and generates the summarization. The experiments on public dataset and the self-collected dataset show that the performance of the summarization generation model considering lexical feature is better than that of the contrast model. The ROUGE-1, ROUGE-2 and ROUGE-L scores on the public dataset are improved by 0.024, 0.033 and 0.030, respectively. Therefore, the generation of summarization is not only related to the semantics and themes of the article, but also to the lexical features.The proposed model provides a certain reference value in the research of generating summarization of integrating key infromation.

The Chinese classification methods based on LSTM can correctly identify the category oftext, but such classification methods mainly focus on learning the text fragments related to the theme without aiming at other aspects of the words in context, especially the implicit feature information contained in the part-of-speech. In order to use the part-of-speech information of words effectively to learn a lot of context-dependent feature information and then improve the performance of text classification, this paper proposes a Chinese classification method combining part-of-speech information, which can easily learn implicit features from words and their part-of-speech.To verify the effectiveness of the attention-based BILSTM model with part-of-speech for Chinese classification tasks, this paper designs a series of comparative experiments and conducts on source-open dataset. The experimental results show that the attention-based BILSTMmodel with part-of-speech has better performance on Chinese classification than some baselines, and it proves that the Chinese classification model proposed in this paper is effective.This indicates that identifying the category of text is not only highly correlated with the semantic information of the words, but also has a great relationship with the information of the words' part-of-speech.

Abstract: In view of the gap between the inflow-performance forecasted by the traditional equation and the actual inflow-performance, experiments of gas-gas alternating seepage flow and gas-water alternating seepage flow were carried out, and a modified equation of inflow-performance relationship considering gas permeability change due to water invasion was established. This equation was verified by example analysis. The experimental results show that: gas permeability of net pay, when water-invaded reservoirs are converted into underground gas storage, is different from that during actual gas field development, in that the value of permeability decreases gradually in gas-water zone while the value increases gradually and even higher than that of original gas field in gas zone as a result of gas water alternative flooding in gas storage. Then one parameter describing the extent of permeability change when gas reservoirs are converted into gas storage was defined based on the result of the experiments, and eventually the traditional equation of inflow-performance relationship was modified to forecast inflow-performance in gas storage based on data from field development. Example analysis result shows that: the modified equation considers the effect of changing permeability, its results agree with the actual inflow-performance, while the result of traditional equation differs greatly from the actual inflow-performance. Key words: underground gas storage, well inflow-performance, percolation experiment, permeability change, gas-water zone, gas zone

Similar words analysis is one of the important aspects in the field of natural language processing, and it has important research and application values in text classification, machine translation and information recommendation. Focusing on the features of Sina Weibo's short text, this paper presents a language model named as POS-CBOW, which is a kind of continuous bag-of-words language model with the filtering layer and part-of-speech tagging layer. The proposed approach can adjust the word vectors' similarity according to the cosine similarity and the word vectors' part-of-speech metrics. It can also filter those similar words set on the base of the statistical analysis model. The experimental result shows that the similar words analysis algorithm based on the proposed POS-CBOW language model is better than that based on the traditional CBOW language model.

With the rapid development of social network, people can present or share their opinions and emotions on the micro-blog, and as a result, large amount of blog texts and emotional information of different topics are produced. As the traditional text mining algorithms cannot deal with these blog texts with fractional contents and different personal emotions effectively, this paper presents a novel weighting computation approach based on the key characters of micro-blog posts to get the dynamic interests of bloggers on different times. Furthermore, based on the emotion classification, the emotion transition tendency analysis can also be implemented. Experimental results show the feasibility of the presented approach.

The types of subtle buried-hill resevoirs and their accumulation characteristics are analyzed based on the exploration progression of subtle buried-hill reservoirs in the Jizhong Depression in recent years. These reservoirs are classified into buried-hill head reservoirs that are easy to discover, and buried-hill-slope reservoirs and intra-buried-hill reservoirs that are difficult to discover. Effective seal is the key factor for the formation of intra-buried-hill reservoirs, and hill controlling faults are the important hydrocarbon migration pathways. As for buried-hill-slope reservoirs, two sealing conditions must be satisfied, including argillaceous caprocks on top of the unconformities and intra-buried-hill barriers on the flank, as well as unconformities acting as primary migration pathways. Simulation experiments on petroleum migration and accumulation using a 2D experimental model show that the preferential charging location of hydrocarbons is determined by the charging force and the permeability ratio of transport pathways (faults or unconformities) to buried-hill reservoirs. When faults or unconformities are highly efficient transport pathways, hydrocarbons will preferentially charge the top of the buried hills, thus forming buried-hill head reservoirs. If faults or unconformities are poor in permeability or highly heterogeneous, hydrocarbons will charge the internal layers or slopes of buried hills, forming intra-buried-hill reserovirs or buried-hill-slope reservoirs when there are high-permeability reservoir rocks within the internal layers or slopes of the buried hills. Key words: Jizhong Depression, subtle buried-hill reservoir, accumulation mechanism, accumulation simulation

A derivative decomposed from ginkgolide B (GB) was isolated by semi-preparative chromatography. The results of the high performance liquid chromatography (HPLC) showed that the retention time of the derivative was approximately three times of that of GB with ultraviolet detection (UV), and the derivative can not exist independently without the presence of GB. The UV spectrum showed that the lamdamax of the derivative was 212.1 nm and the epsilonmax was 2.29 x 10(4), which were approximately 100 times higher than those of GB. This means a new conjugation bond has been formed and that means that the conjugation bond's pii-pi* electron jump occured in the derivative. Liquid chromatography-mass spectrometry showed the molecular ion peaks of the derivative in the positive mode and negative mode were m/z 429.1 (M + Na)+ and m/z 405.2 (M-H)-, respectively. The relative molecular mass of the derivative is 406.2. The derivative showed the same mode of fragment ion as GB in the mass spectrum, which might be due to the loss of a H2O from GB. Thermal stability of GB was greater than that of the derivative. They were both easily to be dissolved in dilute alkali. When the pH gradually increases, the ring-opening speed of the derivative is higher than that of GB. The derivative was obviously affected by solvent and higher temperatures. When GB was dissolved in PEG, the peak of the derivative disappeared at 50 degrees C for 15 h or at 120 degrees C for 4 h. Besides the main peak of GB, there appeared a small peak with a retention time of 1.2-3.0 min after heating 4 h at 120 degrees C. The result showed that the derivative was in a high energy state, and GB had a better stabilization. They coexisted and converted to each other. Under some specific conditions, the derivative could all convert to GB.

Centella asiatica (L. ) Urban is a tropical medicinal plant with a long history of therapeutic uses. Madecassic acid and terminolic acid, which are a pair of structural isomers, are two constituents of Centella asiatica. A method using reversed-phase high performance liquid chromatography in which beta-cyclodextrin (beta-CD) was the additive in mobile phase has been developed for separation of the structural isomers and determination of madecassic acid. The two compounds can be isolated with high resolution on a C18 reversed-phase column with the addition of beta-CD in the mobile phase. The separation mechanism of the isomers was discussed. It was assumed that the separation of the isomers might have been resulted from different inclusion forces of complexes with beta-CD. The effects of beta-CD concentration and the pH of mobile phase on resolution were investigated. It was found that the resolution of the isomers increased with the increase of beta-CD concentration when the mobile phase consisted of methanol-water (65 : 35, v/v) at pH 4. The correlation coefficient (r2) of the linear calibration curve between peak area and concentration of madecassic acid was 0.998 9 in the range of 0.1 - 5.0 g/L. This method was successfully used to determine the madecassic acid in triterpenic genins of Centella asiatica.