A real-time early warning method for secondary collapse risk of RC frame semi-ruin structure under aftershock

Abstract

The present application relates to the field of earthquake site rescue warning, and relates to a real-time early warning method for secondary collapse risk of RC frame semi-rubble structure under aftershock. The core content of the method includes four modules. Module 1: using the method of seismic simulation shaking table, the whole process of RC frame structure from perfect to semi-rubble to secondary collapse is realized by simulating main-aftershock action, and time history data of dynamic response of secondary collapse structure is obtained. Module 2: high-dimensional continuous horizontal moving time window method is used to process the time history data of dynamic response of structure and obtain the dynamic response characteristics of RC frame semi-rubble secondary collapse structure. Module 3: K-Means machine learning algorithm is used to cluster the secondary collapse risk type of RC frame semi-rubble structure. Module 4: Informer machine learning algorithm is used to realize real-time early warning of secondary collapse risk of RC frame semi-rubble structure. The present application can scientifically and accurately distinguish and real-time predict the secondary collapse risk of RC frame semi-rubble structure under aftershock.

Description

Technical Field

[0001] This invention relates to the field of earthquake on-site rescue early warning, and specifically to a method for real-time early warning of the risk of secondary collapse under aftershocks in RC frame semi-ruin structures. Background Technology

[0002] Actual earthquake damage shows that reinforced concrete (RC) frame structures are prone to partial collapse with a survival space at the bottom under rare earthquakes, threatening not only the lives of trapped personnel but also the lives of rescue workers. For example, more than 60 rescue workers perished at the scene of the 2008 Wenchuan 8.0 magnitude earthquake in my country; and more than 100 rescue workers and volunteers died in the aftershocks of the 1985 Mexico City 8.1 magnitude earthquake. The main reason for this is the lack of scientific and effective methods for identifying, predicting, and providing early warning of the risk of secondary collapse under aftershocks in partially collapsed structures with a survival space during earthquake rescue operations. Providing accurate prediction signals and sufficient escape time during rescue operations is of great significance for ensuring the safety of rescue workers and improving the efficiency of rescuing trapped personnel.

[0003] Currently, there are two types of methods for achieving safety early warning of building ruins: indirect and direct methods. Indirect methods utilize earthquake early warning (EEW) to provide on-site earthquake warnings. When an aftershock occurs and destructive aftershock waves reach the potentially damaged earthquake rescue site, the EEW system sends an earthquake warning to the rescue area. Once rescue personnel receive the warning signal, they immediately evacuate. However, earthquake early warning methods still have limitations, such as blind spots, underestimation of the magnitude of large earthquakes, and a discrepancy between accuracy and speed. Missed strong aftershocks can seriously threaten the lives of rescue personnel, and false aftershocks, especially misreporting smaller aftershocks as larger ones, can lead to repeated evacuations, reducing the crucial rescue time. Direct methods involve monitoring and analyzing changes in the structural dynamic response of building ruins under aftershocks or external disturbances to predict their hazard. For example, at emergency rescue sites, levels or total stations are used to observe changes in parameters such as displacement, velocity, acceleration, and tilt angle of the building ruins to provide early warning. However, measurement and early warning methods such as levels or total stations are still unable to enable emergency identification and accurate prediction of secondary collapse of building ruins at earthquake rescue sites, making it difficult to provide rescuers with more time to escape.

[0004] To provide rescuers with sufficient time to escape, accurate prediction of future risks from building ruins is crucial. Current research by scholars both domestically and internationally primarily focuses on predicting long-term risks that develop slowly. For example, deploying sensor systems at key monitoring points in old, dangerous buildings to monitor the dynamic response time histories of structures, such as displacement and tilt angles, in real time enables dynamic safety monitoring of these buildings. However, these studies focus more on the slow-developing damage mechanisms of dangerous buildings and the long-term prediction of slow-developing risks. Furthermore, the predictive factors are relatively singular, making it difficult to effectively predict risks arising from sudden emergencies and complex factors.

[0005] The scientific and accurate assessment and prediction of the risk of secondary collapse of semi-ruin structures under aftershocks is characterized by urgency, sensitivity, multi-factor nature, real-time capability, and predictability. Urgency manifests in the sudden nature of secondary collapses caused by aftershocks; sensitivity arises from the accumulation of minor damage, thus requiring high sensitivity in predicting secondary collapses of semi-ruin structures; and real-time capability stems from the high frequency of aftershocks during post-earthquake rescue operations, necessitating real-time monitoring. However, relevant research in this area is currently lacking both domestically and internationally. Summary of the Invention

[0006] The present invention aims to provide a scientific and effective method for real-time early warning of secondary collapse risk under aftershocks in RC frame semi-ruin structures, so as to protect the lives of rescuers and improve the rescue efficiency of trapped people.

