A fracture failure at the root of an aeroengine combustor fuel nozzle supply tube was occurred during the flight test. In order to figure out the cause of the fault and improve the reliability and safety of the combustor fuel nozzle supply tube, fault analysis is performed. The results show that the main reason for the fracture of the fuel nozzle supply tube in the combustor is the large initial stress caused by the assembly. During engine operation, the superposition of the assembly stress and the thermal expansion stress lead to the large static stress, which finally resulting fatigue fracture under the engine vibration. According to the causes of the failure, various improvement measures including structural improvement, technology improvement, assembly process improvement and other improvement measures are employed, which greatly reduced the assembly stress and effectively avoided the recurrence of the same kind of failure. Fault return-to-zero is completed.

In order to illustrate the characteristics of combustion and emission in the dual fuel aero-engine, the combustion process in the combustor of an aero-engine which respectively burning kerosene and the mixed fuel including 50% kerosene and 50% methane were reaction kinetic simulated by adopting the reduced reaction mechanism of n-decane which was chosen as a surrogate fuel for kerosene. Furthermore, the characteristics of combustion and the main emissions formation of this aero-engine which respectively burning two different fuels were contrastive analyzed. The results show that, compared with the condition that the aero-engine burning kerosene, the flow distribution characteristics in a tube of the combustor is the same as the condition that the aero-engine burning 50% kerosene and 50% methane. However, the high temperature area in the main combustion zone is shorted, and the maximum temperature in the flame tube and the average temperature at the outlet of this flame tube are decreased by 15K and 11K, respectively. Moreover, compared with the condition that the aero-engine burning kerosene, the concentration field distribution characteristics of fuel is the same as the condition that the aero-engine burning 50% kerosene and 50% methane. However, the maximum concentration of fuel is increased, the maximum concentrations of C2H2, C2H4, C4H4 are decreased greatly, and the average concentrations of CO2 and CO at the outlet of this flame tube are decreased by 12% and 33%, respectively, but the average concentration of H2O is increased by 13%.

in order to more realistically simulate the controlled deflection motion of ailerons, this paper takes a 2D aileron as an example. The dynamic overset grid technology is used to coupled solve the unsteady Navier-Stokes (NS) equation in the aileron deflection controlled by proportional-integral- derivative (PID) feedback, and the flow field characteristics during the aileron deflection are obtained. First, the steady and unsteady aerodynamic characteristics of the NACA0012 airfoil were simulated using the dynamic overset grid solution algorithm and the Fluent software package, and then the dynamic overset grid solution algorithm was used to perform the PID-controlled 2D aileron deflection motion. Numerical simulation results show that the numerical simulation method is reliable and effective. This method can provide a reference for the design of the control algorithm and can reflect the unsteady flow field sequence changes of the aileron controlled motion. A direct numerical simulation method combining 2D aileron controlled motion simulation and wing flow field solution is initially established, which provides a dynamic solution algorithm for more realistic simulation of the flow field characteristics of moving objects.

In order to make the formation flight of space solar power stations can be reconstructed, this paper first derives the space solar power station formation movement model on the geostationary orbit based on the synchronous root number difference method, and designs a suitable space solar power station formation configuration to ensure that it can maintain normal energy transmission in any environment. Secondly, the simulation of formation reconfiguration was carried out in combination with the traditional loop tracking control algorithm. And an improved loop tracking control algorithm is designed, and simulation verification is carried out. The simulation results show that the improved loop tracking control algorithm can make the reconstructed space solar power station formation configuration not limited to the center point of the initial state. Finally, it is verified by simulation that the flexible body in the space solar power station has little vibration during the orbit change, and has little effect on the space solar power station orbit change and can be ignored.

基于2008-2019年我国制造业企业微观面板数据，实证研究中美贸易摩擦对我国制造业企业技术创新的影响，进而探究中美贸易摩擦对技术创新的影响路径，进一步研究了中美贸易摩擦对高技术和低技术企业技术创新影响的差异。研究发现，中美贸易摩擦在短时间内会抑制我国制造业技术创新，但长期会倒逼企业增加研发投入，增强创新意识，进而促进技术创新；中美贸易摩擦主要通过企业贸易成本增加，导致企业利润下滑，研发投入减少，进而抑制企业技术创新；相比低技术制造业企业，中美贸易摩擦对高技术制造业技术创新的影响更加显著。

