The unmanned aerial vehicle with folding wing can serve in multiple mission profiles，the flight environment is complex and susceptible to low-altitude gusts. A new parameterized aeroelastic model with non-intrusive characteristics was developed to quickly calculate discrete gust response cha-racteristics of Z-shaped folding wing at different folding angles. Firstly，structural and aerodynamic enhancement design was made to the existing Z-shaped folding wing. Secondly，based on the Nastran input file for the flutter calculation of the folded wing in the extended state，the parameterized aeroelastic model under any folding angle was reconstructed. Finally，the effects of parameters such as folding angle，flight speed，gust velocity amplitude and hinge damping between different wing segments on the gust response characteristics of the folding wing were examined. The results show that increasing folding angles can effectively alleviate the gust responses， the increase in flight speed and gust velocity amplitude significantly contributes to the responses of wing tip acceleration and wing root bending moment， the presence of hinge-damping parameters can alleviate the acceleration response of the wing tip，but it also increases the bending moment response of the wing root.

In order to investigate the influence of root slot on the static cascade performance of transo-nic compressor， the high-load diffusion cascade model NACA65-K48 was used as the basis and the influences of root slot with different width and height on the aerodynamic performance and flow structure of the compressor cascade were studied through numerical simulation. The results show that the root slot can effectively weaken the vortex strength on the suction surface， suppress the corner separation along both the pitch and blade height directions， and reduce the flow loss of the cascade. The inhibitory effect firstly increases and then decreases with the increase of the slot width， and it shows a similar trend with the rise in the height. When the width is 12% chord and the height is 5% blade height， the root slot structure of the blade has the most significant effect on improving the flow within the slot， reducing the cascade flow loss by 11.19%.

A specialized verification test for starting characteristics was carried out to investigate the influence of intake pressure loss on starting characteristics of core engine.A pressure stabilizing system was installed in front of the inlet to simulate the generation of intake pressure loss and the starting cha-racteristics of the core engine were recorded.For comparison，the starting characteristics of the core engine were also recorded without installing the pressure stabilizing system.After the experiment，the influence of intake pressure loss on starting characteristics was obtained，and the reasons for intake pre-ssure loss caused by the pressure stabilizing system，along with the calculation method for intake pre-ssure loss were studied.The results show that intake pressure loss reduces the speed rising ratio during the starting process，hurting the starting.The reasons for flow loss in different components of the pre-ssure stabilizing system are different，and various optimization methods for the different components can be used to reduce the pressure loss，such as improving the roughness of the rectifier network and the honeycomb，as well as selecting the diffusion angle of the diffuser section reasonably.

There are a large number of thin-walled components in aerospace products，which deform before the assembly process due to manufacturing errors and components stress release.If all the dimensions or tolerances of the assembled components can still be interchanged within the limit size，it often requires too high requirements for the components，the production cycle is also too long，and the manufacturing cost of the components was also too high.In response to the uncertainty of product components deformation， the modeling method of measured data was conducted，which effectively integrated technologies such as components geometry，material elasticity，physical data，virtual assembly and simulation analysis.Combined with components data collection before assembly，dynamic simulation analysis was conducted on the assembly process of component deformation bodies.So that out of tolerance deformation components can still be assembled normally.Furthermore，by combining big data and artificial intelligence technology，an equivalent model and application software for the assemblability of on-site deformable bodies are provided，realizing the rapid determination of whether deformed components can be assembled based on measured data of components on the production site，ensuring the continuous production of the production line.

In order to meet the jointing requirements of large-scale components，TA15 wire was used as raw material to join TA15 base metal prepared by laser deposition manufacturing （LDM） and hot isostatic pressing （HIP） with tungsten inert gas （TIG）.The results show that there are differences in the microstructure of different regions of the jointing specimens.The microstructure of the HIP base metal is equiaxed grains，and the internal structure of the grain is bimodal structure.The heat affected zone A （HAZ-A） of HIP base metal presents equiaxed grains.The TIG jointing zone presents β columnar grains，and the intracrystalline is fine basketweave structure.The LDM zone presents β columnar grains，and the basketweave structure inside the grains is coarser than that in the TIG jointing zone.The grain shape of heat affected zone B （HAZ-B） of LDM base metal is still columnar grains.The tensile strength and elongation of the jointed specimens at room temperature are 1 046.3±13.7 MPa and 6.2%± 0.6 % respectively.Compared with the base metal，the tensile strength of the jointed specimens increases and the plasticity decreases.The fracture position of the jointing specimens is HAZ-B，and the fracture mechanism belongs to semi-cleavage and semi-ductile fracture.

