A seismic isolation support full life cycle causality monitoring method and system

Abstract

This invention discloses a method and system for causal monitoring of seismic isolation bearings throughout their entire life cycle. The method includes: collecting production and testing data of the seismic isolation bearings to form a priori parameter package bound to the bearing ID; collecting monitoring data of the seismic isolation bearings during their service life and performing multimodal self-calibration, increasing the sampling rate when abnormal events are triggered; constructing a causal inference model of the mechanical performance evolution of the seismic isolation bearings based on the preset priori parameter package and monitoring data, and dynamically updating the causal inference model through causal identification, counterfactual normalization, and Bayesian parameter learning; inputting the priori parameter package and monitoring data of the target seismic isolation bearing into the causal inference model, and outputting the predicted mechanical performance parameters of the target seismic isolation bearing. This invention can achieve real-time and accurate prediction of the mechanical performance of seismic isolation bearings during their service life.

Description

Technical Field

[0001] This invention relates to the field of structural health monitoring technology, and in particular to a method and system for monitoring the causal relationship of seismic isolation bearings throughout their entire life cycle. Background Technology

[0002] Seismic isolation bearings are core load-bearing components of seismic isolation and damping systems in engineering structures such as buildings, bridges, and hydraulic structures. They are widely used in important facilities such as water intake towers, gate foundations, factory frames, and municipal bridges. The stability of their mechanical properties directly affects the safe service and long-term durability of the superstructure under loads such as earthquakes and vibrations. The service environment of seismic isolation bearings is usually quite harsh, subjecting them to the coupled effects of multiple factors such as long-term periodic fluctuations in temperature and humidity, water leakage or immersion, long-term constant loads, seismic impacts, and random vibrations. This can easily lead to a series of performance degradation problems, such as aging of rubber materials, delamination of interlayer bonds, yielding and degradation of the internal lead core, contamination and wear of the sliding contact surface, loosening of anchor bolts, and abrasion damage to the limiting components. These problems result in a decline in the core mechanical properties of the bearings, significantly reducing the overall seismic isolation effect of the structure and creating potential safety hazards.

[0003] Currently, the performance testing of seismic isolation bearings still relies primarily on traditional manual periodic inspections. This method can only observe the surface condition of the bearings, such as damage and deformation, and cannot accurately obtain real-time changes in internal mechanical properties. The test results are significantly biased and lagging. Offline testing of bearing performance requires complex operations such as disassembly and in-situ loading. This is not only labor-intensive and cumbersome, but also poses safety risks such as working at heights and temporary unloading of the structure. Furthermore, it can affect the normal use of the structure, thus limiting its practical application.

[0004] With the development of structural health monitoring technology, devices such as accelerometers, displacement gauges, and pressure sensors are gradually being used for bearing condition monitoring. However, these devices can only collect basic physical quantities such as acceleration, displacement, and axial compression, and cannot directly infer the core mechanical properties of the bearing. Most existing active sensing monitoring methods rely on data-driven deep learning models, which require massive amounts of data to support modeling. The model construction is complex, and the on-site deployment is difficult. Furthermore, they cannot distinguish between reversible performance fluctuations caused by environmental factors and irreversible performance degradation caused by material deterioration, which can easily lead to false performance warnings and make it difficult to achieve accurate and reliable monitoring of the mechanical properties of seismic isolation bearings throughout their entire life cycle. Summary of the Invention

[0005] To address the aforementioned issues, this invention proposes a method and system for monitoring the causal relationship of seismic isolation bearings throughout their entire lifespan, enabling real-time and accurate prediction of the mechanical properties of the bearings during their service life.

[0006] The technical solution adopted in this invention is as follows: A method for monitoring the causal relationship of seismic isolation bearings throughout their entire life cycle includes: Collect production and testing data of seismic isolation bearings to form a priori parameter package that is bound to the seismic isolation bearing ID; The system collects monitoring data of the seismic isolation bearings during their service life and performs multimodal self-calibration, increasing the sampling rate when abnormal events are triggered; the monitoring data includes axial compression, displacement, temperature and humidity, and water quality. Based on the pre-set parameter package of the seismic isolation bearing and the monitoring data, a causal reasoning model for the evolution of the mechanical performance of the seismic isolation bearing is constructed, and the causal reasoning model is dynamically updated through causal identification, counterfactual normalization and Bayesian parameter learning. The prior parameter package of the target seismic isolation bearing and the monitoring data are input into the causal inference model, and the predicted mechanical performance parameters of the target seismic isolation bearing are output.

[0007] Furthermore, the process of collecting factory production and testing data for seismic isolation bearings to form a priori parameter package bound to the seismic isolation bearing ID includes: Small-sample orthogonal design experiment: Based on the standard shear test, hysteresis tests at multiple environmental points are added, including temperature T, relative humidity RH, and loading frequency. f This forms a multidimensional excitation matrix; Reversible / Irreversible Kernel Decomposition: Decomposing the hysteretic characteristic Y into reversible environmental kernel and irreversible degradation kernel:

[0008] In the formula, As a reversible environmental core, It is an irreversibly degraded core. t For service time, To accumulate cycle energy, It is cyclic energy; Hysteresis fingerprint and prior parameter package generation: Multi-point force-displacement samples are used to form a hysteresis fingerprint hash, which is then combined with the initial performance parameters to form a prior parameter package and bound to the seismic isolation bearing ID.

[0009] Furthermore, the initial performance parameters include: reversible environmental core. Initial horizontal equivalent stiffness K eq0 Initial equivalent damping ratio ξ eq0 Initial yield force F y0 Initial post-yield stiffness K p0 Initial elastic modulus G0, initial friction coefficient μ0, initial bolt preload bolt0, and initial limit gap lim0.

[0010] Furthermore, the process of collecting monitoring data during the service life of the seismic isolation bearings and performing multimodal self-calibration includes: Axial compression acquisition: Force sensors are arranged on the seismic isolation bearing, bearing connecting plate or supporting component to acquire the axial compression on the seismic isolation bearing in real time or periodically, and obtain axial compression time history data; Displacement acquisition: Two horizontal displacement gauges are installed on the seismic isolation bearing to record the relative displacement between the upper and lower connecting plates; accelerometers are arranged to record the response curve of the seismic isolation bearing under earthquake or vibration. Temperature and humidity data collection: When the seismic isolation bearing is equipped with an isolation device on its outer side, i.e., it is not in contact with water, the humidity and temperature of the environment in which the seismic isolation bearing is located are recorded; when the seismic isolation bearing is not equipped with an isolation device on its outer side, i.e., it is submerged in water, the water temperature T and water quality parameters are recorded, including electrical conductivity. Dissolved oxygen (DO), pH value, and turbidity (NTU); Water Quality Degradation Index (WDI) calculation: The WDI is calculated by normalizing water temperature and water quality.

