Search

CN-121996404-A - Seismic vibration model construction and evaluation method based on big data

CN121996404ACN 121996404 ACN121996404 ACN 121996404ACN-121996404-A

Abstract

The invention relates to the technical field of seismic engineering, in particular to a method for constructing and evaluating a seismic vibration model based on big data, which comprises a hardware and rule component, a platform architecture component (a relational/non-relational database, a data warehouse, a metadata center, an association rule/clustering algorithm module, an LSTM/GRU seismic risk prediction model, a resource scheduling model, a scientific research/emergency/public differentiation application function module), a model construction component (a finite difference/finite element multi-scale field model, a building group/transmission tower structure-field coupling sub-model, a seismology-deep learning mixed training frame and a model parameter phasing correction module), an evaluation component and iteration optimization component.

Inventors

  • LI ZHENHUA

Assignees

  • 上海钛擎机器人有限公司

Dates

Publication Date
20260508
Application Date
20251204

Claims (10)

  1. 1. A method for constructing and evaluating a seismic vibration model based on big data is characterized by comprising the following steps: S100, intelligent acquisition and deep treatment of multi-source heterogeneous seismic data, acquisition of seismic monitoring data, geological and seismic source data, structure and disaster data, transmission to a data center through wired and wireless double links, scheduling of hardware resources based on a Lasso algorithm, and adoption of Raft consensus algorithm to ensure data consistency; S200, constructing a three-layer architecture of a large earthquake data platform, classifying and storing structured and unstructured data by a storage layer through a relational database and a non-relational database, integrating a data warehouse and associating a metadata center, deploying association rule algorithm, clustering algorithm and an LSTM/GRU-based earthquake risk prediction model by a processing layer, and allocating processing tasks by combining a resource scheduling model; s300, constructing a multiscale site model by combining a finite difference method and a finite element method, constructing a structure-site coupling sub-model aiming at a building group and a power transmission tower, and correcting model parameters in stages by adopting a seismology theory and deep learning mixed frame training model; S400, multi-dimensional model evaluation and dynamic risk evaluation are carried out, a training set and a testing set are divided, a pull Ding Chao cube sampling expansion sample is adopted, a structural failure probability is calculated based on poisson binomial distribution through an index evaluation model such as simulation precision and response spectrum consistency, a multi-intensity seismic scene is simulated, and a non-standard model is optimized; And S500, performing feedback type iterative optimization on the platform and the model, periodically updating data and metadata, adjusting the platform function according to user feedback, retraining the model by using full data every 2 years, and dynamically updating the seismic velocity model based on generating the data for supplementing extreme working conditions to the countermeasure network.
  2. 2. The method for constructing and evaluating a seismic vibration model based on big data as set forth in claim 1, wherein in S100, the collision intensity is calculated during data collision processing Vi is the parameter value of the ith data source, And when the collision intensity is less than or equal to 0.3, combining with geological data verification, and when the collision intensity is more than or equal to 0.5, manually checking.
  3. 3. The method for constructing and evaluating the seismic vibration model based on big data according to claim 1 is characterized in that in S200, a relational database of a storage layer is MySQL and stores structural data such as seismic event parameters and station information, an unstructured database is MongoDB and stores unstructured data such as seismic waveform files and BIM models, and a data warehouse supports multidimensional query according to a reliability weight and a time range.
  4. 4. The method for constructing and evaluating the seismic vibration model based on the big data according to claim 1 is characterized in that in S300, model parameters are corrected in a staged mode, namely, the self-vibration period and the elastic displacement of a structure monitored by small vibration in an elastic stage are corrected through a particle swarm optimization algorithm, and the structural skeleton curve is corrected through big-vibration historical damage data and an incremental dynamic analysis curve library in an elastoplastic stage.
  5. 5. The method for constructing and evaluating the seismic vibration model based on the big data according to claim 1, wherein in S400, the simulation precision index comprises a Root Mean Square Error (RMSE) and a correlation coefficient R, wherein R is more than or equal to 0.8, RMSE is less than or equal to 10% of industry mean value, the response spectrum consistency index comprises a displacement response maximum value Sd and a velocity response maximum value Sv, and the deviation between the simulation value and the actual value is less than or equal to 15%.
  6. 6. The method for constructing and evaluating the seismic vibration model based on big data according to claim 1, wherein in S500, the frequency of periodically updating data is that real-time seismic waveform data are updated every second, structural monitoring data are updated every month, geological and historical earthquake damage data are updated every 5 years, and when the seismic velocity model is dynamically updated, conventional quarter updating occurs, and emergency updating occurs within 72 hours after Mw is greater than or equal to 5-level earthquake.
  7. 7. The method for constructing and evaluating a seismic vibration model based on big data as claimed in claim 1, wherein in S100, the reliability weight is calculated by weighting the station precision (0.8-1.0), the data integrity (0.6-1.0) and the conflict processing result (0.7-1.0), the formula is w=0.4w_ +0.3w_complete+0.3w_conflict, and the data with weight <0.6 is marked as low quality and collected again.
  8. 8. A seismic vibration model construction and evaluation system based on big data, for implementing the method according to any one of claims 1-7, comprising the following modules: The system comprises a data acquisition and treatment module, a platform architecture module, a model construction module, an evaluation and evaluation module and an iteration optimization module; The data acquisition and management module acquires data of earthquake monitoring, geology, earthquake focus, structure and disasters, transmits the data to the data center through wired and wireless double links, schedules hardware resources based on a Lasso algorithm, adopts Raft consensus algorithm to ensure data consistency, cleans and conflict processes the data, unifies a format and a coordinate system, attaches a credibility weight label and dynamically updates a management rule base; The platform architecture module is used for constructing a three-layer architecture, classifying and storing structured and unstructured data by using a relational database and a non-relational database by using a storage layer, integrating a data warehouse association metadata center, deploying association rules, clustering algorithms and LSTM/GRU earthquake risk prediction models by using a processing layer, and allocating tasks by combining a resource scheduling model; The model construction module is used for constructing a multi-scale site model by combining a finite difference and finite element method, constructing a structure-site coupling sub-model aiming at a building group and a power transmission tower, adopting a seismology theory and deep learning mixed frame training model and correcting parameters in stages; dividing a training set and a testing set, sampling an expanded Latin hypercube sample, evaluating a model through simulation precision and response spectrum consistency indexes, calculating the structural failure probability based on poisson binomial distribution, simulating a multi-intensity seismic scene and optimizing a non-standard model; And the iterative optimization module is used for updating the data and the metadata regularly, adjusting the platform function according to the feedback of the user, retraining the model by using the full data every 2 years, and dynamically updating the earthquake velocity model based on generating the data for supplementing extreme working conditions to the countermeasure network.
  9. 9. A computer device comprising a processor and a memory, wherein the memory stores at least one instruction, at least one program, code set, or instruction set, the at least one instruction, the at least one program, the code set, or instruction set being loaded and executed by the processor to implement a method of constructing and evaluating a seismic vibration model based on big data as claimed in any one of claims 1 to 7.
  10. 10. A computer readable storage medium having stored therein at least one instruction, at least one program, code set, or instruction set, the at least one instruction, the at least one program, the code set, or instruction set being loaded and executed by a processor to implement a method of constructing and evaluating a seismic vibration model based on big data as claimed in any one of claims 1 to 7.

