Study On Structural Seismic Safety Monitoring And Smart Diagnosis For Isolated Gymnasium Structure

As the elementary residence for the public,building security is the fundamentals of a smoothly operated city.The rapid progress of urbanization has triggered the increase of multi-story public buildings,some of which have been brought into service for decades.Therefore,in order to ensure the safety and reliability of the in-service building in high-intensity seismic regions,conducting structural strengthening with seismic isolation technology and structural health monitoring technology is a necessity.In this study,some baseisolated bearings have been designed for strengthening a gymnasium in high earthquake intensity area.Based on theoretical analysis,experimental verification,local test and numerical simulation,a series of study was conducted to study its seismic safety monitoring and smart diagnosis methods.The main work is as follows:(1)Considering the updating tendency on smart monitoring and vibration-based structural damage identification,a new deep learning(DL)method based on convolutional auto-encoder was proposed,to make the best of the auto-encoder(AE)and 1-dimensional convolutional neural network(1D-CNN).To assess the effectiveness of the method,a series of real-time monitored data from the ASCE Benchmark model was used to train and optimize the parameters of the DL network.After training and optimizing,we used five groups of vibration response data with different damage states to test network identification ability.The result revealed that diverse groups of data in high dimensional Euclidean space have different inherent characteristics,which can be extracted with a convolutional auto-encoder.Moreover,these low-dimensional data can also be reconstructed with a transpose convolutional decoder algorithm.The reconstruction ability of the network decreases with the damage state develops.To quantify the reconstruction ability of the DL network,two functions,mean square error and Pearson correlation coefficient,were used to compare input raw data and output reconstructed data.The quantifying results have been proved to meet the five corresponding damaged states.(2)Based on the unsupervised DL theory,a new data-driven isolation story diagnosis method was proposed,which can diametrically identify and locate the damage and earthquake incidents.To test its functionality,a numerical finite element model for the gymnasium structure was established.Damage model 1 considered the seismic performance degradation of the isolation story while Damage model 2 considered the damage brought by local failure of the seismic isolation devices.Response data acquired from model 1 mixed with 10～40%noise can be distinguished by DL network and judge damage and undamaged state.Local damage in model 2 can also be identified by DL network and the safety score in damage location dropped from over 90 to less than 40.So the method can locate and evaluate local damage.On top of the numerical analysis,the accidental earthquake can also be recorded and judged by the smart monitoring algorithm.(3)To aggregate different type of dynamic monitoring data,a safety probability&AHP method was used to comprehensively evaluate the gymnasium structured with various monitored data,including acceleration,displacement,strain stress and tilt angle.The method caters to the requirements on seismicisolation evaluation and upper-structure evaluation.After updating the FE model to elastoplastic model through OpenSees program,we conducted dynamic historical analysis with 15 natural seismic waves and recorded the response data as multi-level early warning limitation.Finally,the multi-level smart diagnosis was realized with the aggregated method.(4)Based on the smart diagnosis methods and digital-twin city operation philosophy,we developed a platform for gymnasium seismic safety monitoring and smart diagnosis.The network topology frame was designed with various sensors and the platform was integrated with B/S network frame.The interactive administration interface was programmed with Python-GUI and the data visualization interface was programmed with Javascript and HTML5.The function of the platform include:1)raw database and IoT administrative system;2)DL safety evaluation system;3)structural health monitoring system;and 4)frontend visualization system with GIS-BIM-SHM technology.