[0011] In the formula, For weights updated based on Bayesian methods, the quantities marked with "-" are the normalized risk coefficients; Multimodal self-calibration: periodic solution Least squares regression, combined with reversible environment kernel Temperature compensation and sensor drift diagnosis are performed; exceeding limits triggers coefficient self-update and quality downweighting. The output quantity to be calibrated , and All are regression coefficients. As a baseline reference value, d For the amount of sensor drift, T For temperature.

[0012] Furthermore, increasing the sampling rate when an abnormal event is triggered includes: When an earthquake, rapid thermocline change, or displacement / acceleration exceeds a preset threshold occurs, the sampling rate is automatically increased and event features are extracted and archived separately from normal data; the event features include hysteresis area, peak / mean square energy, and residual displacement.

[0013] Furthermore, the causal reasoning model for the evolution of the mechanical performance of the seismic isolation bearing, constructed based on the pre-set parameter package of the seismic isolation bearing and monitoring data, includes: Using the pre-set parameter package of the seismic isolation bearing and the service monitoring data as input, and the measured mechanical performance data as output, a causal reasoning model for the evolution of the mechanical performance of the seismic isolation bearing is established based on a causal Bayesian network. The preset seismic isolation bearing refers to a reference bearing that is from the same batch, has the same specifications, and is in a similar service environment as the target seismic isolation bearing, and serves as the basis for model construction; the reference bearing is subjected to 100% shear performance testing at a set frequency to obtain its measured mechanical performance data; The target seismic isolation bearing is the monitoring object and is not removed from the original structure during service. By inputting its prior parameter package and online monitoring data, the mechanical performance prediction results of the target seismic isolation bearing are obtained.

[0014] The difference between the preset seismic isolation bearing and the target seismic isolation bearing is that the preset seismic isolation bearing is not in service, but in a simulated service state of the target seismic isolation bearing, so as to facilitate the acquisition of its mechanical performance test data.

[0015] Furthermore, the causal identification includes: using a pre-set set of prior parameters for the seismic isolation bearing and historical monitoring data during service, and employing a constraint-based algorithm or a scoring-based algorithm to determine the initial causal structure between variables; The initial causal structure includes: Reversible layer: ; In the formula, For the reversible component of the hysteresis characteristic, This is a reversible environmental nucleus, where T is temperature and RH is relative humidity. f For loading frequency, t Service life; Irreversible layer: ; In the formula, For irreversible deterioration of state variables, Let WDI be the state evolution function, and WDI be the water quality deterioration index. For time step, This refers to the perturbation term during the irreversible degradation process; Output layer: ; In the formula, Y is the hysteresis characteristic quantity. h As the output mapping function, L can identify the load or input. It is cyclic energy; Automatic Minimum Backdoor Set: Based on the preset cause-effect graph, variable time sequence constraints, and seismic isolation bearing mechanical mechanism constraints, the set of adjustment variables that satisfies the backdoor criterion is selected from the candidate set of adjustment variables. Then, the set with the smallest size and statistical stability that meets the preset conditions is selected from the set of adjustment variables that satisfies the backdoor criterion as the automatic minimum backdoor set. Subsequently, based on the automatic minimum backdoor set The causal effect of the load or input L on the net performance quantity can be calculated by residual regression or backdoor adjustment integral formula; the automatic minimum backdoor set is updated according to data distribution drift, abnormal event termination or monitoring condition changes; The backdoor criteria include: adjusting the variable set to not contain descendant nodes of the identifiable load or input L, and blocking the identifiable load or input L from reaching net energy. The non-causal path between them, the net energy The performance quantity after removing the effects of reversible environmental factors.

[0016] Furthermore, the counterfactual normalization includes: Determine the set of environment variables E and benchmark environment E 0, the set of environment variables E This includes one or more of the following: temperature, humidity, loading frequency, water temperature, and water quality damage index; the reference environment. E 0 represents the stable environment during the initial service phase; a reversible environment core is fitted based on the results of multi-environment hysteresis tests conducted at the factory. And based on the currently observed hysteresis characteristics Calculate net observations :

[0017] The net observation measurement As observational evidence, it is used to update irreversibly deteriorated state variables. The net observation measurement This is equivalent to normalizing the current observation results to a baseline environment. E The counterfactual observation following 0, that is, the one corresponding to... The performance observation results are as follows.

[0018] Furthermore, the dynamic update of the Bayesian parameter learning includes: Using the factory test results of seismic isolation bearings as the prior distribution and real-time monitoring data during service as observational evidence, the conditional probability table (CPT) or regression coefficients between nodes are continuously updated through the Bayesian parameter learning method to achieve dynamic evolution estimation of bearing performance status. After obtaining new observation data, perform the following operations:

[0019] In the formula, J These are the mechanical performance parameters of the seismic isolation bearing. A set of environment variables. For the irreversible deterioration state variables of the seismic isolation bearing, For a given set of environment variables Time-irreversible degradation of state variables The posterior probability distribution, Given an irreversibly deteriorated state variable Mechanical performance parameters The conditional probability distribution; This means performing a marginal integration over all possible irreversibly deteriorating state variables to obtain a given set of environment variables. Lower mechanical property parameters The posterior predicted distribution.

[0020] A seismic isolation bearing life-cycle causal monitoring system includes: The seismic isolation bearing factory information acquisition module is configured to collect seismic isolation bearing factory production and test data to form a priori parameter package bound to the seismic isolation bearing ID; The seismic isolation bearing service environment information acquisition module is configured to collect monitoring data during the service life of the seismic isolation bearing and perform multimodal self-calibration, and increase the sampling rate when abnormal events are triggered; the monitoring data includes axial compression, displacement, temperature and humidity, and water quality; The causal reasoning model construction and update module is configured to construct a causal reasoning model of the mechanical performance evolution of the seismic isolation bearing based on the pre-set a priori parameter package and monitoring data of the seismic isolation bearing, and dynamically update the causal reasoning model through causal identification, counterfactual normalization and Bayesian parameter learning. The seismic isolation bearing performance change prediction module is configured to input the prior parameter package of the target seismic isolation bearing and the monitoring data into the causal inference model, and output the predicted results of the mechanical performance parameters of the target seismic isolation bearing.

