Early warning and intelligent monitoring system for full-life multi-disaster coupling damage of fragmented prefabricated mixed tower

Abstract

The invention discloses a fragmented prefabricated mixed tower full-life multi-disaster coupling damage early warning and intelligent monitoring system, and relates to the technical field of structure health monitoring and damage assessment, and the system comprises a data collection module which obtains the stress data and environmental difference performance data of each fragmented prefabricated mixed tower under complex conditions, carries out the comprehensive processing of the stress distribution and environmental response, and carries out the early warning and intelligent monitoring of the whole-life multi-disaster coupling damage of each fragmented prefabricated mixed tower. Obtaining initial damage state distribution of each fragment; the damage transfer modeling module is used for constructing a damage transfer network diagram according to the initial damage state distribution and aiming at the contact surface action and the connection part influence between the fragments, and determining a potential damage transfer path; according to the full-life multi-disaster coupling damage early warning and intelligent monitoring system for the fragmented prefabricated mixed tower, the damage state of a prefabricated structure in a complex environment is effectively predicted, and an important basis is provided for structural safety assessment and maintenance decision making.

Description

Technical Field

[0001] This invention relates to the field of structural health monitoring and damage assessment technology, specifically to a prefabricated multi-hazard coupled damage early warning and intelligent monitoring system for the entire life cycle of segmented prefabricated mixed towers. Background Technology

[0002] In the field of modern engineering, the research and application of precast structures occupy a crucial position, especially when facing the challenges of complex environments and multiple disasters. Their stability and durability are directly related to the safety and service life of the project. Precast structures are widely used in various large-scale engineering projects due to their high efficiency and energy saving characteristics. However, how to accurately assess their damage state under complex conditions has become a core issue in ensuring project safety.

[0003] However, current methods for damage assessment of precast structures still have significant shortcomings. Many existing solutions rely too heavily on macroscopic data of the overall structure, neglecting the subtle interactions and damage transmission mechanisms between components. When faced with multiple coupled hazards, this approach struggles to capture how localized damage gradually affects overall performance, leading to significant discrepancies between assessment results and actual conditions, particularly in predictive capabilities under dynamic environments. Focusing on specific challenges, damage transmission between segments of a precast structure has become a critical issue that urgently needs to be addressed. Due to differences in the performance of each segment under stress, material properties, and environmental influences, damage is often not isolated but spreads to adjacent segments through contact surfaces or connections. This diffusion effect not only exacerbates localized damage but may also pose potential stability risks to the overall structure. At a deeper level, the complexity of this transmission relationship lies in the lack of a unified quantitative method to describe the damage state and its interactions, making it difficult to accurately determine the cumulative path and ultimate impact range of damage under multiple hazard conditions. Summary of the Invention

[0004] The purpose of this invention is to provide a multi-hazard coupled damage early warning and intelligent monitoring system for the entire life cycle of segmented prefabricated mixed towers, which expresses the damage state of each segment in a precise mathematical form and reveals the laws of damage transmission and mutual influence between different segments through reasonable calculation methods.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a segmented prefabricated hybrid tower full-life-cycle multi-hazard coupled damage early warning and intelligent monitoring system, comprising a data acquisition module, which acquires stress data and environmental difference performance data of each segment under complex conditions, and comprehensively processes the stress distribution and environmental response to obtain the initial damage state distribution of each segment; a damage transmission modeling module, which constructs a damage transmission network diagram based on the initial damage state distribution and the influence of contact surfaces and connection points between segments to determine potential damage transmission paths; a feature extraction module, which extracts features based on the potential damage transmission paths and the influence of diffusion trends and multi-hazard coupling to obtain the direct impact weight of local damage on adjacent segments; and an evolution analysis module, which constructs an inter-segment damage model based on the direct impact weight of local damage on adjacent segments. The system employs a dynamic evolution model for damage transmission to determine the phased distribution characteristics of accumulated damage; a risk prediction module, if the phased distribution characteristics of accumulated damage exceed a preset threshold, simulates the damage transmission and diffusion trend to obtain a preliminary risk distribution for overall performance prediction; a key node identification module, based on the preliminary risk distribution of overall performance prediction, tracks the evolution of risk distribution for the coupling effect of multiple disasters to determine the key time nodes for quantifying damage status; an assessment calculation module, using data from key time nodes, constructs a multi-dimensional index system for quantifying damage status to obtain comprehensive damage assessment results for prefabricated structures under different disaster scenarios; and a feedback correction module, based on the comprehensive damage assessment results, constructs a feedback correction mechanism to obtain an optimized distribution of damage assessment accuracy through iterative updates of the multi-dimensional index system.

[0006] Preferably, the data acquisition module acquires stress data and environmental difference performance data of each segment under complex conditions, comprehensively processes the stress distribution and environmental response to obtain the initial damage state distribution of each segment. This includes constructing a segment model of the prefabricated structure, acquiring stress data and environmental difference performance data of each segment under complex conditions, using data acquisition tools to initially organize multi-source data, obtaining the original stress distribution and environmental response records of each segment; based on the original stress distribution and environmental response records, using fusion technology to comprehensively process the multi-source data, performing weighted analysis on the stress data and environmental difference performance data to determine the stress distribution characteristics and environmental response patterns of each segment; through the analysis of stress distribution characteristics and environmental response patterns, obtaining the stress concentration areas and environmentally sensitive points of each segment under complex conditions, and using a support vector machine algorithm for key areas. The system is categorized to determine potential initial damage locations. If the classification results show that the stress concentration area of ​​a segment exceeds a preset threshold, a deep comparison of the environmental response pattern of that segment is performed to obtain its correlation coefficient with the environment, thus determining the range of initial damage. Based on the range of initial damage, secondary verification is performed on the stress distribution characteristics and environmental response patterns. Anomalies in the multi-source data are filtered using the information entropy calculation method to obtain preliminary results of the damage state distribution of each segment. By integrating the preliminary results of the damage state distribution, comprehensive performance data of each segment under complex conditions is obtained. Data backtracking is performed on anomalies and potential damage locations to determine the final initial damage state distribution. Based on the final initial damage state distribution, data visualization tools are used to dynamically map the stress distribution and environmental response of each segment, obtaining a panoramic view of the damage distribution of each segment under complex conditions.

[0007] Preferably, the damage transmission modeling module, based on the initial damage state distribution and considering the effects of contact surfaces and connections between segments, constructs a damage transmission network diagram to determine potential damage transmission paths. This includes classifying and organizing the damage distribution data between segments using data integration tools based on the initial damage and state distribution, obtaining damage distribution characteristic data for each segment, and identifying key differences in the distribution characteristics; analyzing the force characteristics of contact and connection points between segments based on the damage distribution characteristic data, obtaining mechanical action data for contact surfaces and connection areas, and determining the distribution range of the effects; constructing a damage transmission network model based on the distribution range of the effects, and digitally mapping the interactions between segments using graph analysis to obtain a network graph of interaction relationships; and based on the interaction relationships... The network graph is analyzed to identify potential damage transmission paths. Path tracing technology is used to simulate the transmission direction and determine the main nodes and branches of potential paths. For the main nodes and branches of potential paths, the role and influence of each node in the transmission process are analyzed. If the transmission intensity between nodes exceeds a preset threshold, the path is prioritized to obtain priority transmission paths. Based on the priority transmission paths, combined with state distribution and piecewise contact data, data comparison is performed on key areas along the paths to obtain the dynamic changes in damage transmission along the paths and determine the risk areas in the transmission process. For the risk areas in the transmission process, the support vector machine algorithm is used to classify the damage distribution characteristics within the area to obtain the risk level distribution after classification and determine the key monitoring scope of the risk areas.

