Abstract:

Abstract: Low-resource few-shot speech recognition is an urgent technical demand faced by the speech recognition industry. The framework technology for few-shot speech recognition is first briefly discussed in this article. The research progress of several important low resource speech technologies, including feature extraction, acoustic model, and resource expansion, is then highlighted. The latest advancements in deep learning technologies, such as generative adversarial networks, self-supervised representation learning, deep reinforcement learning, and meta-learning, are then focused on in order to address few-shot speech recognition on the basis of the development of continuous speech recognition framework technology. On that basis, the problems of limited complementarity, unbalanced task and model deployment faced by this technology are analyzed for the subsequent development. Finally, a summary and prospect of few-shot continuous speech recognition are given.

Abstract: Aiming at the steganalysis of color images, a deep convolutional network applied to the steganalysis of color images is proposed by introducing channel-wise convolution, multiple activation module and adversarial mechanism. Channel-wise convolution can avoid weakening irrelevant noise signals and retain additional steganographic embedded features; multiple activation modules use various activation functions to nonlinearly map convolution results and make different feedback for embedded traces to enrich the diverse expressions of embedded features; adversarial mechanisms can divide content information features and steganographic embedding features from domain categories, thereby separating additional steganographic existence features. Experiments are carried out on the PPG-LIRMM-COLOR dataset for various steganographic algorithms. The proposed algorithm is 1. 83% - 4. 99% higher performance than the best performance in the control methods. Results verify the effectiveness of the proposed color image steganalysis method.

Abstract: Aiming at the problems that the traditional sensors are not timely in detecting the fire of underground parking lot and the object detection is not effective in detecting small flame targets, an improved YOLOv5 fire detection algorithm was proposed. In order to improve the detection effect of the detection algorithm on small flame targets, small target detection layer was added to YOLOv5s network backbone; In order to enhance the expression of flame features, an interval attention structure based on CA attention mechanism was proposed; In order to improve the positioning accuracy and reduce the rate of missed target detection, GIoU was replaced by CIoU. Three groups of ablation experiments and one group of contrast experiment were designed to verify the effectiveness of this improvement. The experimental results show that the mAP0. 5 and R of the algorithm on the user-defined dataset are 92% and 96. 9%, respectively. Compared with YOLOv5s model, the mAP0. 5 of the proposed algorithm on the customized flame data set is increased by 1. 8 percentage, and R is increased by 2. 0 percentage. The weight size of the proposed algorithm is only 16. 4 MB, and the frame rate can reach 113 frames per second. It has smaller model volume and faster detection speed, and can accurately detect small flame targets, effectively improving the fire prevention capability of the underground parking lot.

Abstract: To address the problem that traditional classification models focus on majority class while ignoring minority class for classifying imbalanced data, An imbalanced data evidential classification method with composite reliability is proposed. This method improves the classification ability of the model for each imbalanced test sample by evaluating the global reliability and local reliability of the classification model. First, the method implements under sampling for majority class multiple times. The sampled subsets combine with minority class to form multiple training subsets. Multiple classification models are trained using these subsets. The maximum mean discrepancy measures the difference of data distribution before and after sampling, which can measure global reliability of the classification results obtained by classification models. Then, the local reliability of the classification result of the test sample in the training set is evaluated by using its nearest neighbors. The test sample and its nearest neighbors have similar data distribution and spatial structure. The smaller the deviation between the classification result of the classification model and the ground truth, the greater the local reliability of the classification result obtained by the classification model. Finally, under the framework of evidential reasoning, the global reliability and local reliability are combined as composite reliability factors to discount the classification results obtained from different classification models. Partial probability values are assigned to completely unknown classes to represent the uncertainty of classes. Dempster-Shafer (DS) rule is employed to fuse the classification results after multiple discounts for decision analysis. The experimental results show that the average FM and GM of the proposed method for the classification of 12 imbalanced data in KEEL and UCI database are 80. 18% and 87. 24%, respectively, which are 8. 1% and 4. 99% higher than those of other best imbalanced data classification methods, respectively. This proves the effectiveness of the proposed method in classifying imbalanced data.