[0021] The beneficial effects of this invention are as follows: 1. This invention integrates the manufacturing and testing data of the bearing into a priori parameter package with a unique identifier, and merges multi-source monitoring data during service life. This enables the integrated association of data throughout the entire life cycle of the bearing, fully adapting to the initial performance characteristics and specific service environment of a single bearing, and improving the accuracy of performance monitoring and prediction from the data source.

[0022] 2. This invention collects axial pressure, displacement, temperature, humidity and water quality parameters differently depending on whether the support is in a water environment. It quantifies the environmental erosion effect through the water quality damage index, automatically increases the sampling rate and completes multimodal self-calibration under abnormal events, effectively eliminates errors such as sensor drift and environmental interference, ensures the stability of monitoring data, and completely retains the response characteristics of the support under normal and extreme working conditions.

[0023] 3. This invention constructs a causal reasoning model that includes reversible and irreversible layers. By decoupling the immediate environmental impact and long-term degradation effect through kernel decomposition, and combining automatic minimum backdoor set optimization and counterfactual normalization processing, it eliminates false alarms caused by rapid environmental changes, accurately identifies the true degradation trend of the support, and solves the technical problem that traditional monitoring cannot distinguish between reversible and irreversible changes.

[0024] 4. This invention uses factory test data as the prior distribution and real-time monitoring data as observational evidence. It achieves dynamic model updates through Bayesian parameter learning, which can complete model iterative optimization without a large number of samples, reducing the difficulty of modeling and on-site deployment, and realizing real-time and dynamic estimation of the mechanical performance of the support.

[0025] 5. This invention can directly output predicted results of the core mechanical performance parameters of the bearings, reducing the frequency of manual inspections, significantly reducing the workload and operational safety risks, and realizing the automation and intelligence of seismic isolation bearing monitoring. Simultaneously, relying on counterfactual intervention analysis, it can quantify the impact of environmental and load adjustments on bearing performance, providing quantitative basis for the optimized design of seismic isolation structures, bearing operation and maintenance decisions, and the formulation of protection strategies, comprehensively improving the level of operation and maintenance management and structural safety assurance capabilities throughout the entire life cycle of seismic isolation bearings. Attached Figure Description

[0026] Figure 1 This is a flowchart of a method for monitoring the causal relationship of a seismic isolation bearing throughout its entire life cycle, according to Embodiment 1 of the present invention. Detailed Implementation

[0027] To provide a clearer understanding of the technical features, objectives, and effects of the present invention, specific embodiments are now described. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention; that is, the described embodiments are only a part of the embodiments of the invention, not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0028] Example 1 like Figure 1 As shown, this embodiment provides a method for monitoring the causal relationship of a seismic isolation bearing throughout its entire life cycle, including: Collect production and testing data of seismic isolation bearings to form a priori parameter package that is bound to the seismic isolation bearing ID; Collect production and testing data of seismic isolation bearings to form a priori parameter package that is bound to the seismic isolation bearing ID; The system collects monitoring data of the seismic isolation bearings during their service life and performs multimodal self-calibration, increasing the sampling rate when abnormal events are triggered; the monitoring data includes axial compression, displacement, temperature and humidity, and water quality. Based on the pre-set parameter package of the seismic isolation bearing and the monitoring data, a causal reasoning model for the evolution of the mechanical performance of the seismic isolation bearing is constructed, and the causal reasoning model is dynamically updated through causal identification, counterfactual normalization and Bayesian parameter learning. The prior parameter package of the target seismic isolation bearing and the monitoring data are input into the causal inference model, and the predicted mechanical performance parameters of the target seismic isolation bearing are output.

[0029] Preferably, the step of collecting factory production and testing data of the seismic isolation bearings to form a priori parameter package bound to the seismic isolation bearing ID includes: Small-sample orthogonal design experiment: Based on the standard shear test, hysteresis tests at multiple environmental points are added, including temperature T, relative humidity RH, and loading frequency. fThis forms a multidimensional excitation matrix; Reversible / Irreversible Kernel Decomposition: Decomposing the hysteretic characteristic Y into reversible environmental kernel and irreversible degradation kernel:

[0030] In the formula, As a reversible environmental core, It is an irreversibly degraded core. t For service time, To accumulate cycle energy, It is cyclic energy; Hysteresis fingerprint and prior parameter package generation: Hysteresis fingerprint hashes are constructed from multi-point force-displacement samples and combined with initial performance parameters to form a prior parameter package, which is then bound to the seismic isolation bearing ID. More preferably, the initial performance parameters include: a reversible environmental core... Initial horizontal equivalent stiffness K eq0 Initial equivalent damping ratio ξ eq0 Initial yield force F y0 Initial post-yield stiffness K p0 Initial elastic modulus G0, initial friction coefficient μ0, initial bolt preload bolt0, and initial limit gap lim0.

[0031] It should be noted that by using small-sample experiments and kernel decomposition to accurately extract the initial performance characteristics of the support, exclusive prior data is formed, providing a stable and reliable data foundation for subsequent causal inference.

[0032] Preferably, the step of collecting monitoring data during the service life of the seismic isolation bearing and performing multimodal self-calibration includes: Axial compression acquisition: Force sensors are arranged on the seismic isolation bearing, bearing connecting plate or supporting component to acquire the axial compression on the seismic isolation bearing in real time or periodically, and obtain axial compression time history data; Displacement acquisition: Two horizontal displacement gauges are installed on the seismic isolation bearing to record the relative displacement between the upper and lower connecting plates; accelerometers are arranged to record the response curve of the seismic isolation bearing under earthquake or vibration. Temperature and humidity data collection: When the seismic isolation bearing is equipped with an isolation device on its outer side, i.e., it is not in contact with water, the humidity and temperature of the environment in which the seismic isolation bearing is located are recorded; when the seismic isolation bearing is not equipped with an isolation device on its outer side, i.e., it is submerged in water, the water temperature T and water quality parameters are recorded, including electrical conductivity. Dissolved oxygen (DO), pH value, and turbidity (NTU); Water Quality Degradation Index (WDI) calculation: The WDI is calculated by normalizing water temperature and water quality.