[0008] Preferably, the feature extraction module extracts features based on the potential damage propagation path and the coupling effect of the diffusion trend and multiple disasters. This yields the direct impact weights of local damage on adjacent segments, including potential paths propagated through the damage. Data acquisition tools are used to monitor the diffusion process of local damage in real time, acquiring dynamic feature data to determine the initial range and direction of the diffusion. Based on the initial range and direction of the diffusion, time series analysis is used to process the dynamic feature data, identifying the fluctuation patterns and key time points of the diffusion trend. Based on these fluctuation patterns and key time points, the coupling effects of multiple disasters are analyzed to obtain damage aggravation characteristics under the coupling effects, and to determine the superposition effect of damage under different disasters. By analyzing the cumulative effects of damage under different disasters, and considering the impact on adjacent segments, a data comparison tool is used to perform a stratified analysis of damage distribution, revealing the direct impact on adjacent segments. Based on the direct impact on adjacent segments, a weight allocation model is constructed to determine the impact weight value of each segment in damage transmission, identifying key areas for weight distribution. Using these key areas, a logical mapping method is employed to digitize the impact paths, identifying the priority ranking of damage transmission, based on the correlation between local damage and adjacent segments. Finally, based on the priority ranking of damage transmission, information integration tools are used to classify and organize path data for key nodes along paths from high to low priority, determining the distribution of risk areas along the paths.

[0009] Preferably, the evolutionary analysis module constructs a dynamic evolutionary model of damage transmission between segments based on the direct impact weights of local damage on adjacent segments. This model determines the phased distribution characteristics of damage accumulation, including the correlation between local damage and adjacent segments. Data acquisition tools are used to monitor damage distribution data in real time to obtain preliminary impact data of local damage on adjacent segments and determine the initial boundaries of the impact range. Based on the initial boundary data, a preliminary framework for weight allocation is constructed to determine the impact weight distribution of each segment in damage transmission. Finally, based on the impact weight distribution and relevant information from the accumulation path, a path tracing tool is used to analyze the potential directions of damage transmission and determine the transmission dynamics. The key nodes are distributed in the data; based on the key node distribution, an evolutionary model framework is constructed to address the dynamic changes in transmission, and evolutionary data of damage accumulation at different time periods are obtained; for the evolutionary data, the stage characteristics of damage accumulation are analyzed, and if the change amplitude of stage characteristics exceeds a preset threshold, the data is stratified to determine the significant difference regions of stage distribution; based on the significant difference regions, information integration tools are used to classify and organize the data for the characteristic changes of stage distribution, and detailed classification results of distribution characteristics are obtained; based on the classification results of distribution characteristics, combined with the logical mapping of feature judgment, the distribution characteristics of damage accumulation are digitized, and the priority ranking of the final distribution characteristics at different stages is determined.

[0010] Preferably, the risk prediction module, if the phased distribution characteristics of damage accumulation exceed a preset threshold range, simulates the damage transmission and diffusion trend to obtain the preliminary risk distribution for overall performance prediction. This includes continuously monitoring the phased distribution of damage accumulation using a dynamic evolution model, and obtaining preliminary simulation data for potential directions when the characteristics exceed the threshold. Based on the preliminary simulation data, a path tracing tool is used to analyze the changing characteristics of potential directions and determine key areas of risk distribution. Based on the distribution of key areas and combined with the overall performance prediction data, a multi-stage analysis framework for damage transmission is constructed to obtain dynamic change information of phased distribution. If the dynamic change information of phased distribution shows characteristics exceeding a preset threshold, the change information is processed hierarchically using an information integration tool to determine the priority areas of damage accumulation. Based on the distribution characteristics of priority areas and combined with the simulation results of the dynamic model, a data mapping technique is used to perform secondary calibration of the potential directions of transmission and diffusion to obtain calibrated risk distribution data. Based on the calibrated risk distribution data and the overall performance prediction data, a multi-dimensional analysis matrix is ​​constructed to determine the distribution weights of damage accumulation at different stages. Based on the analysis results of the distribution weights and combined with the output data of the evolution simulation, the phased distribution of damage accumulation is continuously updated to obtain the final prediction data adjustment scheme.

[0011] Preferably, the key node identification module predicts the initial risk distribution based on overall performance, tracks the evolution of the risk distribution in response to the coupling effect of multiple disasters, and determines the key time nodes for quantifying the damage state. This includes continuously monitoring the distribution pattern of the initial risk using time series analysis technology to obtain the changing characteristics of the distribution pattern in different time periods and determine the initial evolution trend. Based on the initial evolution trend, combined with the coupling effect of complex conditions and multiple disasters, a pre-established disaster impact model is used to decompose the evolution trend in multiple dimensions to obtain risk fluctuation data under the coupling effect. For the risk fluctuation data, if the fluctuation amplitude exceeds a preset threshold range, the fluctuation data is processed in layers using information integration tools to determine the potential impact of multiple disasters on the damage state. Based on the potential impact of the damage state, and combined with the correlation between quantification nodes and key times, data mapping technology is used to locate the impact on the time axis and obtain the distribution characteristics of key time nodes. For the distribution characteristics of key time nodes, and combining the correlation between evolution patterns and distribution patterns, the distribution characteristics are dynamically updated using data comparison tools to determine the priority ranking of evolution patterns at different stages. Based on the priority ranking results, and considering the correlation between overall performance and damage state, regression analysis technology is used to perform secondary calibration of the ranking results to obtain calibrated risk distribution weights. For the calibrated risk distribution weights, and combining the correlation between key times and quantification nodes, information fusion tools are used to comprehensively process the weight data to determine the quantification results of the damage state of overall performance at different time nodes.

[0012] Preferably, the assessment calculation module constructs a multi-dimensional indicator system for quantifying damage status using data from key time nodes, and obtains comprehensive damage assessment results for prefabricated structures under different disaster scenarios. This includes classifying information at key time and node points using data integration tools, obtaining the characteristic distribution of the classified time nodes, and determining the variation pattern of the characteristic distribution in different time periods; based on the variation pattern of the characteristic distribution, combined with segmented transmission and dynamic characteristics, using a pre-established feature mapping model to perform hierarchical analysis of the variation pattern, obtaining hierarchical dynamic characteristic data; and for the hierarchical dynamic characteristic data, combining damage status and quantitative indicators, using information fusion tools to perform multi-dimensional comparison of the dynamic characteristic data, and determining the damage status in different dimensions. The distribution is quantified; if the quantified distribution exceeds a preset threshold, the distribution is structured using data filtering tools, taking into account the multidimensional system and prefabricated structure, to obtain the structured damage distribution characteristics; based on the structured damage distribution characteristics, combined with the disaster scenario and comprehensive assessment, logistic regression analysis is used to match the feature data to the scenario, determining the matched scenario damage assessment results; for the matched scenario damage assessment results, combined with damage data and status analysis, the assessment results are deeply analyzed through an information integration platform to obtain the final quantitative distribution of damage status; based on the final quantitative distribution of damage status, combined with key time and node data, a data update tool is used to dynamically adjust the quantified distribution, judging the adaptability of the adjusted distribution characteristics at the time nodes.

[0013] Preferably, the feedback correction module constructs a feedback correction mechanism based on the comprehensive damage assessment results. Through iterative updates of the multi-dimensional indicator system, it obtains an optimized distribution of damage assessment accuracy. This includes: using the comprehensive assessment results and data integration tools to perform preliminary classification of the damage results, obtaining the classified damage distribution characteristics; based on the classified damage distribution characteristics, combined with the feedback mechanism, using a pre-established mapping model to perform deviation analysis on the distribution characteristics, determining the deviation distribution range; for the deviation distribution range, combining overall performance and prediction deviation, if the deviation range exceeds a preset threshold, then using information comparison tools to perform multi-dimensional decomposition of the deviation data, obtaining the decomposed deviation characteristics; based on the decomposed deviation characteristics, combined with multi-dimensional indicators and system iteration, using data update tools to adjust the deviation characteristics layer by layer, obtaining the adjusted indicator distribution.

[0014] Preferably, the feedback correction module constructs a feedback correction mechanism based on the comprehensive damage assessment results. Through iterative updates of the multi-dimensional indicator system, the optimized distribution of damage assessment accuracy is obtained. This includes, for the adjusted indicator distribution, combining the update operation and assessment accuracy, performing a deep comparison of the distribution data through an information fusion platform to determine the accuracy distribution after comparison; based on the accuracy distribution after comparison, combined with the optimized distribution and performance analysis, if the accuracy distribution does not meet the preset standard, the distribution is structured using a data filtering tool to obtain the structured optimized distribution characteristics; for the structured optimized distribution characteristics, logistic regression analysis is used to finally calibrate the feature data to determine the calibrated comprehensive performance distribution.