Abstract: In order to explore the pyrolysis characteristics and thermal stability of different forms of T800 multi-layer carbon fiber/epoxy resin composites, a thermogravimetric-differential thermal synchronous analyzer was used to perform different heating rates on the bulk and powder samples in a nitrogen atmosphere. The program temperature control process obtained the influence of the heating rate on the mass loss of the two types of samples, and carried out the pyrolysis kinetic analysis. The results show that there is only one stage of pyrolysis of the bulk and powder samples, that is, the pyrolysis of epoxy resin under anaerobic conditions. The initial decomposition temperature, termination decomposition temperature, and maximum weight loss rate temperature of the two forms of samples increased with the increase of the heating rate, the pyrolysis temperature range became wider, and the peak weight loss rate became larger. Using Kissinger analysis method and Flynn-wall-Qzawa analysis method for kinetic calculations, the results are basically the same. The apparent activation energy of the two forms of samples gradually increase with the progress of the reaction, and the difficulty of the reaction increases; compared with the bulk sample The powdered sample is slightly stable in the initial stage of pyrolysis, but the stability of the two is basically the same during the whole process.

The issue of vibration sensors placement of space solar power station, which has the characteristic of large size, low stiffness and dense frequency, was studied. Based on the modal assurance method (MAC), the optimal sensor placement scheme for this kind of large space structure was carried out referring to the Fisher information matrix, modal confidence criterion and other evaluation indexes. Moreover, the computation results of sensor placement of the algorithm were compared with these of the effective independence (EI) approach and modal kinetic energy (MKE) method to find the optimal solution of sensor placement problem of this kind of large space structure. The results show that compared with the EI and MKE methods, the sensor location calculated by MAC method has better modal distinction and the measurement points are scattered to reduce the redundant configuration. The eigensystem realization algorithm (ERA) was used to identify the modal parameters based on the aforementioned optimization results. The calculation results illustrate that the sensor placement scheme, which is based on the MAC method, has good ability in parameter identification and is suitable for the output signal acquisition in the parameter identification of space solar power station.

Electrochemical machining is a widely used non-traditional processing technology, which has the characteristics of simplicity, high efficiency, safety and environmental protection. In order to optimize the electrochemical machining parameters, a multi-objective optimization method based on the combination of neural network and genetic algorithm was proposed. Artificial neural network and back propagation algorithm are used to build a prediction model for the test. According to the errors of training samples and test samples, the prediction accuracy of the network model is analyzed. The neural network model was optimized by using multi-objective optimization method, non-dominated sorting genetic algorithm II. The experimental results show that the model has high applicability and accuracy in predicting the surface hydrophobicity of TA15. The prediction error rate of the best result is only 8.0293%, and a Pareto optimal set is predicted. It can be used to predict and analyze the surface hydrophobicity of processed materials in actual processing.

Considering the multi-parameter, uncertainty and conflict of aero-engine condition monitoring data, a health condition assessment model based on D-S evidence theory and RBF neural network is established. Firstly, the key sensor parameters were selected, normalized and quantified, and the weights of each parameter were calculated. Secondly, the membership state was trained as the input of the neural network to get the output of each sub-module. According to the information fusion and decision-making rules, the specific health status grades were obtained. Finally, an example was given to verify the feasibility and effectiveness of the method in order to improve the aviation safety level.

At present, China's rural areas have long entered the era of rural revitalization strategy, and have achieved comprehensive poverty alleviation. According to the government work report on March 5, 2021, a five-year transition period will be established for counties that have been lifted out of poverty, so as to ensure that Poverty-Stricken counties that have been lifted out of poverty aligns with poverty alleviation and alleviation work in an orderly manner. Rural tourism has become an effective way to eliminate poverty and realize rural revitalization because of its drive to rural economy. Based on the rural tourism has rich resources of poverty county - Wei country as the research object, focusing on residents' perception of the micro perspective, rural tourism boost rural revitalization of Wei country economic performance, social and cultural performance, environmental performance, three dimensions research, using factor analysis to the Wei country rural tourism boost rural revitalization of the carding performance analysis and empirical research. The study on rural tourism performance evaluation and promotion path of poverty-stricken counties under the background of rural revitalization strategy is of great significance for exploring the orderly connection between poverty alleviation and rural revitalization of poverty-stricken counties, promoting the upgrading and adjustment of rural tourism industrial structure, and finally realizing rural revitalization.