Abstract: In order to grasp the variation law of secondary flow in helical tube, the local intensity and development law of secondary flow in helical tube were studied, and the dimensionless helical intensity H was proposed as the local secondary flow intensity index. By means of numerical simulation, the average volume of H H m is taken and compared with the Dean number describing the overall secondary flow intensity. The results show that under different Reynolds numbers, curvature and pitch, the correlation coefficient between H m and Dean number is about 1, and there is a good proportional relationship, indicating that H can accurately measure the intensity of local secondary flow. At the same time, the influence of geometric parameters on the development law of secondary flow in the helical tube is explored. According to the intensity of secondary flow, the flow in the helical tube can be divided into development stage, transition stage and stable stage. The secondary flow increases rapidly in the development stage, reaches its maximum value in the transition stage, and fluctuates periodically in the stable stage. The increase of Reynolds number will significantly increase the intensity of secondary flow, but will not affect the location of secondary flow three stages. The increase of curvature will slightly increase the intensity of the secondary flow and extend the distance of the secondary flow to the stable stage. The pitch has little influence on the intensity and development of secondary flow.

Abstract: A typical three-dimensional model of MOLB-type SOFC was developed, and the solution was based on the finite element method. In view of the problem of uneven distribution of reactants in porous electrode and limited output performance, the effects of rib widths and cathode thicknesses on the mole fraction distribution of gas, temperature distribution, performance curve and polarization of the MOLB-type SOFC were investigated with the aim of improving the mass transfer in the porous electrode and enhancing the output power. The results showed that the ratio of the rib width to the cathode thickness λ has an important effect on the mass transfer of gas in porous electrode and the output performance of the cell. The smaller rib width is conducive to the full diffusion of the reactants in the porous electrode, which can improve the uniformity of gas distribution, reduce the concentration polarization of the cell, and improve the output performance of SOFC. As the cathode thickness increases, the temperature of SOFC increases, which reduces the activation polarization and improves the performance of SOFC. Considering the synergistic effect of rib width and cathode thickness on SOFC performance, it is found that a small value of λ can improve the uniformity of the gas distribution in the electrode, reduce the total polarization loss of SOFC, and improve the output performance of the cell. When λ decreases from 10 to 3. 33, the maximum output power density of SOFC increases by 53. 68%.

Abstract: Taking the high-speed pin insertion machine as the research object, the mechanical analysis of its pin insertion mechanism is carried out, the cylindrical cam groove surface equation is established. The contact mechanics model of the cylindrical cam roller and the groove, and the geometric model of the roller rolling in the groove are established. Based on Hertz contact theory, the maximum contact stress of the contact point between roller and groove is calculated at each moment in the course of the cam rotation. Based on the Miner linear fatigue cumulative damage theory, the dangerous point in the circumferential direction of the roller is judged, and the dangerous point is taken as the object to realize the quantitative calculation of the fatigue life of the roller. Through a specific example, the influence of roller diameter, cam diameter, contact length, and cam rotation speed on the contact fatigue life of the cam roller is illustrated. The calculation results show that: when the diameter of the cam roller increases from 10 mm to 26 mm, the fatigue life increases by 14. 1 times; when the diameter of the cam increases from 50 mm to 140 mm, the fatigue life increases by 7. 5 times; when the contact length increases from 5 mm to 15 mm, the fatigue life of the cam roller increases by 26. 0 times; when the cam speed increases from 600 r·min - 1 to 1 300 r·min - 1 , the fatigue life decreases to 0. 449% of the original.

Abstract: Aiming at the problem of low temperature corrosion on the surface of the heat exchanger in the waste incineration system, the corrosion process of sulfuric acid molecules on the surface of nickel alloy coatings was simulated by molecular dynamics research methods, the corrosion rates of different coatings were explored, and the corrosion process model was established. The results show that H2 SO4 dissociates two hydroxide ions on the surface of nickel metal coating, and the decomposed O—H ions and O—S—O ions are captured by the Surface of Ni metal seperatively. O in O—H ions forms bond chains with Ni atoms only. Compared with aluminum oxide, nickel alloy and nickel chromium alloy, nickel chromium alloy has the best inhibition effect on the initial slagging layer, while nickel alloy coating has the worst inhibition effect. The selection of coating materials with strong oxidation resistance can effectively reduce the phenomenon of low temperature corrosion on the surface of the heating surface of the waste incineration system.