[0007] To achieve the above objectives, the technical solution of the present invention is as follows:

[0008] A real-time early warning method for the risk of secondary collapse of RC frame semi-ruin structure under aftershocks is proposed. The method uses earthquake simulation shaking table test to obtain the dynamic response time history data of RC frame structure under main shock-aftershock action, and uses machine learning method to realize the real-time early warning of the risk of secondary collapse of RC frame semi-ruin structure under aftershocks.

[0009] This method consists of 4 modules:

[0010] (1) Module 1 is the module for acquiring the dynamic response time history data of the secondary collapse structure of the RC frame semi-ruin structure.

[0011] The earthquake simulation shaking table method was used to realize the entire process of RC frame structure from intact structure to semi-ruin and then to secondary collapse under main shock and aftershock. The dynamic response time history data of RC frame semi-ruin structure under aftershock and secondary collapse were obtained by monitoring key points.

[0012] (2) Module 2 is a multi-dimensional continuous horizontal shift time window data processing module.

[0013] A multi-dimensional transverse continuous time window method is used to process the time history data of structural dynamic response in order to obtain more accurate and richer structural response characteristics.

[0014] (3) Module 3 is the risk classification module for secondary collapse of RC frame semi-ruin structures.

[0015] The K-Means machine learning algorithm was used to classify the risk of secondary collapse of RC frame semi-ruin structures under aftershocks;

[0016] (4) Module 4 is a real-time early warning module for the risk of secondary collapse of RC frame semi-ruin structures.

[0017] The Informer machine learning algorithm was used to provide real-time early warning of the risk of secondary collapse of RC frame semi-ruin structures under aftershocks.

[0018] The proposed method for early warning of secondary collapse risk under aftershocks in RC frame semi-ruin structures involves designing a real-time early warning implementation process for RC frame semi-ruin structures under aftershocks at earthquake rescue sites. First, key monitoring points of the RC frame semi-ruin structure at the earthquake site are identified, and these points are marked with infrared light. High-speed cameras are then deployed in easily accessible locations to monitor and collect time-history data of the structure's dynamic response under aftershocks. Next, the time-history data of the RC frame semi-ruin structure's dynamic response is transmitted from the high-speed cameras to a machine learning data processing server for computation and analysis to obtain the secondary collapse risk category and real-time prediction results.

[0019] The proposed method for early warning of secondary collapse risk under aftershocks in RC frame semi-ruin structures equips earthquake rescue personnel with wireless portable terminal electronic devices. The analysis results from the data processing server are transmitted to the portable electronic devices, enabling real-time risk level identification and prediction of secondary collapse risk under aftershocks in RC frame semi-ruin structures at earthquake sites, and providing real-time early warning of secondary collapse.

[0020] The proposed method for early warning of secondary collapse risk under aftershocks in RC frame semi-ruin structures incorporates K-Means and Informer machine learning algorithms based on high-dimensional continuous time window data within a machine learning data processing server. This method enables real-time early warning of secondary collapse risk in RC frame semi-ruin structures.

[0021] The proposed method for early warning of secondary collapse risk under aftershocks in RC frame semi-ruin structures uses the mean of the cumulative vertical average displacement and the cumulative vertical average velocity absolute value over time at key measuring points as the risk correlation coefficient for secondary collapse.

[0022] The proposed method for early warning of secondary collapse risk under aftershocks in RC frame semi-ruin structures uses a program algorithm that classifies and predicts secondary collapse risk levels in real time through four modules, and then integrates this program algorithm into a data processing server.

[0023] The proposed method for early warning of secondary collapse risk under aftershocks in RC frame semi-ruin structures uses the vertical yielding deformation of emergency rescue support structural components as an evaluation index for secondary collapse under aftershocks in RC frame semi-ruin structures.

[0024] The proposed method for early warning of secondary collapse risk under aftershocks of semi-ruin RC frame structures employs a high-dimensional continuous transverse time window method to process the structural dynamic time history response data, extracting statistical features of 15 parameters, including the vertical displacement, vertical velocity, absolute value of vertical velocity, and vertical acceleration data for each group, as well as their variance, variance increment, and maximum value.