[0033] In the formula, For weights updated based on Bayesian methods, the quantities marked with "-" are the normalized risk coefficients; Multimodal self-calibration: periodic solution Least squares regression, combined with reversible environment kernel Temperature compensation and sensor drift diagnosis are performed; exceeding limits triggers coefficient self-update and quality downweighting. The output quantity to be calibrated , and All are regression coefficients. As a baseline reference value, d For the amount of sensor drift, T For temperature.

[0034] It should be noted that multi-dimensional data acquisition adapts to different service environments, and self-calibration processing eliminates sensor drift errors, ensuring the stability and accuracy of monitoring data.

[0035] Preferably, increasing the sampling rate when an abnormal event is triggered includes: When an earthquake, rapid thermocline change, or displacement / acceleration exceeds a preset threshold occurs, the sampling rate is automatically increased and event features are extracted and archived separately from normal data; the event features include hysteresis area, peak / mean square energy, and residual displacement.

[0036] It should be noted that high-frequency data collection under abnormal conditions can completely preserve key response data, and categorized archiving facilitates subsequent specialized analysis, effectively improving the effectiveness of abnormal condition identification.

[0037] Preferably, the causal reasoning model for the evolution of the mechanical performance of the seismic isolation bearing, based on the pre-set parameter package of the seismic isolation bearing and monitoring data, includes: Using the pre-set parameter package of the seismic isolation bearing and the service monitoring data as inputs, and the measured mechanical performance data as outputs, a causal reasoning model for the evolution of the mechanical performance of the seismic isolation bearing is established based on a causal Bayesian network.

[0038] The preset seismic isolation bearing refers to a reference bearing that is from the same batch, has the same specifications, and is in a similar service environment as the target seismic isolation bearing, and serves as the basis for model construction; the reference bearing is subjected to 100% shear performance testing at a set frequency to obtain its measured mechanical performance data; The target seismic isolation bearing is the monitoring object and is not removed from the original structure during service. By inputting its prior parameter package and online monitoring data, the mechanical performance prediction results of the target seismic isolation bearing are obtained.

[0039] The difference between the preset seismic isolation bearing and the target seismic isolation bearing is that the preset seismic isolation bearing is not in service, but in a simulated service state of the target seismic isolation bearing, so as to facilitate the acquisition of its mechanical performance test data.

[0040] It should be noted that easily detachable bearings (i.e., the preset seismic isolation bearings) are selected as modeling samples to reduce the difficulty of obtaining model data and ensure the authenticity and effectiveness of model training data.

[0041] Preferably, the causal identification includes: using a pre-set set of prior parameters of the seismic isolation bearing and historical monitoring data during service, and employing a constraint-based algorithm or a scoring-based algorithm to determine the initial causal structure between variables.

[0042] More preferably, the initial causal structure includes: Reversible layer: ; In the formula, For the reversible component of the hysteresis characteristic, This is a reversible environmental nucleus, where T is temperature and RH is relative humidity. f For loading frequency, t This refers to the length of service.

[0043] Irreversible layer: ; In the formula, For irreversible deterioration of state variables, Let WDI be the state evolution function, and WDI be the water quality deterioration index. For time step, This refers to the perturbation term in the irreversible degradation process.

[0044] Output layer: ; In the formula, Y is the hysteresis characteristic quantity. h As the output mapping function, L can identify the load or input. It is cyclic energy.

[0045] Automatic Minimal Backdoor Set First, based on the causal Bayesian network structure of the seismic isolation bearing, the time sequence of variable acquisition, and mechanical mechanism constraints, a candidate set of adjustment variables C is constructed. This candidate set of adjustment variables C includes axial compression P, temperature T, relative humidity RH, water quality deterioration index (WDI), service time t, and cumulative cycle energy. And one or more of the abnormal event state variables; then, select the adjustment variable set Z from the candidate adjustment variable set C, so that the adjustment variable set Z does not contain descendant nodes of the identifiable load or input L, and block L to the net energy after removing the reversible environmental effects. Non-causal paths between variables; among multiple sets of adjustment variables that satisfy the backdoor criterion, select the set with the fewest variables and whose BIC scores meet preset conditions as the automatic minimum backdoor set. .

[0046] More preferably, the backdoor criterion includes: adjusting the variable set to not contain descendant nodes of the identifiable load or input L, and blocking the identifiable load or input L from reaching the net energy. The non-causal path between them, the net energy The performance quantity after removing the effects of reversible environmental factors.

[0047] After obtaining the automatic minimum backdoor set, the net energy of the identifiable load or input L is calculated using residual regression or backdoor adjustment integral formula. The causal effect is determined by the rolling window conditional independence test, which is used to test the stability of the candidate causal structure and the set of adjustment variables. The BIC score is used to rank the candidates among multiple sets, rather than as the sole criterion for the backdoor criterion.

[0048] It should be noted that the two-layer structural equation decouples reversible and irreversible effects, and the automatic minimum backdoor set effectively reduces the estimation variance, which can improve the accuracy of causal identification and the stability of model operation.

[0049] Preferably, the counterfactual normalization includes: Determine the set of environment variables E and benchmark environment E 0. Wherein, the set of environment variables... E This includes one or more of the following: temperature, humidity, loading frequency, water temperature, and water quality damage index. The reference environment... E 0 represents the stable environment during the initial service phase. A reversible environment core is fitted based on the results of multi-environment hysteresis tests conducted at the factory. And based on the currently observed hysteresis characteristics Calculate net observations :

[0050] The net observation measurement As observational evidence, it is used to update irreversibly deteriorated state variables. The net observation measurement This is equivalent to normalizing the current observation results to a baseline environment. E The counterfactual observation following 0, that is, the one corresponding to... The performance observation results are as follows.

[0051] It should be noted that counterfactual normalization can eliminate false alarms caused by environmental fluctuations, accurately identify the true deterioration state of the bearing, and significantly improve the reliability of monitoring results.