In order to improve the accuracy of machine tool spindle reliability prediction, an improved algorithm based on optimized cascade feedforward neural network for CNC machine tool spindle reliability prediction is proposed. Combine the mini-batch gradient descent algorithm with the cascade feedforward neural network prediction method to optimize the neural network to increase the accuracy of prediction. Perform a preliminary analysis on the collected spindle reliability data, select the first nth reliability data and the fault time point t corresponding to the (n+1)th reliability data as the input variables of the neural network, and the (n+1)th reliability data is used as the output variable to complete the training of the reliability prediction model and the error comparison analysis of the test set data. The simulation analysis result of the example shows that the absolute value of the maximum relative error of the reliability prediction value obtained by this method is 2.41%, which is less than 3%, indicating that the prediction method has high accuracy. The maximum relative error of the prediction results obtained by other prediction methods such as BP neural network is greater than 3% or even more than 10%. Compared with them, this method can achieve more accurate prediction of CNC machine tool spindles and provide a theoretical basis for studying the reliability of CNC machine tools.

Human body tracking and abnormal behavior recognition algorithms based on deep learning need to extract features of human targets in the process of implementation. In order to optimize the twice repeated feature extraction process and improve the work efficiency of the network, a joint human tracking and abnormal behavior recognition algorithm is proposed. The real-time multi-domain convolutional neural network (RT-MDNet) is used to track human targets. Different abnormal behaviors are used as different operating domains for human tracking, and extract common features of human targets to achieve real-time human tracking. At the same time, the high-dimensional feature map output by the convolutional layer of the RT-MDNet network is extracted, and the feature map is combined with the long and short-term memory network (LSTM) to realize the classification of abnormal behaviors by grasping the timing information between features. Six abnormal behaviors were selected from the CAISA behavior analysis data set provided by the Chinese Academy of Sciences to train and test the network model. The experimental results show that the model can accurately track human targets in real time, and can also classify the behavior of the tracking targets, with a recognition rate of 88%. The combined algorithm of human tracking and abnormal behavior recognition based on deep learning uses the characteristics of human targets for tracking and recognition, realizing the organic combination of tracking and recognition.

High pressure water jet is a high-tech cleaning technology that has been widely concerned in the industry in recent years. Compared with the traditional cleaning method, it can clean the parts with varied shapes and complex scaling, and can be carried out flexibly in the occasion of narrow space and complex environment. By means of high pressure water jet structure and related parameters were analyzed, and the simulation of the water jet impact carbon deposition layer numerical model, identified the high pressure water jet cleaning method, the principle of carbon deposition layer, explore the use of high pressure water jet all-round cleaning carbon layer technology applications and the feasibility of research under the effect of water jet, clean up the whole process of carbon deposition layer of basic theory and the test method. A gun chamber cleaning machine based on high pressure water jet is designed to clean carbon, copper and nylon in all directions in the gun chamber, and can achieve the reciprocating movement of telescopic rotation in the gun chamber. The design scheme, structure and working principle of the gun chamber cleaning device are emphasized. The feasibility of using water jet to clean carbon deposition was verified by designing corresponding experiments, and the conclusions can be used as the basis for the design and optimization of cleaning process and related equipment of precision parts by high pressure water jet.

Honeycomb aluminum core sandwich composites are widely used in aerospace field because of their excellent mechanical properties, good heat insulation and shock absorption properties. However, the complex structure forms bring difficulties to the follow-up structural health monitoring and residual life assessment. The inverse Finite Element Method (iFEM) has been proved be well suited for the shape sensing and full-field reconstruction of stress and strain. In practical application, the available number of sensors is limited by various factors such as cost. Therefore, reasonable selection of the number and location of sensors is the key parameter for accurate reconstruction of structural deformation. By comparing the displacement reconstruction results of three different strain layout schemes on the refined honeycomb plate model and the homogeneous sandwich plate model, this paper seeks the best honeycomb plate reconstruction strategy. The numerical results show that based on the homogeneous sandwich plate model, the deformation of honeycomb plate can also be accurately reconstructed by using a small amount of strain data (scheme 3), while the refined honeycomb plate model requires a large amount of strain input

Abstract：The study of cervical cell segmentation is import to the diagnosis of cervical cancer, and a large number of overlapping cells have brought great challenges to this research. This paper proposes a new cervical cell instance segmentation method to solve this problem. This method initializes the contour for each cell based on the result of semantic segmentation and establishes an energy function for each contour. The contour evolution is completed by minimizing the energy function, which solves the problem of segmentation of overlapping cells in TCT images. First, improve the U-net neural network to complete the precise semantic segmentation of cervical cell images, and divide the image pixels into three categories: cells, nuclei, and background. Secondly, on the basis of semantic segmentation, identify free lying cells and cell clumps in the image and directly extract the contours of the free lying cells. Finally, for the cell clumps containing overlapping cells, the Snake active contour model is used to extract the contour of each cell to achieve cell instance segmentation. The experimental results show that the proposed algorithm can perform effective instance segmentation on cervical cell images containing overlapping cells.

