To achieve quantitative detection of bolt loosening angles through single frame images, a method based on color segmentation and connected domain feature processing was designed. Firstly, a method for performing nonlinear stretching, normalization and optimal threshold segmentation on a component successively in the Lab color space was designed to segment and represent the red anti-loosening line image of the bolt loosening angle. Secondly, the morphological operations were performed on the image by using the open operation. Then, the orientation vector of the connected domain in the anti-loose line image was determined by computing the geometric moments. Finally, the bolt loosening angle was determined through the four-quadrant arctangent function. The results demonstrate that the precise measurement of the bolt loosening angle through a single frame image can be achieved by this detection algorithm, with a maximal relative error of 1.80%, its accuracy meets the needs of engineering practice and has strong engineering application value.

Since the fault vibration data collected in the real engineering may be accompanied by noise, traditional diagnostic models are difficult to identify fault categories. To address this problem, a rolling bearing fault diagnosis research method based on channel and spatial reconstruction and progressive convolutional neural networks (CSRP-CNN) was proposed.The model utilized channel and spatial reconstruction convolution（CSConv）to reduce the redundant information of channels and space in fault features, and reduced the complexity and computation to improve the performance; using the convolutional block attention module (CBAM), attention enhancement operation was carried out in the channel and spatial dimensions to make the model pay attention to the important fault feature information; and the progressive convolutional network structure was used in the shallow layer of the network, which would fuse the previous fault feature information with the current input to obtain the richer feature information. The performance of CSRP-CNN was evaluated by two different datasets of Case Western Reserve University (CWRU) and machinery fault simulator magnum (MFS-MG). After the noise and ablation tests, it is verified that CSRP-CNN has strong robustness and the effects of CSConv, CBAM and progressive convolutional neural network (PCNN) on the model noise immunity performance.

In order to gain insight into the stability of the tracking frame structure of shipborne large-aperture telescopes,the stability of typical ground-level telescope tracking frames was studied. According to the external load borne by the equipment in the case of ship, the external load was parameterized and entered into the finite element software. The pretreatment software and finite element software were used to analyze the structural deformation under static wind load. Then,the natural frequency of the structure was solved, and a simple response spectrum analysis calculation was proposed instead of the tedious random response analysis to analyze the stability of the equipment under dynamic wind load and wave excitation. According to the stress and deformation values obtained from the results, it was ensured that the shipborne telescope tracking frame theoretically meets the strength requirements and design accuracy requirements under shipborne conditions. Under the static wind load, the maximum stress value of the tracking frame structure is about 14.07 MPa, which was less than the yield strength of steel 355 MPa, the maximum deformation variable was about 0.02 mm, which was less than the design accuracy error coaxiality ϕ0.1 mm, and the natural frequency 1st-6th order mode value was 40.15, 49.65, 66.86, 82.93, 91.38,115.89 Hz. Under dynamic wind load, the peak value of structural stress was 3.92 MPa and the maximum deformation variable was 0.01 mm, and under the excitation of ocean waves, the peak of structural stress was 5.88 MPa and the maximum deformation variable was 0.02 mm, which was less than the yield strength and design accuracy error coaxiality of steel. The error between the modal value obtained by the modal test and the calculated modal value is within 10%. Combining theoretical simulation and practical tests, the tracker structure can work normally under shipborne conditions.

Under the medium and high speed of spherical hybrid sliding bearings, the dynamic pressure effect causes the fluid in the wedge space to whirl, and the vibration characteristics of the rotor may be affected by the oil whirl, so the rotation accuracy of the main shaft is reduced. The spherical hybrid sliding bearings with the orifice throttle was divided into cylindrical and conical whirl. The lubrication mathematical model and rotor dynamic model were established and solved simultaneously, the journal center trajectory and vibration amplitude were obtained. The influence of the centroid offset distance and initial deflection angle on the vibration characteristics of the rotor system were studied.The results show that,compared with the pure cylindrical whirl, the stability of the journal center trajectory decreases and the vibration amplitude increases after considering the conical whirl.With the increase of the centroid offset distance, the stability and vibration amplitude of the journal center trajectory decrease greatly.With the increase of the initial deflection angle, the stability and vibration amplitude of the journal center trajectory decrease only slightly.It can be concluded that changing the distance of the centroid offset has more influence on the stability of the journal center trajectory and the vibration characteristics of the rotor system than changing the initial deflection angle.

The creep behavior of 7050-T7451 aluminum alloy under different temperatures and stresses was studied by the uniaxial tensile creep test. Based on continuum damage mechanics, a constitutive model describing the creep behavior of aluminum alloy at high temperature was established. The model took into account the precipitation coarsening, dislocation multiplication/annihilation and microvoid formation during the creep process, and introduced the corresponding damage factor evolution formula to reflect these three damage processes. In addition, the model took into account the additional damage caused by stress increase to reveal the effect of stress on the creep damage. Based on the comparative analysis of the test results and the predicted results, it was verified that the established physical constitutive model can accurately describe the creep behavior of 7050 aluminum alloy under different temperatures and stresses.