[0015] As can be seen from the above technical solution, the present invention has the following beneficial effects:

[0016] This segmented precast hybrid tower full-life-cycle multi-hazard coupled damage early warning and intelligent monitoring system acquires stress and environmental data by constructing a segmented model and uses multi-source data fusion technology to obtain the initial damage state distribution. Based on this distribution, a damage transmission network diagram is constructed to determine potential transmission paths. Combining local damage diffusion characteristics and the coupled effects of multiple hazards, a dynamic evolution model is established to determine damage accumulation characteristics. When a threshold is exceeded, the system simulates further diffusion trends to obtain the overall performance risk distribution. For the coupled effects of multiple hazards under complex conditions, the system tracks the evolution of risk distribution and determines key time nodes. Finally, a multi-dimensional index system is constructed to achieve comprehensive damage assessment under different disaster scenarios, and the assessment accuracy is optimized through a feedback correction mechanism. This invention can effectively predict the damage state of precast structures in complex environments, providing an important basis for structural safety assessment and maintenance decisions. Attached Figure Description

[0017] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] like Figure 1As shown, this invention provides a technical solution: a prefabricated multi-tower full-life-cycle multi-hazard coupled damage early warning and intelligent monitoring system, including a data acquisition module to acquire stress data and environmental difference performance data of each segment under complex conditions, and to comprehensively process the stress distribution and environmental response to obtain the initial damage state distribution of each segment; a damage transmission modeling module to construct a damage transmission network diagram based on the initial damage state distribution, considering the contact surface effect and connection part influence between segments, and to determine the potential damage transmission path; a feature extraction module to extract features based on the potential damage transmission path, considering the influence of diffusion trend and multi-hazard coupling, and to obtain the direct impact weight of local damage on adjacent segments; and an evolution analysis module to construct the damage transmission between segments based on the direct impact weight of local damage on adjacent segments. The system employs a dynamic evolution model to determine the phased distribution characteristics of accumulated damage; a risk prediction module, if the phased distribution characteristics of accumulated damage exceed a preset threshold range, simulates the trend of damage transmission and diffusion to obtain a preliminary risk distribution for overall performance prediction; a key node identification module, based on the preliminary risk distribution of overall performance prediction, tracks the evolution of risk distribution for the coupling effect of multiple disasters to determine the key time nodes for quantifying damage status; an evaluation calculation module, using data from key time nodes, constructs a multi-dimensional index system for quantifying damage status to obtain comprehensive damage assessment results for prefabricated structures under different disaster scenarios; and a feedback correction module, based on the comprehensive damage assessment results, constructs a feedback correction mechanism to obtain an optimized distribution of damage assessment accuracy through iterative updates of the multi-dimensional index system.

[0020] This system uses a segmented prefabricated hybrid tower structure as its object. A data acquisition module acquires real-time mechanical and environmental response data for each segment under various external stresses and environmental factors. The raw data is normalized and feature-fused to form an initial damage state distribution map for each segment. A damage transmission modeling module constructs a damage transmission network graph including nodes (segments) and edges (contact relationships). A feature extraction module extracts key features from the network paths. An evolutionary analysis module constructs a dynamic evolution model based on time-series data and influence weights. A risk prediction module simulates diffusion trends and provides an overall risk distribution. A key node identification module identifies sensitive points in risk changes. An assessment calculation module constructs a multi-dimensional index system to quantify the degree of damage in each dimension, forming a comprehensive damage assessment under disaster scenarios. Finally, a feedback correction module iteratively updates the index weights through error feedback between historical assessment results and new observation data.

[0021] The data acquisition module obtains stress data and environmental difference data for each segment under complex conditions. It comprehensively processes the stress distribution and environmental response to obtain the initial damage state distribution of each segment. This includes constructing a segmented model of the prefabricated structure, acquiring stress data and environmental difference data for each segment under complex conditions, and using data acquisition tools to initially process the multi-source data to obtain the original stress distribution and environmental response records for each segment. Based on the original stress distribution and environmental response records, a fusion technique is used to comprehensively process the multi-source data, performing weighted analysis on the stress data and environmental difference data to determine the stress distribution characteristics and environmental response patterns of each segment. Through the analysis of stress distribution characteristics and environmental response patterns, stress concentration areas and environmentally sensitive points in each segment under complex conditions are identified. A support vector machine algorithm is used to segment key areas. The system classifies and identifies potential initial damage locations. If the classification results show that the stress concentration area of ​​a segment exceeds a preset threshold, a deep comparison of the environmental response pattern of that segment is performed to obtain its correlation coefficient with the environmental differences, thus determining the range of initial damage. Based on the range of initial damage, secondary verification is performed on the stress distribution characteristics and environmental response patterns. Anomalies in the multi-source data are filtered using the information entropy calculation method to obtain preliminary results of the damage state distribution of each segment. By integrating the preliminary results of the damage state distribution, comprehensive performance data of each segment under complex conditions is obtained. Data backtracking is performed on anomalies and potential damage locations to determine the final initial damage state distribution. Based on the final initial damage state distribution, data visualization tools are used to dynamically map the stress distribution and environmental response of each segment, obtaining a panoramic view of the damage distribution of each segment under complex conditions.

[0022] In the implementation process, the design or supervision unit first provides the building information model (BIM) files and construction drawings for the segmented precast concrete tower structure, clearly defining the geometry, material type (e.g., C50 high-performance concrete), installation direction, and connection points of each segment. Next, a full-size three-dimensional structural model is created using finite element method (FE) software, dividing the structure into segments proportionally. To improve analysis accuracy, each segment is divided into at least 10 stress calculation sub-units, with strain gauges and temperature / humidity sensors placed at the center of each sub-unit. The sensor placement scheme is determined through structural mechanics analysis, ensuring that strain gauges are placed in the area of ​​maximum stress and its vicinity, while temperature / humidity sensors are placed on the outer surface and internal nodes. Data acquisition officially begins after the structure reaches a stable state following construction, with all sensors uploading information in real time through the data acquisition module. The sampling frequency is set to 100 times per second, meaning 100 sets of strain values ​​and environmental data are recorded per second. The collected raw data first enters the preliminary processing flow, including time synchronization (ensuring consistent recording time from all sensors), outlier removal (e.g., removing data points whose changes within 1 second exceed 5 times the average of the previous 10 seconds), numerical normalization (converting strain values ​​to units of microstrain), and data integrity checks. Each data point must have at least 98% integrity; otherwise, the corresponding segment is marked as missing data. After preliminary processing, all strain and environmental data are categorized by segment number and then enter the weighted fusion calculation stage. Stress characteristic calculations include three indicators: maximum stress value, average stress value, and stress standard deviation. The maximum stress is the converted stress corresponding to the point with the highest value among all strain gauges in that segment; the average stress is the average value of all points within a 60-second time window; and the stress standard deviation is the standard deviation of the stress difference among all points under this average value. Environmental characteristic calculations include temperature change amplitude, humidity change rate, and response hysteresis time. The temperature change range is the difference between the highest and lowest temperatures within 10 minutes; the humidity change rate is the change in humidity within 1 minute divided by 60 seconds; and the response lag time is the time interval between a sudden change in strain data and an abrupt change in the environment, measured in seconds. These six indicators are normalized before being incorporated into a weighted model. The weights are determined by expert scoring, where organizational structure engineers, safety engineers, and data analysts rate the impact of each indicator from 1 to 10, and the average score is normalized to obtain the weight coefficients. In this example, the total weight for stress characteristics is 0.6, and the total weight for environmental characteristics is 0.4. The weighted composite response value for each segment is the sum of the normalized six indicators multiplied by their corresponding weights, resulting in a response index between 0 and 1.