Description

Seismic vibration model construction and evaluation method based on big data Technical Field The invention relates to the technical field of seismic engineering, in particular to a method for constructing and evaluating a seismic vibration model based on big data. Background In the fields of earthquake monitoring, disaster prevention and reduction, along with the development of sensor technology, internet of things and deep learning, earthquake related data form a multi-source heterogeneous complex system, which comprises real-time waveform data of an earthquake monitoring table network, stratum and fault parameters of geological exploration, a BIM model of an engineering structure, historical earthquake hazard records and the like. However, the prior art has a significant bottleneck in practical application: Firstly, the data acquisition link depends on a single transmission link, data interruption is easy to be caused by equipment failure or environmental interference, an intelligent resource scheduling mechanism is lacked, the utilization rate of hardware resources is less than 50%, meanwhile, multi-source data conflict processing depends on manual verification, the efficiency is low, the data reliability cannot be quantized, and a high-quality data source is difficult to form; Secondly, the existing platform architecture mostly adopts a single storage mode, the differential storage requirements of the structured seismic parameters and unstructured waveform files cannot be adapted, the data processing and application services are disjointed, and multi-role requirements of scientific researchers, emergency departments and the like are difficult to respond quickly; Thirdly, the seismic vibration model is used for independently constructing a site or a structural model, depth coupling is not realized, simulation precision can only meet macro area analysis, fine evaluation of local key areas cannot be supported, model evaluation only pays attention to simulation errors, structural damage probability and risk level matching degree are ignored, meanwhile, an extreme seismic working condition data supplementing mechanism is lacked, and the prediction capability on rare seismic events is weak; Fourthly, platform functions and model parameter updating depend on manual triggering, geological condition change and newly-added data cannot be dynamically adapted, precision attenuation is obvious after long-term use, and reliable support is difficult to continuously provide for seismic scientific research and emergency decision; therefore, a need exists for an integrated construction method that covers the full life cycle of data. Based on the problems, the invention provides a method for constructing and evaluating a seismic vibration model based on big data. Disclosure of Invention Aiming at the defects of the prior art, the invention provides a seismic vibration model construction and evaluation method based on big data, which solves the problems that a multi-source seismic data acquisition link is single and easy to break, resource scheduling is low-efficient, data reliability is difficult to quantify, a platform architecture cannot adapt to heterogeneous data storage and multi-role demands, the seismic model has low coupling degree, single evaluation dimension, weak extreme working condition prediction and lag between platform and model update. In order to achieve the above object, one of the present invention is to provide a method for constructing and evaluating a seismic vibration model based on big data, comprising; The intelligent acquisition and the deep treatment of multi-source heterogeneous seismic data acquire seismic monitoring data, geological and seismic source data, structure and disaster data, and the data are transmitted to a data center through wired and wireless double links, hardware resources are scheduled based on Lasso algorithm, and Raft consensus algorithm is adopted to ensure data consistency; The method comprises the steps of constructing a three-layer architecture of a large earthquake data platform, classifying and storing structured and unstructured data by a storage layer through a relational database and a non-relational database, integrating a data warehouse and associating a metadata center, deploying an association rule algorithm, a clustering algorithm and an LSTM/GRU-based earthquake risk prediction model by a processing layer, and allocating processing tasks by combining a resource scheduling model; constructing a multiscale site model by combining a finite difference and finite element method, constructing a structure-site coupling sub-model aiming at a building group and a power transmission tower, and correcting model parameters in stages by adopting a seismology theory and deep learning mixed frame training model; The method comprises the steps of multi-dimensional model evaluation and dynamic risk evaluation, dividing a training set and a testing set, sampling an expanded