Abstract: In order to improve the detection ability of tiny defects at the interface of diffusion welding of titanium alloys, an ultrasonic C image fusion algorithm based on wavelet transform was proposed. Firstly, the ultrasonic C image to be fused is decomposed into low-frequency and high-frequency parts by wavelet transform, and the high and low frequency coefficients are differentiated and fused according to the different characteristics of the high and low frequency coefficients, and the inverse wavelet transform is used for the fused coefficients to get the fused image. Finally, in order to improve the contrast of the image and enrich the details of the image, the improved homomorphic filter is used to enhance the fusion image, so as to obtain the fusion result image. By preparing different types of artificial defect samples for testing, and quantitatively comparing the length of defects detected by the algorithm in this paper and conventional ultrasonic C scan, the experimental results show that for small linear defects and weak bond defects, the ultrasonic C reconstructed by the algorithm in this paper can be used. The average errors of the images were 2 mm and 4. 2 mm, respectively, and the average errors of conventional ultrasound C scan was 8 mm and 9. 5 mm, respectively. Therefore, the ultrasonic C image reconstructed by the proposed algorithm in this paper can reflect the defect information more accurately.

Abstract: The problem of source number estimation was a key issue in blind source separation (BSS) , because the number of sources directly affected the effect of BSS. To solve this problem, this paper proposed a single-channel blind source number estimation algorithm that took the information entropy as the statistical evaluation index, and used the information entropy to measure the information quantity of the source signal to determine the source number. To calculate the information entropy of the estimated source signals, firstly, the Gaussian mixture model (GMM) was used to fit their distributions; Secondly, samples obeying the target distribution were sampled and the entropy is calculated based on the Markov chain Monte Carlo (MCMC) algorithm; Finally, the source number was obtained by minimizing the average information entropy of the source signal. A series of experiments based on simulation data and real communication data show that the proposed algorithm has strong robustness and can estimate the number of sources with 94% accuracy, thus verifying the effectiveness of the algorithm.

Abstract: The multi-sensor scheduling problem of dual unmanned flight platforms has a complex coupling relationship, and how to reasonably match flight segments, sensors and targets is an intricate part of the problem. To solve the multi-sensor scheduling problem of dual unmanned flight platforms, firstly, this paper proposes a flight segment task allocation mechanism, which assigns tasks through the position relationship between the flight segment and the target. This mechanism can effectively split the tasks and solve the task coupling problem of the two platforms. Secondly, pre-processing targets and sensors are carried out to identify scarce sensor resources and independent targets in advance to guide subsequent scheduling matching. Then, sensor-target matching is carried out to plan the scheduling of sensors based on the task revenue. The scheduling of conflicting segments is adjusted according to the availability of sensor resources. Finally, task migration is carried out for targets who still need to complete their assigned tasks. The available free sensors in the corresponding flight segments are matched according to the target task remaining to improve the task revenue. In order to verify the effectiveness of the proposed algorithm in this paper, existing algorithms were compared through simulation experiments on 10 test problems. The experimental results show that this algorithm outperforms existing algorithms in task revenue and stability with a significant reduction in running time.

Abstract: Catalytic combustion is one of the most effective purification technologies to remove soot particles from diesel engine exhaust. In order to improve the mobility of atoms on the catalyst surface and the low-temperature oxidation-reduction ability of the catalyst, the catalytic combustion performance of soot was studied by using alkali metal K-supported SnxCe1- xO2 catalyst. The SnxCe1- xO2 catalyst was obtained by doping Sn and Ce according to the ratio of different substances by co-precipitation method. The activity of the catalyst was evaluated in a micro fixed bed temperature programmed reactor, and it was found that the catalyst had the best catalytic performance when n Sn ∶n Ce = 4 ∶6. Then, a series of xK / Sn-Ce catalysts with higher activity were prepared by impregnation, and characterized by XRD, XPS, H2 -TPR, and O2 -TPD. The results showed that the concentration of active oxygen on the surface of loading K was increased, the oxygen transfer rate was accelerated, and the redox ability was improved. When the K loading is 15% ( x = 15%) , the catalyst has the best removal effect on soot, the selectivity of CO2 in the product reaches 94. 3%, and the T90 temperature is 388 ℃ . This research provides an important reference for the design of efficient soot combustion catalysts.