As one of key tasks in the field of remote sensing image processing，object detection has always been a research hotspot.Although significant progresses have been made in this field，the deep learning methods still face significant challenges in dealing with scale changes and complex backgrounds in remote sensing images，which limits the further improvement of detection accuracy to some extent.To address this issue，an innovative object detection method for remote sensing images was proposed，which integrated a saliency guided image adaptive fusion module and improved Faster RCNN to enhance the accuracy of object detection.Firstly，in the image preprocessing stage，a saliency guided image adaptive fusion module was proposed，which effectively integrated the semantic information of the image and shallow fine-grained details，allowing the model to prioritize the object region while minimizing background interference.Secondly，after introducing MobileNetV3 as the feature extractor of Faster RCNN，an attention enhanced feature pyramid network was proposed，which combined attention with upsampling to further enhance target features and output high-quality feature maps，effectively improving the extraction effect of multi-dimensional features and providing more accurate and rich feature information for subsequent object detection tasks.Furthermore，a multi-scale region proposal network was designed，which can more accurately capture the features of objects of different sizes and shapes，thereby enhancing the expression ability of features and effectively improving the detection accuracy of targets.Finally，experiments on the DIOR and ROSD datasets demonstrated that the proposed network model exhibits higher detection accuracy compared to other advanced methods，fully demonstrating its superiority and effectiveness.

In recent years，geo-social networks have attracted a large number of scholars to study them as an important means of abstracting real-world relationships. Among them，the radius bounded k-core search aggregates users who are spatially neighboring and closely related to each other to form subgraph sets，which is widely used in multiple aspects such as advertisement placement and social relationship analysis. However，the high degree of user overlap among subgraphs reduces the search efficiency，and the excessive number of aggregates causes difficulties in user selection. To solve these two problems，the largest radius bounded k-core （LRBK） search problem was proposed，which aimed to search for communities with the largest size that satisfied spatial and cohesion constraints.Combining the known best radius bounded k-core search algorithm RotC+，the basic algorithm was first proposed. Then，an optimization algorithm was proposed by combining effective pruning and optimization strategies. Finally，extensive experiments on five real-world datasets were conducted.The experimental results show that the efficiency of the optimized algorithm is improved by up to about 20 times compared to the basic algorithm.

A method of expanding and improving the hierarchical control structure model （HCSM） of system-theoretic process analysis （STPA） using functional attribute（FA）and directional interaction tag （DIT） was proposed. Based on this method， the hierarchical functional control structure and interaction model （HFCSIM） of the system and essential improvement to STPA was obtained. Through this modification， issues such as the lack of specific methods and forms follow， incomplete interaction information between components， excessive reliance on “Brainstorming” and the difficulty in ensuring model consistency could be solved， and the systematicness， completeness and correctness of the analysis results could be fundamentally ensured. Finally， the effectiveness of the modified method was validated by taking the aircraft wheel braking system as an example.

Based on imprinting theory and upper echelons theory，took 2015—2021 Shanghai and Shenzhen A-share military-civilian integration listed enterprises as research samples，empirical regression methods were used to verify the impact of the CEO’s IT background on the innovation performance of military-civilian integration enterprises as well as the intrinsic mechanism.The results show that CEO’s IT background has a significant positive effect on the innovation performance of military-civilian integration enterprises.CEO’s IT background has a significant positive effect on digital transformation.Digital transformation plays a partial mediating role in the promotion process of CEO’s IT background to the innovation performance of military-civilian integration enterprises.Government subsidy plays a positive moderating role.After the robustness and endogeneity tests，the above findings remain robust.Further heterogeneity analysis reveals that the promotion effect of CEO’s IT background on the innovation performance of military-civilian integration enterprises is more significant in civil-involved military enterprises and the eastern region.

Fuzzy support vector machine is a classification algorithm that combines support vector machine and fuzzy theory.The existing fuzzy support vector machine algorithms can overcome the impact of noise data to some extent，but they have cost sensitivity，leading to inaccurate estimation of the prior distribution of data.A new fuzzy support vector machine algorithm was proposed.When designing the fuzzy membership function of samples，this algorithm better captures the distribution information of data by using the outlier factor constructed by the similarity of sampling points and their neighborhood density.The optimized model is validated using UCI datasets，which proves its good performance.