53 results about "Probability assessment" patented technology

The invention discloses an underground-ground-sky combined rock burst and earth surface deformation early warning method, which comprises the following steps of: S1, arranging an underground monitoring system, and acquiring underground monitoring data; s2, setting a ground monitoring system, and collecting ground monitoring data; s3, setting a sky monitoring system, and collecting sky monitoring data; s4, calibrating the time scale difference of the underground monitoring data, the ground monitoring data and the sky monitoring data through a multi-scale time alignment algorithm; and S5, constructing a multi-stage joint probability assessment framework to carry out mining area rock burst and earth surface deformation risk assessment. According to the method, multi-source monitoring data are integrated through a multi-scale space-time fusion framework, and a full-chain association model of'microcosmic fracture-mesoscopic deformation-macroscopic accumulation 'is constructed. And on the basis of cross-scale time sequence alignment, spatial heterogeneity is optimized and eliminated, a multi-stage joint probability assessment framework quantifies a cross-scale collaborative risk, full-period intelligent management and control of'monitoring-early warning-feedback-optimization 'is realized, and early warning accuracy and timeliness are improved.

The invention discloses a weather big data-based rainstorm and flood composite disaster early warning method, which comprises the following steps of: extracting periodic characteristics and abnormal fluctuation information of rainstorm frequency by acquiring long-term weather data and performing hierarchical processing, and analyzing and determining the change trend of the rainstorm frequency in combination with a time sequence. And constructing an incidence matrix based on the rainstorm frequency change trend and the topographic hydrological characteristic data, determining an initial parameter range of flood probability assessment, introducing dynamic correlation analysis, fusing meteorological data and environmental conditions, and obtaining a dynamic adjustment coefficient of probability assessment. Through mapping the adjustment coefficient and the rainstorm frequency change trend, a quantitative expression of a rainstorm and flood coupling structure is constructed, and an early warning model is obtained based on the quantitative expression. And finally generating multi-scene early warning signal distribution by performing adaptive adjustment on the model response and verifying the accuracy of the model response. According to the invention, accurate early warning of flood disasters in a complex environment is realized, and the timeliness and reliability of early warning are significantly improved.

The invention relates to the field of low-altitude navigation, and particularly provides a low-altitude route conflict and dynamic early warning data processing method and system, a terminal and a medium, and the method comprises the steps: constructing a comprehensive state vector fusing task roles and formation constraints; spatial domain situation structured perception is realized through dynamic spatial domain gridding modeling and spatial domain state attribute updating; performing grid-level macroscopic conflict identification based on airspace state attributes, and screening potential conflict hotspot regions; for aircrafts in the hot spot area, trajectory prediction of intention and dynamics fusion is carried out based on the comprehensive state vector, and conflict risk probability assessment is carried out by adopting space-time probability envelope; and when the conflicting parties belong to the same group, a collaborative collision avoidance strategy taking formation maintenance as a primary target is generated, the strategy preferentially tries to integrally and uniformly avoid, and if the strategy fails, differential instruction distribution based on task role priority is started. According to the method, the high efficiency and the real-time performance of conflict early warning and releasing are improved.

The present disclosure provides a method and system for quickly screening cascading failures that consider causal relationships and timing characteristics, and relates to the technical field of cascading failures in power grids. The method comprises the following steps: obtaining cascading failure scenario data; extracting the causal relationship between an initial short-circuit fault and subsequent faults using logistic regression and a neural network model based on the cascading failure scenario data; converting different states of a fault event into a graph structure, and obtaining time delays between different states of the fault event based on statistics of the cascading failure scenario data, and extracting timing characteristics during the transient evolution of the cascading failure event; obtaining the probability of an initial fault triggering a subsequent fault event based on the causal relationship, and generating a cascading failure scenario using a Monte Carlo sampling method based on the timing characteristics; and performing risk assessment and probability assessment on the cascading failure scenarios generated based on the causal relationship and timing characteristics, and ultimately screening out high-risk cascading failure scenarios.