[0052] Preferably, the dynamic update of the Bayesian parameter learning includes: Using the factory test results of seismic isolation bearings as the prior distribution and real-time monitoring data during service as observational evidence, the conditional probability table (CPT) or regression coefficients between nodes are continuously updated through the Bayesian parameter learning method to achieve dynamic evolution estimation of bearing performance status. After obtaining new observation data, perform the following operations:

[0053] In the formula, J These are the mechanical performance parameters of the seismic isolation bearing. A set of environment variables. For the irreversible deterioration state variables of the seismic isolation bearing, For a given set of environment variables Time-irreversible degradation of state variables The posterior probability distribution, Given an irreversibly deteriorated state variable Mechanical performance parameters The conditional probability distribution; This means performing a marginal integration over all possible irreversibly deteriorating state variables to obtain a given set of environment variables. Lower mechanical property parameters The posterior predicted distribution.

[0054] It should be noted that dynamically updating model parameters to adapt to the performance changes of the support during long-term service can ensure the real-time performance and accuracy of model predictions.

[0055] Accordingly, this embodiment also provides a seismic isolation bearing full life cycle causal monitoring system, including: The seismic isolation bearing service environment information acquisition module is configured to collect monitoring data during the service life of the seismic isolation bearing and perform multimodal self-calibration, and increase the sampling rate when abnormal events are triggered; the monitoring data includes axial compression, displacement, temperature and humidity, and water quality; The causal reasoning model construction and update module is configured to construct a causal reasoning model of the mechanical performance evolution of the seismic isolation bearing based on the pre-set a priori parameter package and monitoring data of the seismic isolation bearing, and dynamically update the causal reasoning model through causal identification, counterfactual normalization and Bayesian parameter learning. The seismic isolation bearing performance change prediction module is configured to input the prior parameter package of the target seismic isolation bearing and the monitoring data into the causal inference model, and output the predicted results of the mechanical performance parameters of the target seismic isolation bearing.

[0056] It should be noted that the system operates in a modular manner throughout the entire process, which facilitates on-site deployment and maintenance, and enables automated and intelligent monitoring of the entire life cycle of the seismic isolation bearings.

[0057] Example 2 This embodiment provides a method for causal monitoring of seismic isolation bearings throughout their entire life cycle, which can be implemented using the following steps: 1. Bearing manufacturing information collection Record detailed information about the support to be monitored during the production process, such as steel plate quality, rubber formula, dimensions, and production process.

[0058] Specifically, according to relevant standards, such as "Rubber Bearings Part 1: Test Methods for Seismic Isolation Rubber Bearings", a 100% shear deformation loading test is conducted on the manufactured bearings to obtain the basic performance parameters of the bearings (such as the horizontal equivalent stiffness K). eq Yield force F y Post-yield stiffness K p and equivalent damping ratio ξ eq wait).

[0059] (1) Small sample orthogonal design experiment. Based on the standard shear test, hysteresis tests at multiple environmental points are added, such as temperature T∈{-10,0,10,20,30,40}°C, relative humidity RH∈{20%,40%,60%,80%}, and loading frequency. f ∈{0.1,0.3,0.5,0.7,0.9}Hz, forming a three-dimensional excitation matrix of temperature-humidity-frequency.

[0060] (2) Reversible / Irreversible kernel decomposition: for hysteretic eigenvalues ​​Y∈{K eq , ξ eq , F y , K p Fitting

[0061] In the formula, It is a reversible environmental core (changing instantly). It is an irreversibly degraded core (initially 0). To accumulate cycle energy, As cyclic energy, t This refers to the length of service.

[0062] (3) Hysteresis fingerprint and prior packet: The hysteresis fingerprint hash is constructed by sampling the force-displacement at multiple points and combined with the {reversible environment kernel}. Initial horizontal equivalent stiffness K eq0 Initial equivalent damping ratio ξ eq0 Initial yield force F y0 Stiffness K after initial yielding p0 The initial elastic modulus G0, initial friction coefficient μ0, initial bolt preload bolt0, and initial limit clearance lim0 are combined into a priori parameter package and bound to the support ID for causal reasoning priors during service.

[0063] 2. Support service environment record (1) Axial compression record. Since the seismic isolation bearings arranged under the water intake tower only face the constant load of the structure's self-weight, the axial compression of the bearings can be regarded as a certain value.

[0064] (2) Displacement recording. Two horizontal displacement gauges are installed on the seismic isolation bearing to record the relative displacement between the upper and lower connecting plates (record the complete displacement-time history curve). In addition, accelerometers are arranged to record the response curve of the bearing under seismic / vibration action (record when the acceleration exceeds the threshold).

[0065] (3) Record the humidity and temperature of the environment where the support is located in detail. When the support is equipped with an isolation device on the outside and is not in contact with water, record the humidity and temperature of the environment where the support is located in detail; when the support is submerged in water (i.e., there is no isolation device on the outside of the support), in addition to recording the water temperature, the properties of the water, such as dissolved oxygen (DO) and conductivity, should also be recorded. pH value and turbidity (NTU), etc.

[0066] (4) Water Degradation Index (WDI): A single index that normalizes water temperature and water quality.

[0067] In the formula, the weights Using historical / expert prior Bayesian updates, WDI enters the backdoor set of the causal model, and the quantity marked with "-" is the normalized risk coefficient (0-1).

[0068] (5) Event triggering high frequency window: When an earthquake, rapid thermocline change or displacement / acceleration exceeds the threshold occurs, the sampling rate is automatically increased to ≥200Hz, and event features such as hysteresis area, peak / mean square energy, and residual displacement are extracted and archived separately from normal data.

[0069] (6) Multimodal self-calibration: periodic solution Least squares regression, together with baseline Temperature compensation and sensor drift diagnosis are performed; exceeding limits triggers coefficient self-update and quality downweighting. Among these, The output quantity to be calibrated , and All are regression coefficients. As a baseline reference value, d For the amount of sensor drift, T For temperature.

[0070] 3. Establishment of a causal reasoning model 3.1 Overall Model Framework Based on the factory performance parameter package obtained in steps 1 and 2 and the multi-source monitoring data of the service environment, a causal reasoning model for the evolution of the bearing's mechanical performance is constructed. This model takes the environment-mechanical performance as the basic level and reflects the dynamic influence of external environmental conditions on the mechanical characteristics of the seismic isolation bearing.