Abstract: In view of the development demand of biomaterials for bone substitute, the development status of polymer materials, metal materials, bioceramic materials and composite materials for bone replacement were summarized, and the advantages of bioceramic materials in mechanical properties, chemical properties and biological properties as bone substitute materials were emphatically introduced. The bioceramic materials play an important role in the field of bone substitute materials due to their similar density and composition to human bone, significant biocompatibility, high mechanical strength, and stable chemical properties. In this paper, the research status of bioceramic materials represented by phosphate, silicate, oxides and non-oxides is systematically discussed, and the advantages and limitations of bioceramic materials and their composites as bone substitute biomaterials are discussed. Finally, the future development direction of bioceramic materials is prospected.

Abstract: In the Pidgeon magnesium smelting process, mastering the laws of magnesium vapor flow and crystallization is the key to obtaining high-quality crystalline magnesium. In this paper, the flow characteristic of magnesium vapor in the compound vertical retort from the reduction reaction zone to the crystallization zone was studied. The calculation method of flow resistance was given by introducing the concept of conductance in hightemperature rarefied gas. The relationship between flow resistance and the retort inner diameter, retort size, vapor pressure was deduced. It is shown that the crystallization process of magnesium vapor is not only affected by temperature and pressure, but also related to the diameter of the crystallizer. Appropriately reducing the inner diameter of the crystallizer can obtain crystalline magnesium with higher density. The total pressure loss of magnesium vapor in the compound vertical retort from the reaction zone to crystallization zone is about 150 Pa, and the average working pressure in the reaction zone is higher than two order of magnitude. In the early stage of magnesium smelting, when the temperature of part of the pellets is lower than 600 ℃ , the magnesium vapor moves in the reaction zone in a “condensation-sublimation” manner. As the temperature of the overall pellet rises above 600 ℃ , the flow resistance of magnesium vapor decreases, and it can smoothly enter the crystallizer for crystallization

Abstract: Aiming at the problem that ciprofloxacin ( CIP ) generated a large amount of ciprofloxacin-containing wastewater in the production and use process, an nitrogen-doped activated carbon (NAC) was prepared from shaddock peel by hydrothermal carbonization and activation with potassium hydroxide to treat ciprofloxacin-containing wastewater. The Box-Behnken central composite design( BBD) of response surface methodology was used to optimize the adsorption conditions of NAC. The result showed that the equilibrium data were perfectly represented by Langmuir and Koble-Corrigan isotherms, and the adsorption process was precisely described by the pseudo-secondorder kinetic model. Besides, the adsorption of CIP on NAC was mainly controlled by hydrogen-bonding, π-π electron-donor-acceptor(EDA) interaction, hydrophobic effect, electrostatic interaction and Lewis acid-base effect. The maximum monolayer adsorption capacity of CIP was 752. 05 mg / g at 298 K, implying that NAC was a promising adsorbent for the removal of CIP from aqueous solution.

Abstract: In order to improve the performance of magnesia dolomite materials, stabilized magnesium dolomite refractory materials whose iron oxide content ( mass fraction) was 0, 0. 5%, 1. 0%, 1. 5%, respectively, were prepared from natural dolomite, high silicon magnesite, silica and iron oxide powder by two-stage calcination and slaking process. The influence of iron oxide content on the physical properties, hydration resistance and coatability adherence of the samples was investigated. The phase composition and microstructure of the specimens were characterized. The results show that, as the iron oxide content increases, the room temperature flexural strength and the room temperature compressive strength increase, the high temperature flexural strength and the adherence strength decrease, and the retention rate of the thermal shock strength of the sample first increases and then decreases. With the iron oxide content increasing to 1. 5%, the hydration weight gain rate of the sample decreases to 0. 12%, and the hydration resistance improves. According to the XRD spectrum analysis, the main crystal phases of the sample are periclase, dicalcium silicate and tricalcium silicate. With the increase of iron oxide content, the diffraction peaks of C2 S and C3 S become stronger and the peak shapes are sharp. It shows that iron oxide can promote the growth of C2 S and C3 S crystals and increase the density of the sample.