The invention discloses a network packet capture and analysis method and device based on artificial intelligence. And constructing a dynamically updated normal traffic baseline by using a lightweight AI model, and extracting real-time traffic features to perform abnormal probability evaluation. And when the abnormal probability exceeds a risk threshold, triggering a precise packet capture strategy. Extracting multi-modal features from the captured abnormal traffic, and fusing the multi-modal features into a high-dimensional feature vector which retains behavior semantics through a deep neural network; and carrying out anomaly classification by using a special AI large model, and carrying out causal reasoning in combination with a knowledge graph to locate a root cause. And generating a disposal strategy based on the root cause, and performing risk grading according to the influence range and the service interruption probability to realize grading closed-loop disposal. According to the method, full-process automation from anomaly perception, intelligent analysis to adaptive disposal is realized, the traditional technical bottlenecks of encrypted traffic detection, unknown threat identification, root cause positioning and the like are effectively solved, and the efficiency and response speed of network security operation and maintenance are remarkably improved.

The invention relates to a slope instability probability evaluation method based on burial depth constraint space random field simulation, and belongs to the technical field of slope engineering. The method comprises the following steps: acquiring shear strength parameters c and phi at different buried depths of a slope soil body in a research area; the obtained shear strength parameters are expanded according to Gaussian process regression in combination with a radial basis function, the reliability of the expanded parameters is determined through K-S inspection, and cross correlation between c and phi is obtained; determining statistical characteristics of the c and phi expansion data; under the burial depth constraint condition, slope model determinacy research is carried out, a safety coefficient is calculated, the model is compared with a traditional model, and the difference between the potential slip plane of the slope under the burial depth constraint and the potential slip plane of the traditional slope is obtained; the traditional Monte Carlo simulation MCS is optimized; according to deterministic analysis and a random field theory, carrying out uncertainty research on the side slope under the burial depth constraint, and calculating the instability probability of the depth evolution random field side slope model.

The invention discloses a multispectral remote sensing disaster monitoring and early warning system using deep learning, and the system comprises a spectral data collection and preprocessing module which is used for continuously recording image data in different time periods based on surface reflection information to obtain an original data set, preprocessing the original data set to obtain a clear image data set; the mode recognition module is used for carrying out mode classification on the clear image data set to obtain a potential disaster related feature combination; the disaster risk assessment module is used for performing time sequence analysis on the potential disaster related feature combination to obtain a probability assessment value associated with the disaster risk; the disaster risk early warning module is used for constructing a quantitative index system associated with the disaster risk based on the probability evaluation value and determining an early warning reference standard based on the constructed quantitative index system associated with the disaster risk; and the closed-loop monitoring module is used for continuously monitoring the multispectral data acquired in real time based on an early warning reference standard to generate early warning information containing risk levels and outputting the early warning information.

The embodiment of the invention discloses a regional water shortage risk and space-time imbalance evaluation method, system and device and a storage medium, and the method and system construct a multi-dimensional water shortage risk evaluation system through the combination of kernel density estimation, a Copula function and a centralized index, and remarkably improve the evaluation precision and practicability. Compared with a traditional parametric distribution method, kernel density estimation does not need to preset a data distribution form, the method can flexibly adapt to complex and diversified supply and demand sequences, high fitting precision is still kept under the condition of insufficient data or extreme distribution, experiments show that the root-mean-square error is reduced by more than 30% compared with the traditional method, and the adaptability to unknown distribution is remarkably enhanced. The joint probability distribution of the water demand and the water supply is quantified through the Copula function, and the severity level is dynamically mapped in combination with the water shortage rate and the duration, so that the limitation of single probability evaluation is solved, the time persistence is brought into risk calculation for the first time, and the evaluation result is closer to the actual water use scene.

The invention relates to the technical field of power systems, and particularly discloses a data-driven power transmission safety grading risk assessment method, which comprises the following steps: constructing an intelligent real-time assessment model of power transmission section limit transmission capability based on a KAN framework; historical operation data and simulation supplementary data are collected to construct a training sample set; training an intelligent real-time evaluation model of the transmission section limit transmission capability through a loss function comprising a TTC evaluation precision loss item and a physical consistency regularization item, and ensuring that the evaluated TTC meets section power flow safety constraints; performing multi-confidence scene sampling on wind power and photovoltaic output to generate a new energy uncertain output scene set under different confidence levels; and inputting the new energy output scene set into a trained intelligent real-time evaluation model of the transmission section limit transmission capability, obtaining TTC probability evaluation sets corresponding to different confidence coefficients, and realizing transmission safety grading risk evaluation. The method has the advantage that intelligent and real-time risk confidence assessment of the operation mode can be realized.

