The electromagnetic performance of phased array antenna is greatly affected by the phased array antenna surface deformation. How to apply the strain measurement data of sparse fiber grating strain sensors to sense the shape of antenna array is the key to realize structural health monitoring and electromagnetic performance control. This paper proposes a virtual sensing method for structural deformation under complex experimental modes. In this method，the complex mode transformation is first used to process the complex mode data obtained from modal testing to obtain the corresponding real displacement modes，and then the full field expansion of finite real displacement modes is realized using mode expansion；Combining the extended real displacement modal and finite element modal data，two virtual sensing equations named CMT-SEREP（complex mode transformation-system equivalent reduction expansion process）and CMT-LC（complex mode transformation-local correspondence），which characterize the relationship between sparse measured strain information and the full field displacement of the structure，have been derived to achieve the real-time estimation of the deformation shape of the antenna structure from sparse measured strain information. Using the developed large phased array antenna array deformation experimental platform，experimental verification of different sensing methods was carried out under three deformation working conditions. Experimental results show that the proposed method can reconstruct the full field displacement of the antenna array structure using sparse strain measurement information，and the sensing accuracy of CMT-LC is higher than that of CMT-SEREP. Compared to the traditional modal method，the relative percentage error of deformation sensing using CMT-LC method has been reduced by at least 6.105%. This method is not only suitable for deformation sensing of non-proportional damping antenna structures，but also suitable for other complex engineering structures，and it has a great application potential.

The analysis of seismic response at seabed sites is a crucial initial step in marine engineering construction. In this study，a fluid-solid weak coupling model is employed to replicate the interaction between seawater and the seabed. Specifically，four representative borehole sections along the proposed tunnel at Qiongzhou strait are chosen to investigate the influence induced by seawater，soft sediments，and bedrock earthquake motion on the seismic responses of the seabed site. A generalized non-Masing constitutive model（DCZ model）is utilized to account for the dynamic nonlinearity of the seabed soft soil. The findings indicate that the suppression effect of seawater on seismic motion in the seabed is limited to depths shallower than 50 m. Furthermore，the suppression effect is more pronounced in the vertical direction compared to the horizontal direction. Additionally，there is a positive correlation between the suppression effect of seawater on seismic motion at the seabed surface and the frequency response phenomenon characterized by high frequency suppression and low frequency amplification in the seabed seismic response. This correlation is influenced by the depth of the seawater. The mean lines of the horizontal and vertical spectrum β obtained by numerical calculation are higher than the design spectrum in the land code in several period ranges，and the possibility of adverse effects induced by seawater and seabed soft sedimentation on the seismic resistance of marine structures should be considered.

For cross domain diagnosis of the label spaces of source domain and target domain are partially overlapped，that is to say，both the target domain and the source domain contain the classes that the other does not have，a cross domain adaptive fusion diagnosis method based on weighted adversarial learning is proposed. As entropy can be used to reflect the characteristics of the shared known classes and unknown classes，two convolutional neural networks with the same structure are introduced to carry out entropy-based weighted adversarial training，which is aim to enhance the ability to identify the shared known classes by extracting the domain-invariant features，as well as the binary cross schemes of the source domain and target domain sample outputs are used to isolate the unknown classes. In addition，the fully connected layer hidden features of these two convolutional neural networks are taken as the input of two label transfer models，and the probability outputs of these three diagnostic models are fused by voting rule. The failure test bench data of mechanical transmission components under variable working conditions and the damage data of selfpriming centrifugal pump are used for analysis and verification，the experimental results show that the proposed cross domain adaptive fusion diagnosis method can distinguish the shared known classes and unknown classes in the target domain more accurately.