[0071] Environmental variables: temperature, humidity, immersion depth, water temperature, water quality parameters (dissolved oxygen, conductivity, pH value, turbidity, etc.); Loading variables: axial compression, shear deformation, loading frequency, etc.; Mechanical performance variable: horizontal equivalent stiffness Yield force F, post-yield stiffness Equivalent damping ratio wait.

[0072] The causal reasoning model uses a directed acyclic graph to describe the causal relationship between variables, that is, the service environment affects the material state, which in turn leads to changes in the mechanical performance parameters of the support.

[0073] 3.2 Causal Identification and Structural Equation Establishment (1) Causal structure identification: Using the support factory data and historical monitoring data during service, the initial causal structure between variables is determined by using constraint-type algorithms (such as PC algorithm) or scoring-type algorithms (such as BDeu scoring); (2) Two-layer structure equation Reversible layer: ; In the formula, For the reversible component of the hysteresis characteristic, This is a reversible environmental nucleus, where T is temperature and RH is relative humidity. f For loading frequency, t This refers to the length of service.

[0074] Irreversible layer: ; In the formula, For irreversible deterioration of state variables, Let WDI be the state evolution function, and WDI be the water quality deterioration index. For time step, This refers to the perturbation term in the irreversible degradation process.

[0075] Output layer: ; In the formula, Y is the hysteresis characteristic quantity. h As the output mapping function, L can identify the load or input. It is cyclic energy.

[0076] Automatic Minimal Backdoor Set After the causal structure is identified, based on the physical meaning and temporal sequence of the variables, variables that occurred before or were jointly affected by upstream variables and can be identified as loads or inputs L are included in the candidate adjustment variable set C. The candidate adjustment variable set C includes axial pressure P, temperature T, relative humidity RH, water quality deterioration index (WDI), service time t, and cumulative cycle energy. And one or more of the abnormal event state variables. For any candidate set of adjustment variables Z, if Z does not contain descendant nodes of L, and can block L from interacting with the net energy after removing reversible environmental effects. If all non-causal paths between Z and L are pointed to by arrows, then Z is determined to satisfy the backdoor criterion.

[0077] Among all candidate adjustment variable sets that satisfy the backdoor criterion, the set with the fewest variables is selected to form the automatic minimum backdoor set. When multiple candidate sets have the same number of variables, the candidate sets are ranked using a rolling window conditional independence test and BIC score, and the set with the highest stability is selected. After obtaining the automatic minimum backdoor set, the causal effect of the identifiable load or input L on the net performance quantity is calculated using residual regression or the backdoor adjustment integral formula. The automatic minimum backdoor set is updated as data distribution drifts, abnormal events end, or monitoring conditions change.

[0078] 3.3 Counterfactual Normalization Determine the set of environment variables E and benchmark environment E 0. Wherein, the set of environment variables... E This includes one or more of the following: temperature, humidity, loading frequency, water temperature, and water quality damage index. The reference environment... E 0 represents the stable environment during the initial service phase. A reversible environment core is fitted based on the results of multi-environment hysteresis tests conducted at the factory. And based on the currently observed hysteresis characteristics Calculate net observations :

[0079] The net observation measurement As observational evidence, it is used to update irreversibly deteriorated state variables. The net observation measurement This is equivalent to normalizing the current observation results to a baseline environment. E The counterfactual observation following 0, that is, the one corresponding to... The performance observation results are as follows.

[0080] 3.4 Establishment of a Causal Inference Model Using a priori parameters and service environment information of a pre-defined seismic isolation bearing as input, and mechanical performance test data as output, a causal inference model is established based on a causal Bayesian network. It is worth noting that the pre-defined seismic isolation bearing refers to a reference bearing from the same batch, of the same specification, and simulating the same service environment as the target seismic isolation bearing (the monitoring object in the structure), serving as the basis for model construction. The pre-defined seismic isolation bearing undergoes 100% shear performance testing at a certain frequency to obtain its measured mechanical performance data. Thus, its prior parameters, service environment information, and mechanical performance data serve as the foundation for establishing the causal inference model (this causal inference model is a data-driven model).

[0081] The target seismic isolation bearing is not removed from the original structure during its service life. By inputting its prior parameter package and online monitoring data, the mechanical performance prediction results of the target seismic isolation bearing are obtained.

[0082] The difference between a preset seismic isolation bearing and a target seismic isolation bearing is that the preset seismic isolation bearing is not in service, but in a simulated service state of the target seismic isolation bearing, so as to facilitate the acquisition of its mechanical performance test data.

[0083] 3.5 Parameter Learning and Dynamic Update By using the results of the bearing factory test as the prior distribution and the real-time monitoring data during service as the observation evidence, the conditional probability table (CPT) or regression coefficients between nodes are continuously updated through the Bayesian parameter learning method to achieve dynamic evolution estimation of the bearing performance status.

[0084] It is worth noting that removing supports from a structure is a time-consuming and labor-intensive process. Therefore, it is recommended to use a subset of supports as the basis for dynamic model updates, and to take into account the ease of support removal during structural design. In this way, the causal inference model can be dynamically updated using factory information, real-time monitoring data, and changes in mechanical properties of a few supports. Subsequently, the updated causal inference model can be applied to predict the state of the remaining supports.

[0085] After obtaining new observation data, perform the following operations:

[0086] In the formula, J These are the mechanical performance parameters of the seismic isolation bearing. A set of environment variables. For the irreversible deterioration state variables of the seismic isolation bearing, For a given set of environment variables Time-irreversible degradation of state variables The posterior probability distribution, Given an irreversibly deteriorated state variable Mechanical performance parameters The conditional probability distribution; This means performing a marginal integration over all possible irreversibly deteriorating state variables to obtain a given set of environment variables. Lower mechanical property parameters The posterior predicted distribution.

[0087] 4. Prediction of bearing performance changes The a priori package of support parameters and historical monitoring data are input into the causal inference model to obtain the final output of the model. The model prediction results include estimates of mechanical properties such as horizontal equivalent stiffness, yield force, post-yield stiffness, and equivalent damping ratio, as well as their confidence levels.