[0023] If the response index of a certain segment exceeds 0.75, the system marks that area as a high-interest area and proceeds to the stress concentration and environmental anomaly analysis stage. Within this area, if the stress increase at each strain gauge point exceeds 50% of its average value over the past 60 seconds within 10 consecutive seconds, it is marked as a stress concentration point. Environmental anomaly points are identified as those where temperature fluctuates by more than 5 degrees Celsius within 1 minute or humidity changes by more than 10% within 5 minutes. If the number of stress concentration points in a segment exceeds 20% of the total number of strain gauges, the segment is marked as a "preliminary suspected damage area." The system then calls a pre-trained support vector machine classification model to classify the area. The model input consists of the aforementioned six indicators plus spatial location, segment number, and construction number. The model output is "potential damage" or "no damage." If the classification result is potential damage, the system proceeds to the deep environmental correlation analysis stage, which extracts the environmental response time series of the segment over the past 24 hours and performs Pearson correlation analysis with adjacent segments. If the correlation coefficient is less than 0.3, indicating a significant difference in environmental response, it is judged as a risk of localized damage caused by environmental differences. At this point, the information entropy anomaly screening step begins. Within each segment, using a 1-minute time window, all strain data are grouped by 0.1 microstrain values ​​to determine statistical frequencies. These frequencies are used as the basis for probability estimation, calculating the probability distribution entropy value. For example, the range of -50 to +50 microstrain is divided into 1000 intervals. The frequency of data occurrences within each interval is then counted and divided by the total number of data points within that time window to obtain the probability value for that interval. A negative logarithmic operation is performed on all non-zero probability values, and the result is multiplied by the corresponding probability value. Finally, the summation is applied to all intervals to obtain the entropy value for that segment within that time window. If the average entropy value of a segment exceeds 30% of the average entropy value of other segments on the same floor, it is identified as a high-anomaly segment. Here, "30%" is an empirical threshold obtained by analyzing engineering data from 50 historical projects, which can accurately identify abnormal fluctuations exceeding the normal range at a 99% confidence level. Then, historical data from these high-anomaly areas are retrospectively analyzed, focusing on data points from the past 60 minutes. If stress values ​​fail to decrease for more than 10 consecutive minutes while environmental responses fluctuate synchronously in the same direction, the area is identified as a "continuously evolving damage area." Finally, the comprehensive response values, classification results, information entropy values, and retrospective analysis results of all identified damaged areas are combined to generate the initial damage status level for that area, categorized as "slight," "moderate," or "severe." The classification is based on a response value exceeding 0.6 for slight, exceeding 0.75 for moderate, and exceeding 0.9 for severe. All results are input into a visualization module to generate a 3D view for each area. The 3D view uses blue for no anomalies, orange for moderate damage, and red for severe damage. The system updates the view every 5 minutes and automatically generates a diagnostic report, listing specific indicator values, anomaly point numbers, damage levels, and recommended treatment methods for each area.

[0024] The damage transmission modeling module, based on the initial damage state distribution and considering the effects of contact surfaces and connections between segments, constructs a damage transmission network diagram to determine potential damage transmission paths. This includes classifying and organizing the damage distribution data between segments using data integration tools, obtaining damage distribution characteristic data for each segment, and identifying key differences in the distribution characteristics. Based on the damage distribution characteristic data between segments, it analyzes the force characteristics of contact and connection points between segments, obtains mechanical action data for contact surfaces and connection areas, and determines the distribution range of the effects. For the distribution range of the effects, it constructs a damage transmission network model, and uses graph analysis to digitally map the interactions between segments, obtaining a network graph of interaction relationships. The system analyzes potential damage transmission paths using a damage mapping technique. Path tracing is employed to simulate the transmission direction and identify key nodes and branches of these paths. For each key node and branch, the influence of its role in the transmission process is analyzed. If the transmission intensity between nodes exceeds a preset threshold, the path is prioritized to identify priority transmission paths. Based on these priority paths, and combined with state distribution and piecewise contact data, data comparison is performed on key areas along the paths to obtain dynamic characteristics of damage transmission and identify risk areas. For these risk areas, a support vector machine algorithm is used to classify the damage distribution characteristics within the areas, obtaining the classified risk level distribution and determining the key monitoring areas for these risk regions.

[0025] In the damage transmission modeling module, the system first reads the initial damage state distribution data generated by the data acquisition module. This data includes the number, coordinates, corresponding damage level value, stress change curve, environmental response data, and segment connection information for each segment. The system represents the damage level of each segment using a value between 0 and 1, where 0 represents no damage and 1 represents severe damage. The data integration tool categorizes and organizes the damage distribution among the segments, pairs each pair of adjacent segments according to their connection relationships, and extracts the damage level difference, damage growth rate difference, and stress fluctuation similarity at time synchronization points between each pair. The system saves this data for each pair of segments as a "damage relationship pair." Subsequently, the system analyzes the structural design drawings and sensor deployment data, identifies all existing segment contact surfaces and connection points, extracts the strain gauge locations and numbers within these areas, and performs mechanical action calculations based on the stress data collected by the sensors. Specifically, on each connection surface, all strain sensor data within the area of ​​that surface are selected, and the average normal stress and average shear stress within that area are calculated, in megapascals (MPA). The average normal stress is calculated using the normal strain data of all sensing points within the region, combined with the material's elastic modulus; the average shear stress is calculated using the tangential strain data at each point. The system performs sliding calculations on these data every minute, recording the fluctuation amplitude and average growth rate over the past 30 minutes. If the difference between the maximum and minimum normal stress values ​​at a connection point exceeds 15% of the historical average normal stress at that point within 30 minutes, or if the shear stress fluctuation exceeds 20% of the maximum historical fluctuation value at that point, then the connection surface is considered to have high influence. The aforementioned percentage thresholds are derived from historical data of the engineering project, ensuring effective differentiation of abnormal transmission paths at a 95% confidence level. Based on the aforementioned processing results, the system establishes network graph nodes for each segment and edges for each pair of adjacent segments. The weight of the edge is equal to the stress fluctuation rate corresponding to the connection surface. The weight value is calculated as follows: the sum of the maximum and minimum values ​​of the normal and shear stresses on the connection surface within the past 10 minutes is subtracted from the sum of the minimum values, divided by the 10-minute average stress level, resulting in a dimensionless value representing the stress instability of the connection surface. The system employs graph analysis to construct a complete structural graph. A breadth-first search algorithm traverses all possible paths, accumulating edge weights for each path and selecting the path with the highest accumulated weight as the potential damage propagation path. For each potential damage propagation path, the system sequentially analyzes each node (segment unit) on the path to identify its influence during the propagation process. Influence is calculated by comparing the damage level change rate of the node with that of its neighboring nodes and calculating its contribution to the damage level increase of downstream nodes. If a node contributes more than 0.5 to the damage level change rate of downstream nodes within 30 minutes (i.e., after the node's change, the average damage level increase of downstream nodes accounts for more than 50% of its own increase), then the node is considered a highly propagating node.When a path contains three or more nodes with high transmissibility, it is marked as a priority path. Within the priority path, the system performs time-series comparisons of the damage status data for each node segment, analyzing the correlation between damage indicators, stress changes, and environmental responses. If the stress waveforms of upstream and downstream nodes in the same path show synchronous peaks within a 5-minute window, and the time difference between the peaks is less than 10 seconds, a dynamic transmissibility relationship is identified. The system then marks this segment as a "dynamic transmissibility risk area" and initiates support vector machine (SVM) model analysis. In the classification analysis, the system constructs a feature vector for each risk area segment, including the aforementioned edge weights, node damage level, volatility, number of historical stress peaks, average shear stress level, environmental response amplitude, and similarity to damage growth with neighboring nodes. The model has been trained using historical project data, outputting classification results as "low risk," "medium risk," or "high risk," corresponding to damage risk level ranges of 0 to 0.3, 0.3 to 0.7, and above 0.7, respectively. The thresholds are derived through statistical distribution classification analysis based on the damage development patterns of 10 typical engineering monitoring projects. Finally, the system outputs the paths of all high-risk nodes and their classification results, updating the transmission network diagram for subsequent evolutionary analysis and early warning. This process is automatically repeated every 10 minutes to ensure the system reflects the evolution and transmission trends of structural damage in real time.