Abstract: In view of the cross-view scene, pedestrian pictures are affected by factors such as illumination, chromaticity changes, style conversion, etc., and the model obtained by person re-identification learning cannot well express the appearance characteristics of pedestrians. This paper modifies the commonly used classification network skeleton Resnet. The introduction of IBN block to deal with the problem of image style conversion. At the same time, a data enhancement scheme of random grayscale erasure is proposed. On the premise of not changing the original image structure information, the gray patch image is used to replace the original image, which effectively improves the impact of illumination on the person re-recognition model. Secondly, Circle Loss is used in model training to replace the linear combination of SoftMax Loss and Triplet Loss, and Center Loss is added to enhance the cohesion of features, while combining warm-up and cosine annealing strategies to dynamically change the learning rate. Finally, the knowledge distillation method is used to learn a smaller model to complete accelerated inference without reducing the accuracy. Experiments show that our proposed method can effectively improve the convergence speed of the model and complete fast cross-domain pedestrian retrieval in real applications. Keyword: Person re-identification; IBN block; Random grayscale erasure; Training strategy

[Abstract]Under the changes of the dynamic environment, SMEs are facing greater crises and challenges than ever before. The 14th Five-Year Plan clearly pointed out that it is necessary to support innovative small, medium and micro enterprises to become the birthplace of innovation, promote small and medium-sized enterprises to enhance their professional advantages, and cultivate specialized and new "small giant" enterprises and manufacturing single champion enterprises. Under the favorable external environment, it remains to be seen whether the growth of the small giant enterprises in Liaoning Province, which is prominent in the manufacturing industry, is good. Whether the small giant enterprises in Liaoning Province succeeds or loses in this battlefield without gunpowder has become the focus of public vision. point. This article first uses the principal component analysis method to find out the main factors affecting the growth index and constructs a growth index model for small and medium-sized enterprises, and then calculates on this basis. Secondly, it summarizes the high-growth enterprises from the two aspects of enterprise scale and industry, and provides ideas for the future development of the small giant enterprises in Liaoning Province from the external environment and internal conditions of enterprise development. [Keywords] small giant enterprises; corporate growth; principal component analysis

Quantum neural network (QNN) is a research field that combines traditional artificial neural network (ANN) with quantum information and quantum computing. QNN has the advantages of both, which is of theoretical importance to improve the shortcomings of traditional ANN and to essentially improve the performance of neural network. In this paper, we propose an optimized variational quantum neural network (VQNN) model based on the ANN model by combining quantum theory and quantum mechanical concepts such as quantum parallel computing, quantum gate circuits, and variational quantum circuits, which is a quantum-classical hybrid computing model consisting of quantum circuits that can operate on Noisy Intermediate-Scale Quantum (NISQ) devices combined with machine learning (ML) strategies. The quantum circuits consist of two parts: the quantum state encoding circuits for encoding classical data into quantum state data; and the variational quantum circuits (VQC) which simulates the classical probability distribution of the learning target and encodes the information into the quantum circuits parameters. Finally, the classical probability output distribution is obtained by measuring the VQC quantum state output, and the parameters of the variational quantum circuits are optimized by using a classical computer. This structure makes it easy to integrate VQC with classical ML. Further, we explored the use of VQNN to build a classifier based on practical applications to apply it in the field of network attack detection. The experimental results show that for the KDD CUP 99 dataset, VQNN has relatively high detection performance, both of which are higher than other classical and the quantum gate circuits neural network model detection comparison models. In addition, this VQNN can be deployed in most of the recent NISQ devices. Also, our proposed VQNN is the first model that can be deployed in NISQ for network intrusion detection.

In this paper, the load calibration test technology of aero-engine multiple tie rod structure was studied, through the finite element analysis, four strain gage combination test schemes were selected as the load calibration strain values, and the effect of the stability of strain data under multiple loading, the load position offset and the connecting pin tightening torque on load calibration strain data were verified by load calibration test. Test result shows that the load-strain curves of the test scheme show the characteristics of nonlinear segment and linear segment combination, and the characteristic curve can be fitted by piecewise function; using the total of single ear side strain values as load calibration strain has higher load calibration accuracy, its deviation between the characteristic curve fitting value and the test measurement value is less than±1%, and the influence of the load position offset and the connecting pin tightening torque is also less than±1%.