The invention discloses a distributed photovoltaic aggregation frequency modulation capability probability evaluation method considering distribution network operation safety constraints, and the method comprises the steps: firstly carrying out the modeling of the probability distribution of distributed photovoltaic power generation power through employing a multivariate Gaussian mixture model, so as to describe the non-Gaussian characteristics and spatial correlation of a plurality of distributed photovoltaic outputs; secondly, taking maximization of distributed photovoltaic aggregation upward and downward power regulation ranges as a target, considering power distribution network line power flow constraints, node voltage safety constraints and inverter capacity constraints, and constructing a second-order cone programming model to realize distributed photovoltaic aggregation frequency modulation capability deterministic evaluation; and finally, proposing a multivariate mixture Gaussian point estimation algorithm, constructing a special sampling set, executing a small amount of deterministic calculation, and combining statistical analysis and weighted mixing to obtain the probability distribution of the distributed photovoltaic aggregation frequency modulation reserve capacity. According to the method, the large-scale distributed photovoltaic polymerization frequency modulation capability is effectively analyzed, the uncertainty is quantified, and the calculation precision and efficiency are both considered.

SYSTEM AND METHOD FOR EVALUATING UNIFORM RESOURCE LOCATOR 5 Disclosed is a system (100) and a method (500) for evaluating a Uniform Resource Locator (URL). The method (500) comprises extracting the URL from a message and assigning an initial risk score. The method (500) further comprises performing a protocol verification to verify a protocol of the URL and a domain assessment operation to determine whether the URL is an IP-address based URL, or FQDN 10 based URL. Furthermore, the method (500) comprises calculating a risk score for the URL based on protocol verification, domain assessment, evaluation of length of the URL, association with reputed domains, analyzing mean value of the URL, and identifying whether an entry corresponding to the URL is stored in a distributed ledger (108). Thereafter, the method (500) comprises determining a probability of 15 the URL being secure based on the calculated risk score and evaluating the credibility of the URL based on the probability.

The present invention relates to a stress intensity limit S t A failure probability assessment method is provided, comprising: obtaining creep rupture test data of specimens under different working conditions, including creep rupture life and creep strain-time curves; determining the standard error of the logarithmic creep rupture life; determining the average stress required for the total strain to reach 1%, the minimum stress at the onset of the third stage of creep, and the minimum stress causing creep rupture under each working condition, as the first criterion stress, the second criterion stress, and the third criterion stress, respectively; determining creep life models corresponding to the first, second, and third criterion stresses; determining creep life models considering safety factors corresponding to the first, second, and third criterion stresses; determining probability parameters corresponding to the first, second, and third criterion stresses based on the creep life model after introducing probability parameters and the creep life model considering safety factors for each criterion; and calculating the failure probabilities of the first, second, and third criterion stresses under different working conditions.

The invention belongs to the technical field of oil and gas drilling, and particularly relates to a casing tripping-in risk quantitative evaluation method considering well bore quality uncertainty, which realizes probability evaluation of casing tripping-in friction resistance by utilizing Monte Carlo simulation and multi-distribution random sampling. The core thought of the method is as follows: the uncertainty of well diameter and dog-leg-degree well bore quality parameters is taken as an entry point, and the accuracy and decision scientificity of casing tripping-in performance evaluation are improved through the process of'probability modeling-related sampling-smooth constraint-friction resistance calculation-risk evaluation '. According to the method, uncertainty and spatial distribution characteristics of key well bore quality parameters such as the well diameter and the dogleg degree can be systematically represented; a large number of random borehole quality parameter sequences conforming to actual geology and engineering rules are generated; on the basis, Monte Carlo simulation is carried out, and probability distribution of casing running friction resistance is calculated; and finally, the risk that the casing can be put down is quantified in a probability form, and a visual probability basis is provided for engineering decision making.

The invention relates to the field of power grid data scheduling, and discloses a power grid business data intelligent scheduling method based on multi-source data fusion, which comprises the following steps: fusing different business data of a power grid for risk assessment and fault probability assessment of the power grid and user loads, and combining a fault probability assessment result to obtain a power grid business data intelligent scheduling result; and carrying out load intelligent scheduling strategy construction and fault correction strategy construction on the user load and the power grid in the high-risk state. The system can accurately and actively identify the operation risk of the power grid, achieves the intelligent load transfer of main and distribution cooperation, can prevent faults in advance, achieves the quick response when the faults occur, effectively reduces the power-off time and range, guarantees the normal power demands of all walks of life and residents, and improves the power utilization efficiency. And meanwhile, unnecessary equipment overload and redundant operation can be avoided, the operation cost of the power grid is reduced, the service life of equipment is prolonged, and the operation economic benefit of the power grid is improved on the whole.