[0025] The proposed method for early warning of secondary collapse risk under aftershocks of RC frame semi-ruin structures uses K-Means based on the time history curve of the absolute value of vertical velocity as the mean of the risk correlation coefficient and sorts the results to classify the secondary collapse risk of RC frame semi-ruin structures into four levels: low risk, medium risk, high risk, and extremely high risk.

[0026] The proposed method for early warning of secondary collapse risk under aftershocks in RC frame semi-ruin structures uses the Informer algorithm to predict the future average cumulative vertical displacement, average cumulative velocity, and average cumulative acceleration, respectively, and maps the prediction results to a given risk level to predict the risk level of data for any future time window in real time.

[0027] The design concept of this invention is:

[0028] This invention employs earthquake simulation shaking table tests to obtain time history data of the dynamic response of an RC frame structure undergoing secondary collapse under mainshock and aftershock conditions. Based on machine learning methods, it achieves real-time early warning of the risk of secondary collapse of the RC frame structure under aftershock conditions. The core content of this invention includes four modules: Module 1: Using an earthquake simulation shaking table, the entire process of an RC frame structure from intact to semi-ruin and then to secondary collapse is simulated through mainshock and aftershock effects, obtaining time history data of the dynamic response of the secondary collapse structure; Module 2: A high-dimensional continuous transverse time window method is used to process the dynamic response time history data of the structure and obtain the dynamic response characteristics of the RC frame semi-ruin secondary collapse structure; Module 3: The K-Means machine learning algorithm is used to cluster the secondary collapse risk types of the RC frame semi-ruin structure; Module 4: The Informer machine learning algorithm is used to achieve real-time early warning of the risk of secondary collapse of the RC frame semi-ruin structure.

[0029] The advantages and beneficial effects of this invention are:

[0030] 1. This invention fills the gap in research on real-time early warning of secondary collapse risk under aftershocks of RC frame semi-ruin structures. It can scientifically and effectively monitor the risk level of secondary collapse of RC frame semi-ruin structures at earthquake rescue sites in real time, and provides valuable reference for emergency, rapid and real-time risk assessment of semi-ruin structures under similar aftershocks or external disturbances.

[0031] 2. This invention provides a scientific and effective real-time early warning of the risk level of secondary collapse of RC frame semi-ruin structures under aftershocks, thereby ensuring the safety of rescue personnel and improving the rescue efficiency of trapped personnel. Attached Figure Description

[0032] Figure 1 This invention presents a flowchart for real-time early warning of secondary collapse risk under aftershocks of RC frame semi-ruin structures at earthquake rescue sites, as well as a method for real-time early warning of secondary collapse risk under aftershocks of RC frame semi-ruin structures.

[0033] Figure 2 yes Figure 1 Module 1 for acquiring time history data of dynamic response of secondary collapse structure in a real-time early warning method for secondary collapse risk under aftershocks of semi-ruin structures with RC frame.

[0034] Figure 3 yes Figure 1 A high-dimensional continuous transverse time window data processing module for a real-time early warning method for the risk of secondary collapse under aftershocks in a semi-ruinous RC frame structure.

[0035] Figure 4 yes Figure 1 The secondary collapse risk classification module 3 of the real-time early warning method for secondary collapse risk under aftershocks of semi-ruin structures with RC frames.

[0036] Figure 5 yes Figure 1 Module 4 of the real-time early warning method for secondary collapse risk under aftershocks of semi-ruin structures with RC frames. Detailed Implementation

[0037] The methods of the present invention will now be clearly and completely described with reference to the accompanying drawings of specific embodiments of the present invention.

[0038] like Figure 1The upper part shows the flowchart of the implementation of the aftershock risk warning for RC frame semi-ruin structures at earthquake rescue sites. Specifically, based on the seismic damage characteristics of RC frame semi-ruin structures, the damage usually occurs at the ends of the RC frame columns. To obtain more accurate dynamic response time history data for RC frame semi-ruin structures, key measuring points are selected at beam-column joints or frame beams that are less prone to damage. To easily mark key measuring points, infrared projection is used to form projection points. A high-speed camera is placed near the RC frame semi-ruin structure on a flat surface, and a seismic isolation device is added to the bottom of the camera bracket to prevent aftershocks from vibrating the camera. The structural dynamic response time history data collected by the high-speed camera is transmitted wirelessly to a machine learning data processing server. The server's computing efficiency must be reliable and robust, and the signal transmission must be real-time and latency-free. The server has a built-in program for classifying and providing real-time warning of the risk of secondary collapse under aftershocks for RC frame semi-ruin structures, and the calculation and transmission results must be controlled within approximately 0.1 seconds. The calculated risk warning results of secondary collapse are transmitted in real time to a portable terminal device carried by rescuers. This device can be a fitness tracker or a watch, which is convenient for rescuers to use.