In view of the difficulty of flanging the GH536 combustion chamber cap and controlling the stress-strain distribution, the flanging forming simulation research is carried out. According to the different punch feed rates and friction factors, three groups of finite element models for flanging forming are established. The influence of the punch feed speed and friction factors on stress and plastic strain distribution of GH536 superalloy cap flanging is explored, the results show that when the punch feed rate is too high, the annular convex and flanging hole connections area are in higher average stress field, cap ring bulge area in the region of the irregular hole connect produces larger bore hole deformation, the plastic strain and the degree of deformation distortion of special-shaped hole is high, and the flanging of the irregular hole is not obvious. When the contact friction coefficient increases, higher shear stress around the irregular hole area is produced, causing the front edge of the irregular hole to be in an excessively high stress state. Distortion such as wrinkles appears in local area. The stress and strain distribution can be decreased by decreasing punch feed speed and friction coefficient, reducing inhomogeneous deformation conditions, and the flanging forming quality is improved.

To evaluate the flow resistance in downdraft free jet wind tunnel, the flow network method is used to simplify the tunnel system into a fluid network composed of typical elements. The flow resistance of each element is calculated by theoretical or empirical method. According to the design standard of wind tunnel, the structural dimensional of the main components are obtained. The calculation results show the variation law of aerodynamics parameters in the system. The flow resistance in the expansion section is the largest. By setting different inlet pressures, the corresponding flow rate and wind speed in the test section are calculated. Combined with the performance curve of the selected fan, at rated fan power, the mach number in the wind tunnel can reach 0.48. The calculated results can provide support for the selection of air source system in the further improvement of test capability.

New energy electric airplanes have received more and more attention. Electric airplanes need to use high-power, high-energy-density batteries as the electric propulsion system. High-power batteries have become one of the main sources of magnetic interference for electric airplanes. In this paper, a Maxwell 3D modeling simulation study is carried out for the magnetic field interference of the high-power battery of the electric aircraft. On the basis of simulation analysis, the RX4E electric aircraft and experimental equipment are used to build a test platform, and to conduct experimental tests on the magnetic field interference of the airborne power system. Through the research and analysis of the experimental data, it is concluded that the magnetic field interference of the high-power battery of the electric aircraft will cause deviations in the operation of the aircraft instrument panel, the on-board magnetometer and the magnetic compass. For the electromagnetic compatibility design of electric aircraft, the research results of this paper have good reference value.

The algorithm of generating inspection snap-gauge is the core of rapid design of composite Skin parts inspection fixture, In the modeling process , the workload of tooling design is very heavy, the design quality varies from person to person and is unstable, in order to solve the above situation, according to the characteristics of inspection card and air groove which depend on the upper boundary and positioning of skin parts , this paper puts forward the recognition method of boundary and positioning features. Meanwhile the design method of gas tank is put forward according to the principle of vacuum adsorption. The application of divide-and-conquer strategy to the calculation of test snap-gauge layout is studied. Design the inspection snap-gauge intelligent avoidance principle. And expounds the judgment method of checking the cutting direction of the inspection snap-gauge. In CATIA environment, the intelligent and rapid design system of composite Skin parts inspection fixture was developed. The examples show that the above discussion provides an effective method for the intelligent and rapid design of composite Skin parts inspection fixture.

To improve the frame rate of streaming video and solve the problem of complex video interpolation algorithm model, a lightweight video interpolation algorithm based on optical flow estimation, referred to as SKFEVI, is proposed. The algorithm only needs to perform optical flow estimation once to calculate the intermediate stream information of two video frames directly, which improves the estimation speed of the intermediate stream, and on this basis, it introduces the attention (SKNet) mechanism for feature fusion, allowing the model to integrate the feature information generated at different scales, which enhances the ability of feature expression and improves the efficiency of information processing. The experimental results show that the PSRN value of the algorithm is increased to 35.56dB, the SSIM value is increased to 0.978, the quality of the synthesized intermediate frame is better, and the streaming video playback is smoother.