[0026] The feature extraction module extracts features based on the potential damage propagation path and the impact of the coupling between the diffusion trend and multiple disasters. It obtains the direct impact weights of local damage on adjacent areas, including potential paths propagated through damage. Data acquisition tools are used to monitor the diffusion process of local damage in real time, acquiring dynamic feature data to determine the initial range and direction of diffusion. Based on the initial range and direction of diffusion, time series analysis is used to process the dynamic feature data, identifying the fluctuation patterns and key time points of the diffusion trend. Based on these fluctuation patterns and key time points, the coupling effects of multiple disasters are analyzed to obtain damage aggravation characteristics under the coupling effects, determining the superposition effect of damage under different disasters. To assess the cumulative effects of different disasters, a data comparison tool was used to perform a stratified analysis of damage distribution, considering the impact on adjacent segments, to determine the direct impact on each segment. Based on the direct impact on adjacent segments, a weight allocation model was constructed to determine the impact weight value of each segment in damage transmission, identifying key areas for weight distribution. Using these key areas, a logical mapping method was employed to digitize the impact paths, identifying the priority ranking of damage transmission, based on the correlation between local damage and adjacent segments. Finally, based on the priority ranking of damage transmission, information integration tools were used to classify and organize the path data for key nodes along the high-to-low priority paths, determining the distribution of risk areas along the paths.

[0027] In the feature extraction module, the system first extracts the numbers of all segments along the aforementioned potential damage transmission path and identifies the starting segment as the "local damage source." The system then initiates a high-frequency data acquisition mode at all sensing points within the damage source segment and its three adjacent segments, acquiring 100 data items per second, including strain values, temperature, humidity, and displacement. Each data set is accompanied by a timestamp and sensor number and stored in a database in real time. For each sensing point, the system calculates its strain change rate over a continuous 60 seconds, which is the average difference between the current time and the data from the previous second, and compares this rate with the historical average change rate over the past 30 minutes. If the average change rate over the current 60 seconds exceeds five times its historical average, and this phenomenon occurs continuously for more than 10 seconds, the sensing point is marked as a "local diffusion front point." The system uses the coordinate set of all "diffusion front points" as input and fits their spatial distribution area using the least bounded ellipse method. The major axis of the fitted ellipse is taken as the primary direction of current damage propagation, and the area inside the ellipse represents the initial propagation range. Next, the system dynamically monitors all segments within the propagation range, extracting the strain change sequence of all strain sensing points within each segment over the past 5 minutes and constructing a time series. A sliding window method is used to analyze the fluctuation trend. The window size is set to 30 seconds, with a step size of 10 seconds. Within each window, the system calculates the mean and standard deviation, and the ratio of the standard deviation to the mean is used as the volatility. If the volatility of three consecutive windows is greater than 1.5 times the average historical volatility of the segment over 30 minutes, the point is considered to be in a state of heightened volatility. If the number of points in this state exceeds 30% of the total number of sensing points in the segment and appears in three or more segments, the system marks the current time point as a "critical time node for propagation trend." Within the critical time node, the system compares this time period with the local disaster database, extracting meteorological records, seismic fluctuation records, and records of sudden temperature changes within the structure. Each type of disaster factor has a set threshold: wind speed exceeding 10 meters per second, temperature change rate exceeding 2 degrees Celsius per minute, and seismic acceleration exceeding 0.05g. If two of the above three conditions simultaneously meet the threshold within the specified time period, and the time difference between the start of data fluctuation and the start of structural strain surge is less than 60 seconds, the system determines that the node has a disaster coupling effect. Subsequently, the system retrieves the strain peak value of the damage source segment under the coupled state and calculates the ratio with the historical 30-minute average value of the same location under the disaster-free state. If the ratio exceeds 2, it is determined that there is a damage aggravation phenomenon. For all adjacent segments on the diffusion path, the system further compares their damage diffusion trends with those of the damage source segment. The system extracts the strain time series of each pair of segments, performs dynamic time warping, and calculates the similarity value. The similarity is between 0 and 1, with a larger value indicating more consistent waveforms.Pieces with a similarity exceeding 0.8 are marked as "highly similar segments." The system records their maximum strain difference, i.e., the difference between the maximum strain values ​​of the two segments at the same time point, and also records the time delay between the waveforms, i.e., the time difference between the strain peaks appearing at the two points, in seconds. Based on the maximum strain difference and time delay, the system calculates an influence score, assigning a weight of 0.6 to the maximum strain difference and 0.4 to the time delay. The influence score is the sum of the normalized values ​​of the two scores multiplied by their respective weights. The higher the score, the greater the influence of the damage source on adjacent segments. The system summarizes the influence scores of all path nodes into an influence weight matrix. Taking the path as the unit, the system calculates the proportion of influence of each segment node in the entire path, i.e., the score of a single node divided by the sum of the scores of all nodes in the path, to obtain the normalized weight value. The nodes with the top 20% weight values ​​in all paths are defined as "weighted key nodes." The system extracts these nodes and their upstream and downstream segments to construct an influence path diagram, identifies all possible transmission paths by number, and sorts them from high to low based on the total weight value to form a damage transmission priority path list. Ultimately, the system analyzes all segments along the top three priority paths, extracting data on strain fluctuations, disaster response, structural deformation records, and damage level changes over the past 60 minutes. This data is then categorized and organized according to time sequence and spatial relationships using information integration tools to construct a complete path risk distribution map. If a segment appears repeatedly on multiple high-priority paths, the system accumulates its risk scores and designates segments that appear more than three times and have a total risk score exceeding 0.75 as "high-risk clusters." These clusters are automatically added to the next phase's key monitoring list, triggering a high-frequency diagnostic mode.

[0028] The evolutionary analysis module constructs a dynamic evolutionary model of damage transmission between segments based on the direct impact weights of local damage on adjacent segments. It determines the phased distribution characteristics of damage accumulation, including the correlation between local damage and adjacent segments. Data acquisition tools are used to monitor damage distribution data in real time, obtaining preliminary impact data of local damage on adjacent segments and determining the initial boundaries of the impact range. Based on the initial boundary data, a preliminary framework for weight allocation is constructed to determine the impact weight distribution of each segment in damage transmission. Finally, based on the impact weight distribution and relevant information from the accumulation path, a path tracing tool is used to analyze the potential directions of damage transmission and determine the key aspects of the transmission dynamics. Key node distribution; based on the key node distribution, construct the framework of the evolutionary model to address the dynamic changes in transmission, and obtain the evolution data of damage accumulation at different time periods; for the evolution data, analyze the stage characteristics of damage accumulation. If the change amplitude of the stage characteristics exceeds the preset threshold, the data is stratified to determine the significant difference regions of the stage distribution; based on the significant difference regions, for the characteristic changes of the stage distribution, use information integration tools to classify and organize the data to obtain detailed classification results of the distribution characteristics; based on the classification results of the distribution characteristics, combine the logical mapping of feature judgment to digitize the distribution characteristics of damage accumulation and determine the priority ranking of the final distribution characteristics at different stages.