In order to improve the load-bearing efficiency of the stiffened conical shell in large launch vehicle, the lightweight design of the stiffened conical shell was carried out via a data-driven multi-fidelity approximate modeling optimization method. Aiming at the problems such as low efficiency and insufficient accuracy of the single fidelity approximate modeling optimization method, a data-driven multi-fidelity approximate modeling optimization framework was built based on variable-fidelity expected improvement (VF-EI) point criterion，and accordingly the optimization design of stiffened conical shell structure was carried out. Based on the finite element models of stiffened conical shells with different mesh sizes, a Co-Kriging multi-fidelity approximate model for the collapse load of stiffened conical shells was established. In the optimization iteration, multi-fidelity sampling points were generated by using VF-EI point criterion, and the global and local approximation accuracy of Co-Kriging multi-fidelity approximation model was improved sequently. Moreover, the optimization efficiency and accuracy of the proposed method were demonstrated by comparing with radial basis function approximation model and Kriging model. Besides, 11. 5% weight reduction of the optimized stiffened conical shell structure is obtained compared with the initial design, which has certain engineering application value.

In view of the difficulty of obtaining durability test specifications for axle components, a method of constructing driving load of the front axle dynamic model was proposed to provide the input for system-level bench test and life verification of front axle components. The life of front axle was obtained based on the whole vehicle dynamics model and the measured road load, and the useful information and life prediction results in the above process were employed to guide the construction of driving load applied on front axle. Firstly, the frequency band of the measured six-component forces at the wheel center was adjusted. Then, with the goal of minimizing the difference between the damage/life of the front axle and the reference value under each working condition, the adjustment coefficient of the amplitude of the three-way forces at the wheel center was optimized by combining the response surface method and genetic algorithm. The optimized three-way forces at the wheel center and the other three-way torques constituted the driving signal of the front axle model. The results show that the constructed wheel center drive signal is used to simulate the dynamic load of the front axle model. The life of the front axle,the failure sequence of risk points and the damage contribution ratio of various road conditions to dangerous points are obtained and in good agreement with the reference results, which verified the effectiveness of the proposed method and provided reference for the drive signal construction and component life evaluation of system-level bench durability test.Finally, the damage distribution of the front axle can be consistent with the reference results by modifying the main damage load of the shaft tube.

The relative curvature of the spur gear mesh point is one of the key geometric parameters of the tooth profile,which has a significant impact on the stress distribution and mesh stiffness of the gear. Starting from the relevant research on constant relative curvature (CRC) gears, three types of non-involute spur gear tooth profiles were constructed based on the relative curvature control strategy considering the time-varying mesh characteristics of gears, and the mesh simulation of gear pairs was implemented. The influence of relative curvature control on the maximum contact stress, maximum bending stress,and mesh stiffness of gears was analyzed. The effectiveness of the control strategy was verified. This work provides some reference for the design of spur gears based on the relative curvature control.

At present, the research on supersonic civil aircraft wings mainly focuses on the low sonic boom design and supersonic drag reduction technologies. There are relatively few studies on the wing structural design. Therefore, a multi-level optimization method for the wing structural design in the preliminary design stage of supersonic civil aircrafts was proposed. It included the parametric modeling of the wing structural layout, the automatic generation of the finite element model for the structural size optimization, construction and training of a surrogate model for the deep neural network. And the optimization was solved based on the deep neural network. The analysis results show that the proposed optimization strategy could quickly design the wing structure of the supersonic civil aircraft. The deep neural network model has higher prediction accuracy than the traditional surrogate model. Thus, the proposed approach can improve the efficiency of the preliminary design for wing structure.

The mechanical response of external rotor direct drive generator under time-varying loads is analyzed theoretically, calculated by simulation and verified by test. Firstly, the source of magnetic pull of the rotor core of the external rotor generator and its variation with load were analyzed. The expression of magnetic pull of the rotor and the characteristics of time and space order were determined. The basic vibration model of the external rotor core was analyzed, and the basic vibration equation of the external rotor core was determined. Then the simulation model of the external rotor generator was established, and the spatiotemporal order characteristics of the magnetic pull and the typical daily variation rule with time were obtained. The magnetic pull density obtained from the electromagnetic field was used as the input load to guide the structure field. The magnetic-solid coupling simulation was carried out, and the deformation and stress distribution and noise response of the external rotor core were calculated and analyzed. Finally, a simulation example of a 13 kW external rotor direct-drive generator proved the correctness of the analysis and simulation. The results of the study determined the time-varying load on the outer rotor core and its mechanical response distribution. It is found that the time and position of the rotor core should be tested emphatically when the generator is running. Noise characteristics under time-varying loads are also analyzed. The analysis provides reference for the maintenance and design of the generator.