TA15 titanium alloy is a metal material widely applied in aerospace and aviation. Superhydrophobic surfaces prepared with TA15 titanium alloy can effectively suppress icing and frosting in aircraft at high altitudes, however, the microstructure of superhydrophobic surfaces is prone to lose of efficacy in wear. To improve wear resistance of superhydrophobic TA15 surfaces, an insulation mesh with predetermined pattern was attached to the base metal by a dry-firm photoresist to set off an electrochemical reaction on the metal surface without an insulation coating, forming a superhydrophobic surface in topology. In electrolyte of 0.2 mol/L NaBr solution, TA15 was electrochemically machined using a square mask with an electrical machining parameter of 0.8 A/cm2 for 12 minutes as its surface was superhydrophobic. The worn surface upwards of 1,200 cm maintained superhydrophobic at contact angle of 150.1o and rolling angle of 21.3o. The EDS maps and SEM images of electrochemically machined TA15 titanium alloy surface showed that the use of a dry-film photoresist mask could effectively improve the wear resistance of the superhydrophobic surface, while the structure generated from mask electrolysis did not damage the hydrophobicity of the original surface. The test showed that the superhydrophobic surface had good self-cleaning and anti-adhesion properties.

Because the traditional wire crimping technology relies on manual adjustment and judgment based on experience, the crimped wire has a wide range of side distance changes, and the uniformity of the finished wire is poor. Based on the traditional wire crimping pliers, an automatic digital wire crimping control system is developed with Omron CP1H programmable logic controller(PLC) as the controller, Fuzzy PID as the core control algorithm, industrial flat panel as the human-machine interface and upper computer. The operator only needs to input a small number of preset pieces before the start to automatically start accurate crimping and measurement. The results shows that compared with the traditional wire crimping, it can realize the measuring the total length of wire movement,manually setting the moving distance of each step of crimping pliers, automatically measuring the diagonal length of hexagon wires,displaying the pressure of each crimping on the interface, automatically adjusting the crimping pressure of crimping pliers and the moving distance of wires.

In order to explore the effect of pressure surface winglets on gas flow in small-sized and high-load axial compressor rotors, this paper used a proven numerical simulation method to discuss the performance of pressure surface winglets on a high-loaded compressor rotor. The result showed that the pressure surface winglet increases the stable working margin of the prototype compressor rotor by 6.19%, which can significantly reduce the strength of the tip leakage flow and the occurrence of secondary leakage, but at the same time reduce the workability of the tip area, which reduced the momentum ratio of tip leakage flow and trailing edge separation flow, which increases the strength of trailing edge separation flow and increases the total compressor loss. In application scenarios that requires lower stabilizing working mass flow rate, the pressure surface winglet is a passive control method that can be used. Key words： Pressure Side Winglets, Stability Margin, High-loaded Compressor Rotor.

Aiming at the problem of aircraft engine health maintenance, this paper adopts the method of reinforcement learning. According to the status monitoring information, event information and maintenance information, the decay process model of aircraft engine is established. The state of aircraft engine is represented by the Markov decision model in discrete state, and the recovery model of aircraft engine under the condition of imperfect maintenance is established. Finally, based on a research example, the q-learning method is used to solve the decay model to obtain the average return of aircraft engines and the state-based predictive maintenance strategy, which has certain practical guiding significance.

In order to study the distribution law of stiffeners in the fan casing of the aeroengine, the maximum casing stiffness can be obtained under the condition that the casing inclusion requirements are satisfied. At the same time, in order to study the weight reduction of the casing without reducing the stiffness of the casing. Take the fan casing of an aeroengine as an example, the position and shape of the stiffeners at the containment of the casing are optimized by using the shape optimization method based on OptiStrcut, and the optimal structure form of the stiffeners is obtained. The optimal distribution area of the material is obtained by using the method of topology optimization. Based on the above optimization results, the fan casing is designed for weight reduction. By removing materials along the longitudinal grooves on the casing and removing invalid and redundant materials, the weight of the whole casing can be reduced by about 12% of the casing weight while ensuring the stiffness of the casing. This provides a new idea for the optimization and weight reduction design of the casing.