[0088] 5. Causal effect assessment and counterfactual reasoning After the model has stabilized, counterfactual analysis is performed by applying do-operators. For example, reducing the water temperature predicts the rate of change of the horizontal equivalent stiffness; reducing the axial compression predicts the rate of change of the equivalent damping ratio. Performing counterfactual reasoning can provide quantitative decision-making basis for the optimal design of seismic isolation structures, bearing maintenance, and protection strategies.

[0089] Accordingly, this embodiment also provides a seismic isolation bearing full life cycle causal monitoring system, including: 1) Sensing and Interface Module. Receives factory information about the support (such as steel plate quality, rubber formula, dimensions, and manufacturing process), as well as service environment information (such as shear displacement, axial pressure, aging time, temperature, humidity, water quality damage index, acceleration, etc.).

[0090] 2) Data preprocessing module. Removes noise and zero drift issues from the monitoring data, and removes missing data portions to meet the requirements of subsequent processing.

[0091] 3) Condition Detection Module. Using the factory information and service environment information of a specific support as input, and the mechanical state as output, a causal inference model is established based on a causal Bayesian network. The factory information and service environment information of the support to be tested are input into the causal inference model to obtain the mechanical performance test results (net observation) of the support under specific conditions.

[0092] 4) Parameter learning and dynamic update module. In real-world environments, as monitoring data for specific supports becomes increasingly abundant, it is necessary to dynamically update the parameter weights in the causal inference model to improve its generalization performance and accuracy.

[0093] 5) Visualization and Warning Module. Displays the implementation prediction results of the bearing's mechanical performance and its confidence interval, determines the bearing's hazard level, and issues an alarm when the threshold is exceeded.

[0094] Example 3 This embodiment is based on embodiment 1: This embodiment provides a computer device, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the causal monitoring method for the entire life cycle of seismic isolation bearings as described in Embodiment 1. The computer program can be in the form of source code, object code, executable file, or some intermediate form.

[0095] Example 4 This embodiment is based on embodiment 1: This embodiment provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the causal monitoring method for the entire life cycle of seismic isolation bearings as described in Embodiment 1. The computer program can be in the form of source code, object code, executable file, or some intermediate form. The storage medium includes any entity or device capable of carrying computer program code, a recording medium, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc.

[0096] The above description is merely a preferred embodiment of the present invention. It should be understood that the present invention is not limited to the forms disclosed herein and should not be construed as excluding other embodiments. It can be used in various other combinations, modifications, and environments, and can be altered within the scope of the concept described herein through the above teachings or related technologies or knowledge. Modifications and variations made by those skilled in the art that do not depart from the spirit and scope of the present invention should be within the protection scope of the appended claims.

[0097] It should be noted that, for the sake of simplicity, the foregoing method embodiments are described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

Claims

1. A method for monitoring the causal relationship of seismic isolation bearings throughout their entire life cycle, characterized in that, include: Collect production and testing data of seismic isolation bearings to form a priori parameter package that is bound to the seismic isolation bearing ID; The system collects monitoring data of the seismic isolation bearings during their service life and performs multimodal self-calibration, increasing the sampling rate when abnormal events are triggered; the monitoring data includes axial compression, displacement, temperature and humidity, and water quality. Based on the pre-set parameter package of the seismic isolation bearing and the monitoring data, a causal reasoning model for the evolution of the mechanical performance of the seismic isolation bearing is constructed, and the causal reasoning model is dynamically updated through causal identification, counterfactual normalization and Bayesian parameter learning. The prior parameter package of the target seismic isolation bearing and the monitoring data are input into the causal inference model, and the predicted mechanical performance parameters of the target seismic isolation bearing are output.

2. The method for monitoring the causal relationship of a seismic isolation bearing throughout its entire life cycle according to claim 1, characterized in that, The process of collecting factory production and testing data for seismic isolation bearings to form a priori parameter package bound to the seismic isolation bearing ID includes: Small-sample orthogonal design experiment: Based on the standard shear test, hysteresis tests at multiple environmental points are added, including temperature. T relative humidity RH and loading frequency f This forms a multidimensional excitation matrix; Reversible / Irreversible Kernel Decomposition: Decomposing the hysteretic characteristic Y into reversible environmental kernel and irreversible degradation kernel: In the formula, As a reversible environmental core, It is an irreversibly degraded core. t For service time, To accumulate cycle energy, It is cyclic energy; Hysteresis fingerprint and prior parameter package generation: Multi-point force-displacement samples are used to form a hysteresis fingerprint hash, which is then combined with the initial performance parameters to form a prior parameter package and bound to the seismic isolation bearing ID.

3. The method for monitoring the causal relationship of a seismic isolation bearing throughout its entire life cycle according to claim 2, characterized in that, The initial performance parameters include: reversible environment core. Initial horizontal equivalent stiffness K eq0 Initial equivalent damping ratio ξ eq0 Initial yield force F y0 Initial post-yield stiffness K p0 Initial elastic modulus G0, initial friction coefficient μ0, initial bolt preload bolt0, and initial limit gap lim0.

4. The method for monitoring the causal relationship of a seismic isolation bearing throughout its entire life cycle according to claim 1, characterized in that, The process of collecting monitoring data during the service life of seismic isolation bearings and performing multimodal self-calibration includes: Axial compression acquisition: Force sensors are arranged on the seismic isolation bearing, bearing connecting plate or supporting component to acquire the axial compression on the seismic isolation bearing in real time or periodically, and obtain axial compression time history data; Displacement acquisition: Two horizontal displacement gauges are installed on the seismic isolation bearing to record the relative displacement between the upper and lower connecting plates; accelerometers are arranged to record the response curve of the seismic isolation bearing under earthquake or vibration. Temperature and humidity data collection: When the seismic isolation bearing is equipped with an isolation device on its outer side, i.e., it is not in contact with water, the humidity and temperature of the environment in which the seismic isolation bearing is located are recorded; when the seismic isolation bearing is not equipped with an isolation device on its outer side, i.e., it is submerged in water, the water temperature T and water quality parameters are recorded, including electrical conductivity. Dissolved oxygen (DO), pH value, and turbidity (NTU); Water Quality Degradation Index (WDI) calculation: The WDI is calculated by normalizing water temperature and water quality. In the formula, For weights updated based on Bayesian methods, the values ​​marked with "—" are the normalized risk coefficients; Multimodal self-calibration: periodic solution Least squares regression, combined with reversible environment kernel Temperature compensation and sensor drift diagnosis are performed; exceeding limits triggers coefficient self-update and quality downweighting. The output quantity to be calibrated , and All are regression coefficients. As a baseline reference value, d For the amount of sensor drift, T For temperature.