The invention discloses an active power distribution network bearing capacity probability modeling method related to virtual power plant demand response, and aims to solve the problems that in the prior art, evaluation of photovoltaic bearing capacity depends on certainty, and an uncertainty processing method and an efficient probability evaluation framework are lacked. According to the method, uncertainty is quantified through a probability density function, a complex random problem is converted into a multi-scene deterministic optimization problem containing key uncertainty information by utilizing Monte Carlo simulation and scene reduction technologies, and demand response of a virtual power plant is integrated as an active control means. And finally, accurately calculating the photovoltaic bearing capacity and the system performance index of the active power distribution network under the condition of meeting multiple security constraints.

The present application relates to railway route selection technical field, and relates to a kind of railway route selection scheme across seismic fault's total probability risk assessment method.The assessment method uses Monte Carlo simulation, establishes the total probability risk analysis model suitable for railway route selection design across fault zone, is based on the theory of performance-based earthquake engineering, integrates fault zone displacement probability risk analysis, railway structure object probability vulnerability analysis and railway fault rear probability economic loss analysis, can be combined with fault zone characteristics (position, geometric shape, seismic intensity) The annual exceedance probability of specific displacement value is calculated, the damage state of different railway structure (usually avoid using bridge mode to cross fault zone, therefore, the present application mainly considers the intersection of tunnel, embankment section and fault zone) under the influence of fault displacement is evaluated, simultaneously, total probability risk analysis model across fault zone and cost minimization model form cost-risk net present value model, and construction cost and cross-fault zone risk are simultaneously evaluated.

This invention discloses a method and system for optimizing medical consignment inventory based on kit integrity constraints. The method includes: constructing a material dependency graph for surgical kits, representing each surgical kit as a graph structure containing nodes and edges; constructing a shortage risk propagation model based on historical borrowing and returning data, calculating the risk transmission coefficient of material shortages on kit availability, and forming a kit integrity probability assessment system; establishing an optimization model aimed at minimizing the probability of kit shortages under total safety stock budget constraints, dynamically allocating the safety stock level of each material within the kit; integrating surgical scheduling data with the inventory calculation model, and implementing differentiated inventory reservation and sharing strategies based on surgical certainty levels to achieve synergistic optimization of kit integrity and inventory costs. This effectively solves the problem that traditional single-material inventory calculation methods cannot guarantee kit integrity, significantly improving material availability and reducing redundant inventory costs.

The invention provides a power grid forest fire potential intensity probability assessment method based on comparative learning, and belongs to the technical field of power grid safety assessment. Comprising the steps of 1, defining parameters; 2, collecting input data and processing the input data; 3, classifying the processed data into dynamic and static variables; 4, building a model; 5, initial data are sent to the step 2, results are sequentially output to the next step until the step 4, and network training is carried out; 6, after training is completed, learning network training results are compared; and step 7, obtaining a power grid mountain fire potential probability evaluation result after training is completed. According to the model, input and target features are aligned to a unified high-dimensional hidden space by using comparative learning in a pre-training stage, uncertainty of potential fire intensity is described by using a multi-quantile head in a regression stage, and a constructed two-stage learning framework is also suitable for other power grid forest fire behavior feature probability modeling and is high in universality and expandability.

The invention belongs to the technical field of power system scheduling and protection, and discloses a non-homogeneous Poisson model-based power distribution system recloser event probability evaluation and rapid standby scheduling method, which comprises the following steps of: constructing a recloser event structure analysis model, and realizing event probability evaluation of a disturbance event through non-homogeneous Poisson process modeling; in combination with system power imbalance and reserve capacity analysis, a coupling relation between disturbance influence and reserve capacity is established; and introducing a conditional risk value mechanism to construct positive and negative symmetric risk constraints, integrating the positive and negative symmetric risk constraints into a joint scheduling optimization model, and performing hourly rolling optimization in combination with income maximization to form a risk-aware scheduling decision. The method can be widely applied to a virtual power plant and other power distribution systems containing distributed resources, effectively improves the risk identification, probability evaluation and standby resource scheduling capabilities of the power distribution system in response to short-time disturbance, and improves the economy, toughness and risk control capability of the system in a multi-disturbance uncertainty scene.