Time-varying modal identification is important to obtain the dynamic characteristics of time-varying engineering structures and realize real-time vibration control and online health monitoring of structural systems. Aiming at the problems of measuring the excitation of engineering structures and low efficiency of time-varying modal identification，an output-only recursive identification method based on dynamic mode decomposition（DMD）is proposed in this paper. To extend the theory of the existing DMD method，this paper draws on the projection approximation subspace tracking algorithm and the sliding-window idea，and proposes a recursive format of the DMD method，which can update the system matrix and the proper orthogonal decomposition basis recursively，and can be further used for the output-only recursive modal identification of time-varying systems. A numerical example of a three-degree-of-freedom structural system with time-varying mass and an experimental setup of a liquid-filled cylindrical structural system with variable mass are respectively designed and built to validate the proposed method numerically and experimentally. The results demonstrate that the proposed recursive DMD method is able to accurately identify the modal parameters of the time-varying structures by only using the measured vibration response data，and has good output-only recursive identification capability.

Currently，gear fault detection based on vibration，acoustic emission，and other signals requires the installation of additional sensors. This approach faces limitations in terms of sensor placement，high sensor cost，and difficulties in analyzing signals due to modulation effects in gear systems with variable transmission paths，such as planetary gearboxes. A method for diagnosing local gear faults using the built-in encoder of the servo motor is proposed in this paper. Using the output signal of the built-in encoder to extract the feature related to local gear faults. The encoder signal acquisition wiring is drawn from the built-in encoder of the servo motor，and a high-speed counter is used to record the time interval between the rising edges of the angular position pulses of the rotary encoder. The instantaneous angular speed（IAS）signal is calculated，and the IAS signal is analyzed and feature extracted in the angular and order domains to achieve local gear fault detection. Taking the planetary gearbox gear localized fault detection as an example，the proposed method is validated through experiments. Results show that using the built-in encoder of the servo motor can effectively achieve the detection of gear partial faults under low and variable speed conditions. This provides a new approach to fault detection of transmission units such as gearboxes in applications driven by servo motors.

The fuzzy domain of traditional fuzzy control is fixed，and the control efficiency will decrease when the dynamic characteristics of the controlled structure or external excitation changes. On the basis of traditional fuzzy control algorithms，a variable universe fuzzy control is designed. The variable universe fuzzy control takes the error and error rate of the controlled structure as input，and the scaling factor as output，achieving adaptive adjustment of the fuzzy domain of the main fuzzy controller. A two-story steel frame structure with a magnetorheological damper as the control device was constructed，and the variable universe fuzzy control system with the displacement and velocity of the first floor as inputs was developed in the dSPACE real-time simulation system. Shaking table tests under different intensities of seismic motion and different additional mass conditions were conducted. The results show that variable universe fuzzy control can adaptively adjust the fuzzy domain，effectively reducing structural displacement，velocity，and acceleration response. When the added mass of the controlled structure and the peak ground acceleration change，the control effect of variable universe fuzzy control is better than that of fuzzy control and OFF passive control.

Control algorithm is a key factor to improve the performance for reducing helicopter vibration. In this paper，according to the multi-frequency characteristics of helicopter vibration，both the response separator and controller are constructed utilizing the adaptive notch filter to establish the adaptive dual-notch control of helicopter structural response. The response separator separates each frequency component from the error response to update the control input of each harmonic independently in time domain. Utilizing a dynamic similarity model of helicopter airframe，simulations and experimental studies of the proposed adaptive dual-notch control algorithm are carried out. The results show that the adaptive dual-notch control algorithm has good control performance and faster convergence rate under the multi-frequency excitations，and it can enhance the robustness of active vibration control system by increasing the critical convergence step size.