[0029] In the implementation of the evolutionary analysis module, the system first receives weighted data from the feature extraction module regarding the direct impact of local damage on adjacent segments. This weight, ranging from 0 to 1, represents the intensity of the current segment's influence on the damage spread of adjacent segments. The system extracts segments with weight values ​​greater than 0.4 as "critical impact sources." This threshold has been validated by a large amount of engineering data, accurately marking the core diffusion region in over 90% of samples. Subsequently, a high-frequency acquisition program is initiated at all strain sensing points of the "critical impact source" and all its adjacent segments, recording 100 sampling points per second, including strain, temperature, and displacement data, which are then stored in the data platform according to time sequence and segment number. The system calculates the average strain growth rate of all sensing points within each segment, with each 10-minute time window as a time interval. This value is calculated by subtracting the minimum strain value from the maximum strain value within that interval and dividing by the time interval of 600 seconds. This growth rate is then compared with the historical 60-minute average growth rate of that segment. If the current growth rate is more than twice the historical average, the segment is marked as a "dynamic response boundary node," and all such nodes constitute the initial boundary of the influence range. After establishing the initial boundaries, the system assigns influence weights to each segment. The weight is calculated as follows: the strain change rate of the segment over the most recent 30 minutes is multiplied by the similarity to the stress waveform of the source segment. The similarity is calculated using a dynamic time warping algorithm. Similarities greater than 0.8 are assigned a weight of 1, those between 0.5 and 0.8 are assigned 0.8, and those below 0.5 are assigned 0.5. The final score is normalized to obtain an influence weight distribution map. This map is a two-dimensional array, with the horizontal axis representing the time period number, the vertical axis representing the segment number, and the values ​​representing the influence weights. Based on this weight map, the system invokes a path tracing tool. Starting from the source node, the tool searches for paths with weights greater than 0.5 along the connection relationships segment by segment, accumulating the weight values ​​of the nodes for each path as its score. Each path contains at most 10 nodes; if the cumulative score of a path exceeds 5, the path is designated as a high-transmission path. The system extracts node damage indicators along the time dimension in all high-transmission paths, with data sets every 10 minutes. The data includes the node's average strain, maximum strain, minimum strain, and fluctuation range, forming a dynamic sequence of the path. The system constructs a matrix based on path and time numbers, with matrix elements representing the slope of indicator changes for each node in each time period, in microstrains per second. The system calculates the coefficient of variation (COP) on the matrix data, which is the ratio of the standard deviation to the mean. If the COP of a path node exceeds twice its historical mean within a certain time period, and remains in this state for two consecutive time periods, this time period is marked as a "stage transition period," and the node with the largest indicator change in the path is marked as a "stage critical node." For example, if the average strain of node 5 in path 3 jumps from 50 microstrains to 130 microstrains in time period 4, while the average strain in the previous period was only 60 microstrains, and the fluctuation direction of other nodes is consistent, then this period is recorded as a stage transition.The system then extracts node data from all transition periods and groups them according to transition magnitude. Each group includes the transition start time, end time, maximum value, and rate of change. The system sorts the data by rate of change from high to low, dividing them into high-change group (rate greater than 10 microstrains per second), medium-change group (between 5 and 10 microstrains per second), and low-change group (less than 5 microstrains per second). The statistical results of the number of segments within each group are used to assess the diffusion intensity and range change rate of that stage. The information integration tool further processes this grouped data, constructing feature items for each stage containing the following fields: average strain change rate, maximum strain increase, fluctuation frequency, number of stage nodes, and number of disaster co-occurrences. Each item is weighted at 0.3, 0.2, 0.2, 0.2, and 0.1, and the final score represents the evolution intensity of each stage. Based on this, the system establishes a "stage intensity ranking table," arranged from high to low score, and assigns "high priority," "medium priority," and "low priority" labels to the corresponding stages. High-priority stages are those with scores exceeding 0.7, medium-priority stages are those with scores between 0.4 and 0.7, and low-priority stages are those with scores below 0.4. The system marks the path nodes corresponding to high-priority stages in red on the structural visualization platform and automatically adds the segments of that stage to the early warning task queue, enabling real-time monitoring. The average strain rate of change is weighted at 0.3 because this parameter directly reflects the change in structural response intensity per unit time, serving as the primary basis for measuring the speed of damage evolution and having a decisive impact on the overall evolution trend. The maximum strain increase is weighted at 0.2 because it can capture extreme response states during the damage process; although a single-point value, it is significant in determining whether a structure is nearing its failure boundary. Fluctuation frequency is also weighted at 0.2 because it reflects instability during the damage process; frequent fluctuations often indicate drastic changes in structural state, possessing a strong risk indication function. The number of stage nodes is weighted at 0.2 based on the consideration of the damage's impact range; a higher number of nodes indicates a wider spread of impact and a more significant impact on the overall structural performance. The number of disaster synergies is weighted at 0.1 because, although an important contributing factor, it often manifests as a background factor, and its impact on a single stage is uncertain, thus receiving a relatively low weight. The overall weighting configuration follows the principle of combining impact degree, measurability, and discriminative effectiveness, ensuring that the indicator system accurately reflects the intensity of the damage evolution process while possessing good engineering applicability and algorithm stability.

[0030] The risk prediction module, if the phased distribution characteristics of damage accumulation exceed a preset threshold range, simulates the damage transmission and diffusion trend to obtain a preliminary risk distribution for overall performance prediction. This includes continuously monitoring the phased distribution of damage accumulation using a dynamic evolution model, and obtaining preliminary simulation data for potential directions when characteristics exceed the threshold. Based on the preliminary simulation data, a path tracing tool is used to analyze the changing characteristics of potential directions and identify key areas of risk distribution. Based on the distribution of key areas and combined with the overall performance prediction data, a multi-stage analysis framework for damage transmission is constructed to obtain dynamic change information of phased distribution. If the dynamic change information of phased distribution shows characteristics exceeding a preset threshold, the change information is processed hierarchically using an information integration tool to determine the priority areas of damage accumulation. Based on the distribution characteristics of priority areas and combined with the simulation results of the dynamic model, a data mapping technique is used to perform secondary calibration of the potential directions of transmission and diffusion, obtaining calibrated risk distribution data. Based on the calibrated risk distribution data and the overall performance prediction data, a multi-dimensional analysis matrix is ​​constructed to determine the distribution weights of damage accumulation at different stages. Based on the analysis results of the distribution weights and combined with the output data of the evolution simulation, the phased distribution of damage accumulation is continuously updated to obtain the final prediction data adjustment scheme.

[0031] During the operation of the risk prediction module, the system first calls the staged damage distribution data provided by the evolution analysis module. This data is presented in matrix form, with row indices representing time period numbers and column indices representing segment numbers in the structure. The values ​​in the matrix represent the damage level of the corresponding segment within that time period, ranging from 0 to 1, with higher values ​​indicating more severe damage. The damage level is calculated by dividing the average strain of all strain sensing points in the segment within that time period by the historical stable average strain of that segment, and then normalizing the result to ensure that the value fluctuates within the range of 0 to 1. The system sets a judgment threshold of 2.5, meaning that when the current average strain of a segment within a certain time period is greater than 2.5 times its historical stable average, the segment is considered to have experienced a damage surge. This threshold is derived from the statistical analysis of health monitoring data from 30 typical precast concrete tower structures, using the upper boundary value of the 95% confidence interval as the judgment standard. The system performs a rolling judgment on the evolution matrix every 10 minutes. If it finds that more than three segments simultaneously meet the above damage surge condition within a certain time period, it triggers the potential diffusion direction simulation program. The program extracts the numbers and spatial coordinates of all over-threshold segments and checks for previously unmarked diffusion path directions. If a new continuous connection exists that does not appear in the historical diffusion map, it is defined as a "potential diffusion path direction." Subsequently, the system calls a simulation algorithm to perform preliminary damage propagation calculations for this direction. The method involves taking the strain growth rate of all nodes on the path in the current time period and calculating the total strain growth rate of the path, in microstrain per minute. If the cumulative strain growth rate of the path exceeds 100 microstrains per minute, it is marked as a high-risk diffusion path. After confirming the high-risk path, the system analyzes the fluctuation trend of each node on the path, determining whether its fluctuation direction is consistent with the damage source node, and calculates the response delay time of adjacent nodes, defined as the time for the downstream node to experience a unidirectional strain surge minus the time for the upstream node. If the delay is less than 30 seconds and more than three consecutive groups of nodes appear, it is considered a path with a consistent trend. The system designates the area where this path is located as a critical risk area. Then, the system loads data from the overall performance prediction module, including changes in the overall structural modal frequency, maximum displacement response, and load center of gravity, in Hertz, millimeters, and millimeters, respectively. The system performs linear fitting between the cumulative damage intensity of the risk area and the three overall indicators mentioned above to determine the trend of the risk area's impact on overall performance. If the fitting residual of each indicator exceeds the preset tolerance of 5% for three consecutive time periods (determined by the sum of the historical residual mean and standard deviation), the area is considered to have a transmission accuracy offset and needs to enter the hierarchical processing procedure for change information.The system categorizes the node data within the abnormal region into three-dimensional information volumes based on path number, node number, and time number. For each data volume, it performs statistical analysis on three dimensions: fluctuation amplitude (the difference between the current maximum and minimum strain values), fluctuation duration (the total number of time periods exceeding 5% amplitude), and node response consistency (the proportion of nodes with consistent fluctuation directions). These three indicators are weighted at 0.4, 0.3, and 0.3 respectively. Based on the scores, each data volume is assigned a risk level label: scores above 0.7 define a Level 1 risk area, 0.4 to 0.7 a Level 2 risk area, and below 0.4 a normal area. Risk levels are used to adjust path scoring priorities. Furthermore, the system performs morphological matching between the current path and all high-risk paths in the historical evolution model database, calculating the similarity between structural layout, node number sequence, and trend sequence. If the average similarity of these three factors exceeds 0.8, the current path is considered to highly overlap with a typical historical high-risk path. The system then corrects the current path score by the difference between the highest historical path score and the current score, ensuring the current model more closely reflects the actual risk distribution. The system then multiplies the risk scores of all nodes by their corresponding overall performance impact factors to generate a risk impact matrix. The matrix dimensions are the number of paths × the number of time periods × the number of nodes. Each data point is the node's risk score multiplied by the slope of the path's fit to the overall performance. The system accumulates the total impact value of each path over time periods, representing the potential impact intensity of each path on the overall structural performance at each stage. If the cumulative impact value of a path exceeds 150% of the average of all paths over three consecutive time periods, it is included in the priority correction path list. The system automatically performs weight enhancement processing on its simulation parameters for the next three time periods, increasing the parameter weight for its participation in the next round of evolution simulation by 20%. Finally, the system outputs the risk prediction data adjustment scheme as follows: updated risk scores for each path node, predicted values ​​of overall performance impact at each stage, and model residual prediction graphs. This scheme is updated every 30 minutes and forms a closed-loop interaction with the evolution analysis module, creating a high-frequency prediction and real-time correction intelligent monitoring mechanism.