[0039] like Figure 1 As shown in the lower half, the real-time early warning method for secondary collapse risk under aftershocks of RC frame semi-ruin structures of the present invention is built into a machine learning data processing server. Specifically, the early warning method for secondary collapse risk under aftershocks of RC frame semi-ruin structures includes four modules. Module 1 is the data acquisition module for secondary collapse under aftershocks of RC frame semi-ruin structures. First, the entire process of RC frame structure from intact structure to semi-ruin and then to secondary collapse ruins under the action of main shock and aftershock is simulated through earthquake simulation shaking table test; then, the structural dynamic response time history data of key measuring points is collected through high-speed camera; then, the vertical yield deformation of emergency rescue support structural components is used as the evaluation index for secondary collapse of RC frame semi-ruin structure; finally, the vertical structural dynamic response time history curve data at the time of secondary collapse is acquired, including vertical displacement time history curve data, vertical velocity time history curve data, and vertical acceleration time history curve data (…). Figure 2 Module 2 is for processing high-dimensional continuous horizontal displacement time window data. First, the data is multidimensionalized by processing the vertical displacement time history data, vertical velocity time history data, vertical acceleration time history data, and vertical velocity absolute value time history curves into variance, variance increment, and maximum value time history data. This yields 15 dimensional variables X = {x1,...,x...} for each window, including the data itself. 15 Then, each data point is divided into N time windows, and the 15-dimensional variable data are distributed into N time windows, where N is the number of structural dynamic response time points; finally, the data is normalized. Figure 3Module 3 classifies the risk of secondary collapse. First, the average of the cumulative vertical displacement and vertical velocity absolute values ​​over time at key measuring points is used as the risk correlation coefficient for secondary collapse. Then, a lower bound K for clustering is selected. min The value is 2, and the upper bound K for clustering is... max The value is 20, and K is executed. min -means algorithm generates K min Each cluster is initially defined, and its quality is calculated using the BIC Score. Then, each cluster is split into two sub-clusters using a 2-Means algorithm. The BIC Score of the split sub-clusters is compared to that of the original clusters. If the BIC Score of the split sub-clusters is better than that of the original clusters, the sub-clusters are replaced with the original clusters, and the number of clusters is updated. This process is repeated until the cluster limit is reached or no further sub-clustering is possible. Finally, each risk correlation coefficient is classified into four risk levels: low risk, medium risk, high risk, and very high risk. Figure 4 Module 4 is for real-time prediction of secondary collapse risk. First, the Informer algorithm is used to predict future average cumulative vertical displacement, average cumulative velocity, and average cumulative acceleration. Then, the prediction results are mapped to a given risk level. Next, the highest risk value is assigned to the current risk level. Finally, the risk level for any future time window is predicted in real time. Figure 5 ).

[0040] The results show that, in order to fill the gap in the research on the risk warning of secondary collapse of semi-ruin structures under aftershocks, this invention uses the method of earthquake simulation shaking table test to obtain the dynamic response time history data of the secondary collapse structure of RC frame semi-ruin structure under aftershocks, and realizes real-time early warning of the risk of secondary collapse of RC frame semi-ruin structure under aftershocks based on machine learning method.