China's promotion of the "dual circulation" of the economy is to complete the mutual promotion and integration of domestic and international economic cycles on the basis of adhering to the domestic cycle. In this context, the importance of corporate innovation capabilities has become more prominent. The article selects the "One Belt One Road" initiative as the policy measurement event for trade liberalization, uses the 2009-2019 A-share listed companies as the research sample, and uses the Differences-in-Differences method to examine the actual impact of the "One Belt One Road" initiative on the innovation of Chinese enterprises, and at the same time in-depth analysis the role of financing constraints in the above influence reveals the internal mechanism of enterprise innovation. The results show that the country's adherence to high-quality opening to the outside world and the significant improvement of enterprises' innovation capabilities through trade liberalization will help promote the formation of a new domestic and international dual-cycle development pattern, and this effect will be achieved by alleviating corporate financing constraints. Further analysis of the heterogeneity of enterprises found that the innovation capabilities of private enterprises, large-scale enterprises, low-tech enterprises, enterprises with separation of two posts, and enterprises in the eastern region have improved significantly. This research conclusion is helpful to understand the positive benefits of the “One Belt One Road” initiative, and provides enlightenment for the government to further improve policies and Chinese enterprises to strengthen the use of policies under the new pattern of “dual circulation” economic development.

讨论具有工件可拒绝的单机公共窗口指派单机排序问题，其中公共窗口指派指的是公共窗口的开始时间和窗口长度都是决策变量。如果工件在窗口期内完工，将不会产生任何额外的费用，否则工件将会提前或者延误费用。若工件被拒绝，则会产生拒绝费用。目标是确定可加工的工件集、可加工工件集中的工件排列顺序和公共窗口的位置使得排序费用（包括提前、延误与公共窗口指派的费用）和拒绝费用的加权和最小，其中权重只与工件被排在序列中的位置有关，即位置权重。证明此问题是多项式时间可解的，并给出了具体求解算法。

Structural graph clustering is an important problem in the domain of graph data management. Given a large graph G, structural graph clustering is to assign vertices to clusters where vertices in the same cluster are densely connected to each other and vertices in different clusters are loosely connected to each other. Due to its importance, many algorithms have been proposed to study this problem. However, all clustering algorithms based on vertex features need to calculate the similarity between each vertex, which wastes a lot of computing resources. In this article, an algorithm named RP (abbreviation of graph clustering algorithm based on representative parameters) is proposed, which eliminates the need to calculate the similarity between all vertices. First, use historical clustering information to create a set of representative parameters. Then, when a clustering task is submitted, use a greedy algorithm to calculate connected components. Finally, graph clustering can be completed quickly. Extensive performance experiments using large graphs and composite graphs prove that the algorithm has high computational efficiency.

Based on Reddy's third-order shear deformation theory, a new buckling analysis model of composite stiffened cylindrical shell is established in this paper. The in-plane displacement in this model is a third-order function, which can naturally meet the free surface condition of the shell without introducing additional shear correction coefficient. Based on the energy principle, the buckling governing equations of stiffened composite cylindrical shells subjected to axial and lateral compression are derived and solved for typical examples. In the part of example analysis, the accuracy of this solution is proved by comparing with the literature solution, the influence of structural layout on bearing capacity is discussed, the similarities and differences between this solution and the classical first-order theoretical solution under different design parameters are studied, and it is proved that there are significant differences between this solution and the first-order theoretical solution when the ratio of diameter to thickness is high. Considering that the displacement field order of this model is higher than that of the first-order theory, it is suggested that this model should be used to calculate the composite stiffened shell with higher diameter-thickness ratio to improve the calculation accuracy

In order to study the interaction between the thermodynamic cycle parameters and flow path parameters of theaeroengine,a iterative model between the thermodynamic cycle parameters and flow path parameters of the aeroengine was established based on the steady-state performance calculation model of aeroengine and the flow path parameters modeling method of the turbofan engine. Based on calculation model, the effects of the thermodynamic cycle parameters such as total supercharging ratio, turbine front temperature and bypass ratio on the flow path parameters and turbine AN2 values of the military turbofan engine with small bypass ratio were studied. The calculation results show that, there was a strong influence of the total pressure ratio, turbine front temperature, bypass ratio on flow path size of the engine and the turbine AN2 value. Under the condition that the size and series remain unchanged ,the increase of the total supercharging ratio will lead to the increase of the aerodynamic load and mechanical stress of the core components ,the increase of the turbine front temperature is conductive to the decrease of the AN2 value of the turbine,and the increase of bypass ratio will lead to the decrease of the size of the core components and the in-crease of the size of the low-pressure turbine. The calculation model can realize the analysis of aero-engine thermodynamic cycle parameters,taking into account the flow path size of components,turbomachinery load and turbine AN2 value ect,and has a good engineering application value in the conceptual design stage of aero-engine overall performance.