[0032] The key node identification module predicts the initial risk distribution based on overall performance. Addressing the coupling effect of multiple disasters, it tracks the evolution of the risk distribution and identifies key time points for quantifying damage status. This includes continuously monitoring the distribution pattern of the initial risk using time series analysis to understand the correlation between overall performance and initial risk, acquiring the changing characteristics of the distribution pattern over different time periods, and determining the initial evolution trend. Based on this initial evolution trend, and considering the coupling effect of complex conditions and multiple disasters, a pre-established disaster impact model is used to decompose the evolution trend in multiple dimensions, obtaining risk fluctuation data under the coupling effect. For the risk fluctuation data, if the fluctuation amplitude exceeds a preset threshold range, information integration tools are used to perform stratified processing of the fluctuation data to determine the potential impact of multiple disasters on the damage status. The assessment process involves several steps: First, based on the potential impact of the damage state, and considering the correlation between quantification nodes and key time points, data mapping techniques are used to locate the impact degree along a timeline, obtaining the distribution characteristics of key time points. Second, considering the correlation between evolution patterns and distribution patterns, data comparison tools are used to dynamically update the distribution characteristics, determining the priority ranking of evolution patterns at different stages. Third, based on the priority ranking, regression analysis is used to perform a secondary calibration of the ranking results, obtaining calibrated risk distribution weights, and finally, considering the correlation between key time points and quantification nodes, information fusion tools are used to comprehensively process the weight data, determining the quantification results of the damage state of overall performance at different time points.

[0033] During the operation of the critical node identification module, the system first imports the overall structural performance indicators from the overall performance prediction module, including three parameters: maximum lateral displacement, first-order natural frequency change, and vertical acceleration response, expressed in millimeters, Hertz, and meters per square second, respectively. Each parameter is updated every 10 minutes, and the system constructs three time series for these parameters. Simultaneously, the system receives preliminary risk distribution data output from the risk prediction module, represented as a two-dimensional matrix. The row index represents the time period number, and the column index represents the segment number within the structure. Each value in the matrix is ​​the segment's risk score for that time period, ranging from 0 to 1. The segment's risk score is calculated by weighting three indicators: strain increment, strain fluctuation rate, and historical damage level within the current time period, with weights of 0.5, 0.3, and 0.2, respectively. The system averages the risk scores of all segments over each time period to form an overall structural risk sequence. This sequence is then time-aligned with the overall performance time series to construct a unified correlation analysis table. Subsequently, the system employs a sliding window algorithm to extract trends from the risk score sequence. The window length is three time periods with a step size of 1. Within each window, the difference between the maximum and minimum score values ​​is calculated as the risk fluctuation amplitude, expressed as the change in score value. If the fluctuation amplitude within a certain time period is greater than twice the average fluctuation amplitude of all windows over the past 30 minutes, and the risk score for that time period increases by more than 0.3 compared to the previous time period, that time period is marked by the system as a "preliminary risk change inflection point." The inflection point identification threshold of 0.3 is a stable lower limit extracted from a large amount of simulation data and measured structural data on the historical risk surge ratio before damage. After identifying multiple inflection points, the system synchronously imports disaster environmental data, including wind speed, seismic acceleration, and temperature changes, recorded in meters per second, meters per square second, and degrees Celsius, respectively, with one set of data per minute. The system extracts and aligns disaster data within a 30-minute range before and after the inflection point of risk change. If wind speed increases by more than 5 meters per second within 15 minutes, or seismic acceleration changes by more than 0.05 meters per square second, or temperature fluctuates by more than 4 degrees Celsius within 30 minutes, and the time difference between the peak of disaster data and the peak of the risk score is less than 20 minutes, then the current risk change is considered to be affected by disaster coupling effects, and this period is marked as the "disaster coupling period." Within the "disaster coupling period," the system calls the built-in disaster impact model, multiplies various disaster factors by their corresponding sensitivity coefficients, and weights them by summing them. The sensitivity coefficient for wind speed is 0.4, the sensitivity coefficient for seismic acceleration is 0.4, and the sensitivity coefficient for temperature is 0.2. The weighted sum is the disaster enhancement coefficient for this period. Subsequently, the system multiplies the disaster enhancement coefficient by the original risk score growth rate to obtain the disaster enhancement risk value and reconstructs the risk evolution trend sequence.The system performs fluctuation intensity identification processing on the sequence, which involves calculating the standard deviation of the disaster enhancement risk value over three consecutive time periods. If this standard deviation is greater than 1.5 times its historical average and meets the condition twice consecutively, the system records this time period as a "high-risk fluctuation segment". All segments with a risk score greater than 0.8 corresponding to "high-risk fluctuation segments" are considered quantified nodes. The system pairs these node numbers with their corresponding time period numbers to form a "quantified node-key time period mapping table". Subsequently, a density clustering algorithm is used to analyze whether there are concentrated distribution areas in these time periods. The density clustering algorithm first treats each key time node as a data point to be clustered, and calculates the time difference between any two nodes using time as the sole dimension. The minimum time interval is set to 10 minutes, and the minimum number of cluster points is 3. The system performs a neighborhood search on all nodes to determine whether at least 3 key time nodes are clustered within 10 minutes. If this condition is met, these nodes are assigned to the same cluster, and the search continues to expand to the neighborhood until no more nodes meet the condition. Ultimately, multiple clusters are formed, each representing a high-density critical time region. If a time period is covered by more than three high-risk nodes, that time period is identified as a "critical time node" and recorded as a "critical time node list." The system then divides the overall structural risk score time series into several evolution stages based on the critical time points, using these critical time points as boundaries. Within each stage, the risk score growth slope is calculated as the risk evolution rate, and the evolution rates are sorted in descending order. The top 30% of stages are marked as "high-priority stages," the middle 40% as "medium-priority stages," and the remainder as "low-priority stages." Simultaneously, the system extracts the performance change values ​​corresponding to each stage from the overall performance time series and constructs a multiple regression model, setting the risk score as the dependent variable and the performance change values ​​as the independent variables, and constructs a fitting function. If the model's coefficient of determination is greater than 0.8 and the mean residual is less than 10%, the model is considered effective, and the evolution priority results of each stage are used as the regression coefficient calibration weights to obtain the "calibrated risk distribution weights." Finally, the system combines all key time nodes, their corresponding risk weights, and the number of quantified nodes to construct a three-dimensional matrix using information fusion tools. Each element of the matrix is ​​the node's risk score multiplied by the risk weight of the time period and the node's propagation level in the path. The propagation level is determined by the path structure and ranges from 1 to 5. The sum of all node matrix values ​​for each time period yields a quantified score of the overall structural damage status for that time period. If the score exceeds 0.75, which is the lower limit of the structure's preset safety tolerance, the system activates an early warning mechanism, outputting the corresponding time point, node number, and overall structural score as input to the response control module.