Claims

1. A method for real-time early warning of secondary collapse risk under aftershocks in a semi-ruin RC frame structure, characterized in that, Seismic simulation shaking table tests were used to obtain the time history data of the dynamic response of the RC frame structure under the action of main shock and aftershock, and machine learning methods were used to realize the real-time early warning of the risk of secondary collapse of the RC frame semi-ruin structure under aftershock. This method consists of 4 modules: (1) Module 1 is the module for acquiring the dynamic response time history data of the secondary collapse structure of the RC frame semi-ruin structure. The earthquake simulation shaking table method was used to realize the entire process of RC frame structure from intact structure to semi-ruin and then to secondary collapse under main shock and aftershock. The dynamic response time history data of RC frame semi-ruin structure under aftershock and secondary collapse were obtained by monitoring key points. (2) Module 2 is a multi-dimensional continuous horizontal shift time window data processing module. A multi-dimensional transverse continuous time window method is used to process the time history data of structural dynamic response in order to obtain more accurate and richer structural response characteristics. (3) Module 3 is the risk classification module for secondary collapse of RC frame semi-ruins structures. The K-Means machine learning algorithm was used to classify the risk of secondary collapse of RC frame semi-ruin structures under aftershocks; (4) Module 4 is a real-time early warning module for the risk of secondary collapse of RC frame semi-ruin structure. The Informer machine learning algorithm is used to predict the risk of data, and the prediction results are mapped to risk levels. The risk level of data in any future time window is predicted in real time, and real-time warnings are issued.

2. The real-time early warning method for secondary collapse risk under aftershocks of RC frame semi-ruin structures according to claim 1, characterized in that, This method designs a real-time early warning implementation process for the risk of secondary collapse of RC frame semi-ruins structure under aftershocks at earthquake rescue sites: First, the key measuring points of RC frame semi-ruins structure at earthquake sites are determined, the monitoring points are marked with infrared, and high-speed cameras are set up in easily monitored locations to monitor and collect the dynamic response time history data of RC frame semi-ruins structure under aftershocks. Then, the dynamic response time history data of the RC frame semi-ruin structure is transmitted from the high-speed camera to the machine learning data processing server for calculation and analysis to obtain the secondary collapse risk category and real-time prediction results.

3. The real-time early warning method for secondary collapse risk under aftershocks of RC frame semi-ruin structures according to claim 2, characterized in that, Equip earthquake rescue personnel with wireless portable terminal electronic devices to transmit the analysis results of the data processing server to the portable electronic devices, so as to realize the real-time risk level identification and real-time prediction of the risk of secondary collapse under aftershocks of RC frame semi-ruin structures at the earthquake site, and to provide real-time early warning of secondary collapse.

4. The real-time early warning method for secondary collapse risk under aftershocks of RC frame semi-ruin structures according to claim 2, characterized in that, The machine learning data processing server incorporates K-Means and Informer machine learning algorithms based on high-dimensional continuous time window data to implement a real-time early warning method for the secondary collapse risk of RC frame semi-ruin structures.

5. The real-time early warning method for secondary collapse risk under aftershocks of RC frame semi-ruin structures according to claim 2, characterized in that, The mean of the cumulative vertical average displacement and the cumulative vertical average velocity absolute value over time at key measuring points is used as the risk correlation coefficient for secondary collapse.

6. The method for real-time early warning of secondary collapse risk under aftershocks in a semi-ruin RC frame structure according to claim 1, characterized in that, This method uses four modules to obtain a program algorithm for classifying and predicting secondary collapse risk levels in real time, and then integrates this program algorithm into the data processing server.

7. The real-time early warning method for secondary collapse risk under aftershocks of RC frame semi-ruin structures according to claim 1, characterized in that, The vertical yield deformation of emergency rescue support structural components is used as an evaluation index for secondary collapse of RC frame semi-ruin structures under aftershocks.

8. The method for real-time early warning of secondary collapse risk under aftershocks in a semi-ruin RC frame structure according to claim 1, characterized in that, The high-dimensional continuous transverse time window method was used to process the structural dynamic time history response data, and statistical features of 15 parameters were extracted, including the vertical displacement, vertical velocity, absolute value of vertical velocity and vertical acceleration data itself, as well as their variance, variance increment and maximum value.

9. The method for real-time early warning of secondary collapse risk under aftershocks in a semi-ruin RC frame structure according to claim 1, characterized in that, Using K-Means to rank the risk correlation coefficients based on the time history curves of the absolute value of vertical velocity, the risk of secondary collapse of RC frame semi-ruin structures is divided into four levels: low risk, medium risk, high risk, and extremely high risk.

10. The method for real-time early warning of secondary collapse risk under aftershocks in a semi-ruin RC frame structure according to claim 1, characterized in that, The Informer algorithm is used to predict the future average cumulative vertical displacement, average cumulative velocity, and average cumulative acceleration, respectively. The prediction results are then mapped to a given risk level to predict the risk level of data for any future time window in real time.

Images