This invention discloses a multispectral remote sensing disaster monitoring and early warning system utilizing deep learning. The system includes: a spectral data acquisition and preprocessing module, used to continuously record image data from different time periods based on surface reflectance information to obtain a raw dataset, and preprocess the raw dataset to obtain a clear image dataset; a pattern recognition module, used to classify patterns in the clear image dataset to obtain potential disaster-related feature combinations; a disaster risk assessment module, used to perform time-series analysis on potential disaster-related feature combinations to obtain probability assessment values ​​associated with disaster risk; a disaster risk early warning module, which constructs a quantitative index system for disaster risk association based on the probability assessment values, and determines early warning reference standards based on the constructed quantitative index system for disaster risk association; and a closed-loop monitoring module, which continuously monitors the real-time acquired multispectral data based on the early warning reference standards to generate early warning information output including risk levels.

The invention relates to the technical field of geological disaster monitoring and early warning, and discloses a geological disaster probability assessment method based on multi-source big data fusion analysis. The method comprises the steps of three-dimensional coordinate system layout and multi-point sensor acquisition, standardization, memory kernel construction, time sequence convolution fusion, anomaly recognition, probability mapping and spatial interpolation, and high-risk area output. By unifying a spatial reference system, the spatial consistency of data of each monitoring point is ensured; the whole-process data quality inspection and standardization are realized, and the dimension unification and comparability of the multi-source heterogeneous data are realized. A finite memory weight model and time sequence weighted convolution are innovatively adopted, the historical cumulative effect is comprehensively reflected, the anomaly discrimination robustness is improved, and the defect that only short-term data is used is overcome. Probabilistic mapping and feature distance-based spatial smoothing endow the risk expression with continuity and adaptivity, and realize scientific determination of regional high-risk partitions.

The application discloses a global sensitivity calculation method of a power system risk event based on subset simulation, establishes a representation method of a risk event in power system uncertainty quantification considering uncertainty of a random source model, applies subset simulation to decompose a rare event to reduce sampling times, and efficiently calculates global sensitivity of uncertainty of probability distribution parameters of a random source in a power system to a probability of a rare event. The scheme greatly reduces a sample size while guaranteeing accuracy of rare event probability evaluation, effectively solves the problem of low efficiency of a traditional Monte Carlo sampling algorithm in simulating a rare event, and can quickly evaluate the influence degree of uncertainty of probability distribution parameters of a random source in a power system to a probability of a rare event of power grid operation.

This invention relates to the field of PCB board technology, and particularly to a method and system for assessing the defect probability of PCB boards. By acquiring multiple feature data and constructing a three-dimensional defect feature map, this invention effectively assesses not only surface defects but also internal structural defects and latent defects. Real-time acquired environmental interference parameters are used to correct the basic assessment factors, ensuring that the defect probability assessment considers not only the material's inherent characteristics but also the impact of environmental factors on the test results. By individually assessing each preset detection area, the defect situation and its impact in different areas can be clearly identified, providing a precise basis for subsequent production adjustments and defect rectification. The calculation method based on single-area defect probability and area ratio ensures the authority and accuracy of the final overall defect probability, providing solid data support for PCB quality inspection.

The invention discloses a nuclear power plant hydrogen risk analysis method and system, electronic equipment and a storage medium, and relates to the technical field of nuclear power plant safety analysis. Comprising the steps of establishing a key parameter probability distribution model of a nuclear power plant hydrogen originating accident; generating an initial condition combination sample set of the key parameters based on a probability sampling method, and constructing a thermal hydraulic calculation working condition matrix; executing high-fidelity simulation of a full-sample working condition, and generating a containment multi-compartment spatio-temporal evolution data set; training a time sequence neural network model by using the data set to realize mapping of an accident process and multi-physical field parameters; and inputting a real-time accident type and a monitoring initial value, driving the model to output hydrogen concentration prediction, and synchronously completing dynamic evaluation of the AICC risk and the DDT risk. According to the invention, through efficient calculation and data analysis means, rapid probability evaluation of diffusion, aggregation and explosion consequences of hydrogen in the containment is realized, and the risk research and judgment speed and decision reliability under severe accidents of the nuclear power plant are significantly improved.