Abstract: In order to control the flight attitude of the fixed-wing UAV stably, this paper uses MATLAB to control and simulate the roll channel of the UAV. Due to the various shortcomings of classical PID control, this paper combines backstepping theory, sliding mode control theory, robust control theory and adaptive control theory in modern control theory. A series of controllers are designed to meet the needs of fixed-wing UAV stability control. Although the backstepping sliding mode controller designed based on the backstepping theory has a good control effect, it brings the problem of control input chattering. In order to solve the chattering problem, a robust backstepping sliding mode controller and an adaptive backstepping sliding mode controller are further designed. The simulation results show that the adaptive backstepping sliding mode controller has the best control effect.

In the field of polymer composites, the type of filler, the amount of filler and the degree of dispersion affect the mechanical properties of composites. As a thermoplastic resin, polycarbonate has good mechanical properties, excellent tensile properties, impact resistance, heat resistance and wear resistance. In order to explore the effects of filler types and amount on mechanical properties of composites, polycarbonate was selected as the matrix in this experiment, and three nano-fillers including graphene, carbon nanotubes and carbon nanofibers were selected to prepare polymer-reinforced polycarbonate composites by twin screw extrusion. It was made into test parts by injection molding and analyzed and characterized by electronic universal testing machine and electron microscope. The results show that all the three fillers can enhance the mechanical properties of polycarbonate, and the carbon nanofiber has the best enhancement effect, and the tensile properties of the composite are the best when the filling amount is 2.5wt%. The tensile strength reached 90.12±2.18MPa, 25.17±3.03% higher than the original polycarbonate material. With the increase of nano-filler content, the young's modulus of the material continues to increase but the tensile limit begins to decrease.

The two-person interactive behavior recognition based on the combination of RGB video and joint point data has broad application prospects. Aiming at the problem that two heterogeneous features of RGB video and joint point data cannot be well integrated, this paper proposed a two-person interactive behavior recognition method that used joint point data to focus on RGB video. Joint point data analysis was used to obtain the main movement parts and attention was paid to the RGB motion features. By in this way, the global appearance information provided by the RGB video and the local motion information obtained from joint point data analysis were fused, and the features of them were effectively represented. The problem in the previous two-person interaction behavior characteristics that all information was treated in a unified manner but the main motion information was ignored could be solved, and the rationality of heterogeneous feature fusion was improved. Convolutional neural network was used to extract deep features and perform two-person interactive behavior recognition. The method is tested under the international standard NTU RGB+D database. The results show that the algorithm improves the recognition accuracy of similar actions, which proves the feasibility of the algorithm.

The dual-throat vectoring nozzle has good vectoring performance. In order to improve the performance of the dual-throat vectoring nozzle, a concept of dual-injection dual-throat vectoring nozzle is proposed. An injection was added to the upper wall to provide the radial velocity while reducing thrust loss, which can increase the thrust vector angle and obtain a larger thrust. The design parameters of the secondary injection of the nozzle were numerically simulated. The results show that, the flow rate and incident angle of the secondary injection will directly affect the internal flow pattern of the nozzle, thereby affecting the thrust vector angle and thrust coefficient. The best thrust vector performance obtained by this method in this paper: The thrust vector angle is 16.5°, and the corresponding thrust ratio is 96.7%.

CThe cultural industry is closely related to the tourism industry. On the one hand, cultural industry promotes the rebirth and innovation of tourism industry and promotes its product connotation. On the other hand, the tourism industry provides a large number of consumer groups for the development of the cultural industry and opens up a huge market space. Based on the data of Liaoning Provincial Bureau of Statistics and Shenyang Municipal Bureau of Statistics, this paper uses the coupling evaluation model to evaluate the coupling situation and coordination type of Shenyang's cultural industry and tourism industry, and discusses the reasons that hinder the coupling development and overall development level of Shenyang's tourism industry and cultural industry. And according to the cultural industry and tourism industry in Shenyang, the paper puts forward some suggestions and measures to promote the high-level coupling development of cultural industry and tourism industry.

Aiming at the problem of insufficient performance of single PID controller in variable cycle aeroengine when variable working conditions, two reinforcement learning thrust control methods are respectively proposed, based on deep Q network and deep deterministic policy gradient algorithms. The research gives two kinds of reinforcement learning controller designs, including engine state input parameter selection, actuator action output parameter selection, reward function setting based on control performance, state-action network convergence condition setting, and experience buffer mechanism setting based on engine characteristics. Simulation results show that the two kinds of variable cycle aeroengine reinforcement learning thrust controllers can obtain good performance indicators under a variety of external commands. The stability and robustness of the intelligent control design are verified, and a stable, fast and accurate control of variable cycle aeroengine thrust during different working points is achieved.