| With the process of urbanization,there are more and more large buildings,and the structure is more complicated.Such structures are easily affected by various natural disasters,environmental conditions and operating loads.If disasters such as collapse occur,they will not only cause property losses,but also threatened people’s lives,causing social panic and causing economic development to be blocked.Therefore,during the early planning and design phase,mid-term construction phase,and later operation management phase of the entire project construction,reasonable monitoring measures should be adopted regularly or continuously tracking the health status of the structure in order to discover structural injuries and alarms in the early stages.The global navigation positioning system is widely used in the monitoring field with the advantages of high positioning accuracy and no need to communicate.Therefore,in order to better play the advantages of the global navigation positioning system and the characteristics of the Beidou constellation heterogeneous,a single-different observation value domain star daily filter algorithm is established to eliminate Beidou multi-path errors.In response to GNSS,it is not sensitive to high-frequency information,and the introduction of the GNSS/acceleration meter resistant multi-rate Karman filtering algorithm is designed to introduce the acceleration meter to improve the robustness of the model while solving the multi-rate sensor data fusion.It is difficult to determine the deformation monitoring and early warning threshold.It proposes a deformation monitoring and early warning model based on the LSTM neural network to identify mutation displacement and achieve effective early warning.The main work of this thesis is as follows:(1)The principle of EMD,EEMD and CEEMDAN signal decomposition algorithms is emphatically studied.The suppression effect of the three algorithms on modal aliasing and the residual situation of Gaussian white noise after signal reconstruction are compared.The boundary IMF component is determined according to the constant product of the energy density of IMF component and its average period.(2)According to the heterogeneity of Beidou constellation,a filtering algorithm of sidereal day in observation range based on single difference residual is proposed.Calculate the satellite orbit repetition cycle time advance through broadcast ephemeris and use it as the multipath error time advance.Based on the assumption of zero mean,the double difference residual is converted into a single difference residual.The CEEMDAN algorithm is used to extract the multipath error models of each satellite in the single difference residual sequence,and the multipath error model of the previous repetition period is corrected to the current repetition period.The experiment shows that after multipath error correction,the positioning accuracy in the E,N,and U directions has been improved by 19.6%,23.9%,and 15.1%,respectively.(3)A GNSS/accelerometer multi rate fusion algorithm is proposed to address the different sampling frequencies between GNSS and accelerometers,transforming multi rate sensor data fusion into a single rate sensor data fusion problem.Construct a multi rate Kalman filter model using accelerometer data as the control vector and GNSS displacement data as the observation vector.Construct an equivalent gain matrix based on predicted residuals to suppress gross errors and improve the robustness of the model.The shaking table experiment shows that the positioning accuracy of this method in the east and north directions has been improved by 4.1%and 17.8%,respectively.(4)In view of the difficulty in determining the early warning threshold of Deformation monitoring,an early warning model of Deformation monitoring based on LSTM neural network is proposed.Use sliding windows to calculate the standard deviation between the test set and the training set,and adaptively calculate the warning threshold.The experimental results show that the algorithm can accurately predict structural deformation and identify sudden displacement,achieving effective early warning. |