An axisymmetric finite element-boundary element coupled computational method is proposed for the dynamics of cylindrical shells with endplates in water. Due to the periodic characteristics of the structure，the movement of the cylindrical shell and end plate can be determined by examining their meridional movement，and the meridians are discretized into several elements for further analysis. By using the energy method，the elemental matrices are obtained and assembled into the global matrices. Then the discrete schemes for structural motion with the finite element is established. Considering the axisymmetric characteristic of the fluid，an axisymmetric boundary element on the meridians is established and the calculation format for fluid-elastic forces on the finite elements is given. The fluid forces are then converted into an equivalent nodal force at the nodes and the effects of the fluid are evaluated by the added mass，damping，and stiffness matrices. By combining the structural motion and the schemes of fluid-elastic forces，an axisymmetric finite element-boundary element coupling method is developed to solve this fluid-structure coupling problem. The calculated results presented in this paper demonstrate good agreement with existing theoretical solutions and commercial analysis software，validating the accuracy of present method. Moreover，this method exhibits higher computational efficiency compared to commercial analysis software. It allows for direct determination of the mass and stiffness matrices of cylindrical shells in fluid，facilitating fast calculation of forced vibration. Based on this method，the wet frequencies and stability characteristics of complex cylindrical shells with end plates in axial flow are analyzed. The results show that the end plates will change the flow velocities on the cylindrical shell and make it more easily to instability.

Based on the three-dimensional finite element slope dynamic analysis model which is validated through the test data of 1：7 indoor physical model for live tree stump，the attenuation characteristics of additional dynamic stress on live tree stump slopes，the stress response characteristics of taproot and lateral root of live tree stumps，and the influence of live tree stumps on slope stability are studied. The dynamic stability mechanism of live tree stump slope is explored. The research conclusion is as follows. The peak value of vertical dynamic soil pressure calculated by the three-dimensional finite element dynamic analysis model is close to the measured results of the indoor physical model. The method and calculation results of the three-dimensional finite element dynamic analysis model of the live tree stump slope established through Midas GTS NX are reliable. The peak value of additional dynamic stress on the slope is affected by the superposition effect of train axle load. The higher the train movement speed，the greater the peak value of vertical dynamic stress. The additional dynamic stress in the ballast layer is the largest，it rapidly decays and diffuses downwards showing a semicircular arc shape. Only small dynamic stresses are transmitted to the two rows of live trees on the upper slope. The degree to which the shear resistance of the taproot of live tree stumps varies at different positions，and the taproot at the foot of the slope is subjected to greater shear stress compared to the taproot at the shoulder of the slope. The taproot is similar to the anti-slide pile to exert its shear resistance ability. The side roots of live tree stumps growing inside the slope are similar to anchor rods，while the side roots growing outside the slope are similar to supports，and they form an anchor-support effect that synergizes with the main root to play a sliding resistance role. The existence of live tree stumps leads to a redistribution of stress in soil，and the shear stress concentration near the live tree stumps. The live tree stumps hinder the transmission of shear stress in soil，inhibit the connectivity of plastic zones，and improve slope stability. The degree of influence on the horizontal dynamic displacement of the slope is related to its distance to the location of the dynamic load，and the closer the distance，the greater the impact. The live tree stumps can reduce the horizontal dynamic displacement at various points on the slope，but the reduction at the foot of the slope is the greatest. The live tree stumps can significantly improve the stability of slopes under train power，and their slope safety factor can be increased by 15%~20%. The potential sliding surface of the live tree stump slope presents a circular arc shape，and the live tree stump moves the plastic zone towards the deep soil layer，thereby improving the stability of the slope. The research results can provide certain guidance for the application of live tree stumps in roadbed slopes.

In the processing of near-fault original seismic acceleration records，how to retain the real ground permanent displacement information is a key problem to be solved in the baseline correction of seismic acceleration records. Based on the analysis and discussion of the validity and applicable scope of existing near-fault seismic acceleration baseline correction methods，this study introduces a smooth slope displacement function model and establishes a new baseline correction method that can reasonably characterize the permanent displacement of near-fault ground motions，and the new baseline correction method is verified by analyzing the baseline correction results of typical near-fault acceleration records. The results show that the new method established in this study improves the fitting accuracy between the displacement function model and displacement time history，reduces the influence of the selection of subjective parameters on the baseline correction results，and the corrected ground permanent displacement is in good agreement with the GPS co-seismic displacement. The ground motion baseline correction method established in this paper can not only automatically deal with the baseline drift of near-fault ground motion，but also reasonably characterize the permanent ground displacement caused by fling-step effect.