This invention discloses a method for assessing the overall performance degradation probability of box girder bridges based on DBN and Gamma distributions, belonging to the technical field of box girder bridge performance degradation probability assessment. The method includes: determining the composition of the box girder bridge; determining the dwell time of each component; calculating relevant parameters of the Gamma distribution probability density function; calculating the cumulative probability density function and cumulative survival probability function of each component; modeling the performance degradation of the box girder bridge to obtain the transition probability; calculating the state probability distribution of each component; forming the probability degradation curves of each component and the overall structure through a dynamic Bayesian network; and assessing the impact of each component on the probability degradation curves of its component and the overall structure. This invention, employing the aforementioned method for assessing the overall performance degradation probability of box girder bridges based on DBN and Gamma distributions, can scientifically predict the performance degradation of box girder bridges from component to component and from component to the whole, thereby predicting the service life of box girder bridges from a probabilistic perspective.

The invention discloses a data-driven micro-grid planning scheme probability evaluation method and system. The method comprises the following steps: acquiring a historical database; training based on a historical database to obtain a generative deep neural network model; based on the generative deep neural network model, generating a scene data set including load data, wind power generation data and photovoltaic power generation data; based on a preset to-be-evaluated micro-grid planning scheme and the scene data set, micro-grid operation simulation is carried out through a preset micro-grid simulation model, and an analogue simulation result is obtained; based on the analogue simulation result, pre-defined evaluation indexes are calculated, and an evaluation index sample set is formed. The system comprises a data acquisition module, a scene generation module, a scene data set generation module, a micro-grid operation simulation module and a performance index probability evaluation and analysis module. According to the method, the result is more comprehensive and credible, and the tail risk can be effectively captured; moreover, the method is very high in adaptability to an uncertain environment.

This invention relates to the field of precipitation forecasting technology, and discloses a precipitation forecasting and probabilistic prediction method, system, equipment, and medium based on physical consistency constraints. The method includes: in the training phase, inputting preprocessed first meteorological data into the model to obtain preliminary prediction results, and constructing physical consistency constraints by combining the physical fields in the results; constructing classification prediction losses for precipitation magnitude intervals; performing joint optimization by combining various losses to obtain a trained precipitation forecasting model; inputting preprocessed second meteorological data into the trained model, and outputting continuous prediction results of future precipitation intensity through a physical enhancement feature construction module, a multi-scale circulation feature fusion module, a spatiotemporal feature fusion module, and a feature reconstruction and prediction module. This invention can improve the stability and generalization ability of the model, and provide direct and usable magnitude event probability information for precipitation forecast results, thereby meeting the needs for precipitation risk classification and early warning and extreme event probability assessment.

This invention relates to the field of transaction data management technology, and in particular to a method and system for transaction data anomaly analysis and processing based on feature correlation dimensionality reduction. This method determines the multi-dimensional original features of transaction data, performs dimensionality reduction based on feature correlation to obtain target features, and calculates the initial anomaly probability by combining historical normal feature deviations to screen out suspected abnormal transactions. It then uses the dimensionality-reduced target features to extract behavioral data from suspected transactions, determines the behavioral anomaly probability and specific abnormal transactions by the deviation from historical behavioral data, and comprehensively assesses the overall transaction risk of the platform based on the initial and behavioral anomaly probabilities. If the risk does not meet the requirements, the entire dataset undergoes anomaly review, and the transaction risk label status is updated accordingly. This method removes redundant noise through feature dimensionality reduction and employs a two-stage probability assessment to achieve precise identification from coarse to fine, effectively reducing the false alarm rate and improving the efficiency and accuracy of anomaly detection.

This application discloses an integrated assessment method and device for multivariate seismic vulnerability of tunnels that incorporates physical priors, relating to the technical field of seismic analysis and seismic risk probability assessment of underground engineering structures. The method includes: generating predicted samples based on a logarithmic domain joint distribution description; inputting each predicted sample into a heteroscedastic moment estimation substitution model to obtain predicted values ​​of the conditional mean and conditional variance of the logarithmic domain demand response; reconstructing and dynamically anchoring the conditional marginal distributions of each response based on the predicted values ​​and residual information, and coupling each anchored conditional marginal distribution to obtain a multi-response joint probability model; calculating the point-state failure probability of each projected sample under a preset damage state based on the multi-response joint probability model, constructing discrete sample pairs; and fitting each discrete sample pair to obtain seismic vulnerability curves under each preset damage state. This application improves the accuracy, robustness, and engineering usability of tunnel seismic vulnerability assessment.