5. The method for monitoring the causal relationship of a seismic isolation bearing throughout its entire life cycle according to claim 1, characterized in that, The increase in sampling rate when an abnormal event is triggered includes: When an earthquake, rapid thermocline change, or displacement / acceleration exceeds a preset threshold occurs, the sampling rate is automatically increased and event features are extracted and archived separately from normal data; the event features include hysteresis area, peak / mean square energy, and residual displacement.

6. The method for monitoring the causal relationship of a seismic isolation bearing throughout its entire life cycle according to claim 1, characterized in that, The causal reasoning model for the evolution of the mechanical performance of seismic isolation bearings, based on the pre-set parameter package and monitoring data of the seismic isolation bearings, includes: Using the pre-set parameter package of the seismic isolation bearing and the service monitoring data as inputs, and the measured mechanical performance data as outputs, a causal reasoning model for the evolution of the mechanical performance of the seismic isolation bearing is established based on a causal Bayesian network. The preset seismic isolation bearing refers to a reference bearing that is from the same batch, has the same specifications, and is in a similar service environment as the target seismic isolation bearing, and serves as the basis for model construction; the reference bearing is subjected to 100% shear performance testing at a set frequency to obtain its measured mechanical performance data; The target seismic isolation bearing is the monitoring object and is not removed from the original structure during service. By inputting its prior parameter package and online monitoring data, the mechanical performance prediction results of the target seismic isolation bearing are obtained.

7. The method for monitoring the causal relationship of a seismic isolation bearing throughout its entire life cycle according to claim 1, characterized in that, The causal identification includes: using a pre-set set of prior parameters for the seismic isolation bearings and historical monitoring data during service, and employing a constraint-based algorithm or a scoring-based algorithm to determine the initial causal structure between variables; The initial causal structure includes: Reversible layer: ; In the formula, For the reversible component of the hysteresis characteristic, This is a reversible environmental nucleus, where T is temperature and RH is relative humidity. f For loading frequency, t Service life; Irreversible layer: ; In the formula, For irreversible deterioration of state variables, Let WDI be the state evolution function, and WDI be the water quality deterioration index. For time step, This refers to the perturbation term during the irreversible degradation process; Output layer: ; In the formula, Y is the hysteresis characteristic quantity. h As the output mapping function, L can identify the load or input. It is cyclic energy; Automatic Minimum Backdoor Set: Based on the preset cause-effect graph, variable time sequence constraints, and seismic isolation bearing mechanical mechanism constraints, the set of adjustment variables that satisfies the backdoor criterion is selected from the candidate set of adjustment variables. Then, the set with the smallest size and statistical stability that meets the preset conditions is selected from the set of adjustment variables that satisfies the backdoor criterion as the automatic minimum backdoor set. Subsequently, based on the automatic minimum backdoor set The causal effect of the load or input L on the net performance quantity can be calculated by residual regression or backdoor adjustment integral formula; the automatic minimum backdoor set is updated according to data distribution drift, abnormal event termination or monitoring condition changes; The backdoor criteria include: adjusting the variable set to not contain descendant nodes of the identifiable load or input L, and blocking the identifiable load or input L from reaching net energy. The non-causal path between them, the net energy The performance quantity after removing the effects of reversible environmental factors.

8. The method for monitoring the causal relationship of a seismic isolation bearing throughout its entire life cycle according to claim 7, characterized in that, The counterfactual normalization includes: Determine the set of environment variables E and benchmark environment E 0, the set of environment variables E This includes one or more of the following: temperature, humidity, loading frequency, water temperature, and water quality damage index; the reference environment. E 0 represents the stable environment during the initial service phase; a reversible environment core is fitted based on the results of multi-environment hysteresis tests conducted at the factory. And based on the currently observed hysteresis characteristics Calculate net observations : The net observation measurement As observational evidence, it is used to update irreversibly deteriorated state variables. The net observation measurement This is equivalent to normalizing the current observation results to a baseline environment. E The counterfactual observation following 0, that is, the one corresponding to... The performance observation results are as follows.

9. The method for monitoring the causal relationship of a seismic isolation bearing throughout its entire life cycle according to claim 1, characterized in that, The dynamic update of the Bayesian parameter learning includes: Using the production and testing data of seismic isolation bearings as the prior distribution and the real-time monitoring data during service as the observation evidence, the conditional probability table CPT or regression coefficients between nodes are continuously updated through the Bayesian parameter learning method to achieve dynamic evolution estimation of bearing performance status. After obtaining new observation data, perform the following operations: In the formula, J These are the mechanical performance parameters of the seismic isolation bearing. A set of environment variables. For the irreversible deterioration state variables of the seismic isolation bearing, For a given set of environment variables Time-irreversible degradation of state variables The posterior probability distribution, Given an irreversibly deteriorated state variable Mechanical performance parameters The conditional probability distribution; This means performing a marginal integration over all possible irreversibly deteriorating state variables to obtain a given set of environment variables. Lower mechanical property parameters The posterior predicted distribution.

10. A causal monitoring system for the entire life cycle of seismic isolation bearings, characterized in that, include: The seismic isolation bearing factory information acquisition module is configured to collect seismic isolation bearing factory production and test data to form a priori parameter package bound to the seismic isolation bearing ID; The seismic isolation bearing service environment information acquisition module is configured to collect monitoring data during the service life of the seismic isolation bearing and perform multimodal self-calibration, and increase the sampling rate when abnormal events are triggered; the monitoring data includes axial compression, displacement, temperature and humidity, and water quality; The causal reasoning model construction and update module is configured to construct a causal reasoning model of the mechanical performance evolution of the seismic isolation bearing based on the pre-set a priori parameter package and monitoring data of the seismic isolation bearing, and dynamically update the causal reasoning model through causal identification, counterfactual normalization and Bayesian parameter learning. The seismic isolation bearing performance change prediction module is configured to input the prior parameter package of the target seismic isolation bearing and the monitoring data into the causal inference model, and output the predicted results of the mechanical performance parameters of the target seismic isolation bearing.

Images