Structural Seismic Response Monitoring And Risk Assessment Algorithm Study

Recently,MEMS chip,wireless communication and cloud computing technology have developed rapidly.Structural health monitoring system based on wireless intelligent sensor network overcomes many shortcomings of the traditional system,such as wire arrangement difficulty,poor maintainability,bad timeliness,limited application scenarios.The new structural health monitoring system has great development prospects.This paper designed and established a structural seismic response monitoring and risk assessment cloud platform prototype for Tsinghua University civil engineering building.The platform is divided into five modules: data collection,data broadcasting,data warehouse,preprocessing and application.This paper studied several algorithms for structural seismic response monitoring and risk assessment.Following works have been done:In this paper,the LSTM neural networks were built to fit the time series,filled the missing values,identified and replaced the outlier values.The algorithm achieved satisfactory results.An automatic structural modal parameter identification method is proposed based on HHT.Structural modal parameter identification is one of the core algorithms of structural health monitoring system.In the process of long-term monitoring,the modal parameter identification algorithm cannot contain subjective parameter selection process.Firstly,K-L divergence and DNN are used to identify and eliminate the false components generated by EMD in the first step of HHT.Then,SSA and Butterworth filter are used to separate the aliasing modes of IMF.The method can be effectively applied to the monitoring data of Tsinghua University civil engineering building.A structural damage detection method based on ARMA time series model and kernel ridge regression is proposed.A vital capability of structural seismic response monitoring and risk assessment system is to detect and quantify the structural damage.The method proposed in this paper effectively uses the regularization and kernel function characteristics of kernel ridge regression.The model has strong fitting performance and generalization performance.A structural safety scoring system based on ARMA time series model and GRA,is established.In addition to structural damage identification and location,the platform needs to evaluate the structural safety more intuitive and understandable.The method proposed in this paper uses GRA of AR coefficients as index,and determines the weight through the separation degrees among indexes.Finally,the automatic modal parameter identification algorithm and safety scoring algorithm are verified by shaking table test monitoring data of a three-story reinforced concrete frame shear wall structure.