Abstract: Taking the yield and purity of baicalin as evaluation indicators, the acid precipitation method, alkali-soluble acid precipitation method, recrystallization method and pharmacopeia purification method were used to purify baicalin, and the best purification method was determined as alkali-soluble acid precipitation method. Plackett-Burman test was used to obtain the key process parameters in the alkali-soluble acid precipitation method, and the regression model between the key process parameters and the key evaluation indicators was established by the Box-Behnken response surface test, and the design space was obtained and verified. The key parameters obtained by screening were the liquid concentration, alkali solution temperature, alkali solution pH, acid precipitation temperature and acid precipitation pH. When the yield and purity of baicalin were both greater than 85%, the operating space were: the alkali solution temperature was 40 ℃ , the pH of alkali solution was 8. 0, the temperature of acid precipitation was 80 ℃ , the concentration of liquid medicine was 11. 53 ~ 14 mg·mL - 1 , and the pH of acid precipitation was 1. 17 ~ 2. 33. The verification results show that the design space was stable and reliable, and it could provide a reference for industrial production of Scutellaria baicalensis extract.

Abstract: In order to study the mechanical properties of the composite pavement, a finite element model of the composite pavement is established based on the overlay reconstruction project of an airport in the central plains region. Through the established composite pavement monitoring system, the mechanical response of the pavement under the action of the target aircraft type is extracted and compared with the simulation data to verify the rationality of the model. Under the action of moving load, the mechanical response of pavement considering different types of aircraft, different interlayer bonding states and different horizontal forces is analyzed. It is found that under the action of B737-800, B767-300ER and B777-300ER, the maximum shear stress of asphalt overlay appears in the upper layer, and with the increase of the number of rounds, the mechanical index of pavement gradually increases, in which the transverse tensile strain at the bottom of asphalt layer and the tensile stress at the bottom of cement concrete slab change significantly; With the increase of interlayer bonding coefficient, the pavement mechanical index shows a decreasing trend, and its influence degree gradually decreases along the pavement depth direction; When there is emergency horizontal braking force, the maximum shear stress between asphalt layers increases by 11. 84%, the maximum transverse tensile strain at the bottom of asphalt layer increases by 6. 05%, and the longitudinal tensile strain increases by 9. 68%.

Abstract: The purpose of this study was to reveal the effect of salt content and initial water content on the thermalphysical properties of silty clay, the freezing temperature and thermal conductivity of silty clay with distinct salt con-tent and water content were measured by indoor temperature test ， and the effects of distinct factors on freezing tem-perature and thermal conductivity of silty clay with distinct salt content and water content were revealed.The resultsdemonstrated that:The freezing temperature and supercooling temperature of NaCl salty silty clay permafrost dimin-ished linearly with the increase of salt content，while the freezing temperature and supercooling temperature ofNa,S0。salty silty clay permafrost first diminished and then increased slightly，and the thermal conductivity of bothsalty silty clays diminished with the augment of salt content.At the moment of the initial water cut was inferior to thesaturated water cut , there was a positive correlation between initial moisture content and freezing temperature andthermal conductivity , and the moisture rate was the dominant element that could affect the freezing temperature , andafter it surpassed the saturated water cut，the salt content became the dominant element. When the rate of waterclose to the liquid limit，the thermal conductivity of the two saline soils increased tardily. At the same time，theempirical formulas of freezing temperature and thermal conductivity in different conditions were given， and the wa-ter-salt-temperature variables were linked with freezing temperature and thermal conductivity , which could provide areference for engineering construction in salty soil distribution areas.