[0034] The assessment and calculation module constructs a multi-dimensional indicator system for quantifying damage status using data from key time nodes. This allows for the acquisition of comprehensive damage assessment results for prefabricated structures under different disaster scenarios. This includes classifying information at each time node using data integration tools, obtaining the categorized time node feature distribution, and determining the variation patterns of these feature distributions across different time periods. Based on these variation patterns, and considering segmented transmission and dynamic characteristics, a pre-established feature mapping model is used to perform hierarchical analysis of the variation patterns, resulting in layered dynamic characteristic data. Finally, for this layered dynamic characteristic data, information fusion tools are used to perform multi-dimensional comparisons of the dynamic characteristic data, combining damage status and quantitative indicators, to determine the quantification of damage status across different dimensions. The distribution is analyzed as follows: If the quantitative distribution exceeds a preset threshold, the distribution is structured using data filtering tools, taking into account the multidimensional system and prefabricated structure, to obtain the structured damage distribution characteristics. Based on the structured damage distribution characteristics, combined with the disaster scenario and comprehensive assessment, logistic regression analysis is used to match the feature data to the scenario, determining the matched scenario damage assessment results. For the matched scenario damage assessment results, combined with damage data and status analysis, the assessment results are deeply analyzed through an information integration platform to obtain the final quantitative distribution of damage status. Based on the final quantitative distribution of damage status, combined with key time and node data, data update tools are used to dynamically adjust the quantitative distribution, judging the adaptability of the adjusted distribution characteristics at the time nodes.

[0035] During the evaluation and calculation module's operation, the system first retrieves all identified key time nodes and corresponding high-risk segment numbers from the key node identification module. These time nodes are represented by timestamps, with each 10-minute period as a time unit. At each time node, the system extracts five characteristic parameters: corresponding structural performance index value, segment risk score, strain increase, frequency change rate, and vertical acceleration fluctuation. All parameters are indexed into a multi-dimensional data matrix, with the matrix dimension being the number of time periods multiplied by the five indicators. The system uses a data integration tool to classify this matrix by time period, calculating the mean, range, and standard deviation of all indicators within each time period to obtain the categorized time node characteristic distribution curve. Next, the system dynamically compares the time series of each statistical indicator with the risk score curve, calculating the Pearson correlation coefficient as the basis for judging the correlation. If the correlation coefficient is greater than 0.7, the statistical feature is considered to have a significant correlation with the evolution of the damage state at the current stage. Subsequently, the system invokes the constructed structural segmentation transmission network model and disaster type dynamic impact model to perform hierarchical analysis on the aforementioned feature distribution sequence. The analysis process is divided into three layers: the first layer is the local segment risk evolution layer, the second layer is the inter-segment contact coupling layer, and the third layer is the overall performance response layer. The system inputs current feature data into each layer, uses a feature mapping model for matching transformation, and outputs the corresponding dynamic characteristic indicators for each layer, such as the local risk increase rate, coupling strength change rate, and overall frequency shift speed. All dynamic characteristics are summarized according to the hierarchical dimension and the time dimension, constructing a three-dimensional dynamic characteristic array. After standardizing each indicator sequence in this array, the system assigns weights to each indicator according to set weight coefficients. The default weights are 0.4 for the local risk indicator, 0.3 for the coupling layer indicator, and 0.3 for the overall response indicator. The weights are derived from the optimal proportion determined by fitting the range of changes in assessment accuracy under different structural monitoring results. After feature weighted fusion, the system obtains the comprehensive structural damage value for each time period, ranging from 0 to 1. The system determines that if the damage value exceeds a set threshold of 0.75 (the midpoint between the structural allowable damage capacity and the safety reserve limit) in any time period, it considers that the structural response has exceeded the limit. In this case, the system uses a data filtering tool to sort and extract the damage values ​​of each segment within the exceeding time period, and performs cluster analysis on the distribution of their locations in the structural space to determine if there are damage clusters. If any cluster contains more than 5 adjacent segments and the average damage value is greater than 0.8, this cluster is defined as a structural damage hotspot. The system regenerates a structured damage distribution map centered on this area, with the map representing the segment number, time, and overall damage value using three-dimensional coordinates.

[0036] The system then performs logistic regression matching between the feature values ​​in the map and pre-constructed multi-hazard scenario templates. Each template consists of typical features of structural response under a specific hazard type. Using the current map data as input variables, the system estimates the matching probability using a logistic regression model. If the matching probability of any template exceeds 0.9, the current structural state is considered to have successfully matched the scenario. The system outputs the category of the successfully matched hazard scenario and its corresponding damage assessment level, which is divided into low risk (damage value less than 0.5), medium risk (damage value between 0.5 and 0.75), and high risk (damage value greater than 0.75). The system further analyzes the structural changes in the spatial distribution and temporal evolution of the damage level for each matching result. Through the information integration platform, gradient difference analysis is performed on the scores of each segment in each time period to determine whether the changes are continuous, drastic, or concentrated. If the rate of change of the difference exceeds 0.2 every 10 minutes, it is marked as a "mutation point." All time periods containing mutation points are back-calculated again to ensure the consistency and dynamic rationality of the scoring results. Finally, the system merges the backtracked scoring data with the original key time point sequence. If the direction of the scoring trend change is consistent after fusion, the system records the quantitative distribution result as the final state and saves it as the "Final Damage State Quantitative Distribution Table". The system then compares and updates this table with the historical scoring standards in the structural database. If there is a deviation of more than 5%, the system calls the data update tool to iteratively correct the standard curve, thus completing the adaptive dynamic adjustment process of the quantitative distribution at each key time point.

[0037] The feedback correction module, based on the comprehensive damage assessment results, constructs a feedback correction mechanism. Through iterative updates of the multi-dimensional indicator system, it obtains an optimized distribution of damage assessment accuracy. This includes: using the comprehensive assessment results, employing data integration tools to perform preliminary classification of the damage results and obtain the classified damage distribution characteristics; based on the classified damage distribution characteristics, combined with the feedback mechanism, using a pre-established mapping model to perform deviation analysis on the distribution characteristics and determine the deviation distribution range; for the deviation distribution range, combining overall performance and prediction deviation, if the deviation range exceeds a preset threshold, then using information comparison tools to perform multi-dimensional decomposition of the deviation data to obtain the decomposed deviation characteristics; based on the decomposed... Deviation characteristics are analyzed by combining multidimensional indicators and system iteration. Data update tools are used to adjust the deviation characteristics layer by layer to obtain the adjusted indicator distribution. For the adjusted indicator distribution, the distribution data is deeply compared through an information fusion platform, combining update operations and evaluation accuracy, to determine the accuracy distribution after comparison. Based on the accuracy distribution after comparison, combined with optimized distribution and performance analysis, if the accuracy distribution does not meet the preset standard, the distribution is structured using data filtering tools to obtain the structured optimized distribution characteristics. For the structured optimized distribution characteristics, logistic regression analysis is used to finally calibrate the feature data to determine the calibrated comprehensive performance distribution.

[0038] In the specific implementation of the feedback correction module, the system first reads the final damage state quantitative distribution data generated by the evaluation calculation module and compares it point by point with the predicted data corresponding to the same time period to calculate the damage error value of all segments. The error value is defined as the measured damage value minus the predicted damage value, with a range of -1 to 1. The system classifies the error values ​​of all segments into five levels: extremely low error, small error, medium error, large error, and extremely high error. The classification criteria are: an absolute error value less than 0.05 is extremely low error, 0.05 to 0.1 is small error, 0.1 to 0.2 is medium error, 0.2 to 0.3 is large error, and greater than 0.3 is extremely high error. This classification is based on historical monitoring error distribution statistics and combined with the tolerance range determined by expert experience. The system maps various types of errors to coordinates in the structural space to generate a preliminary error spatial distribution map. Then, the system performs cluster analysis on the error spatial distribution map. If the number of segments with medium or higher errors in any cluster region exceeds 5 and the average error value is greater than 0.2, then the region is marked as an error concentration area. The system extracts raw monitoring data such as strain, frequency, displacement, and acceleration from all segments within the error concentration area and establishes a segmented multidimensional feature matrix. The matrix dimension is the number of segments multiplied by four feature parameters. Max-min normalization is performed on each feature value to ensure uniform feature scale. Then, the relative error of each feature parameter in each segment is calculated, specifically the difference between the measured value and the theoretical estimate divided by the theoretical value. The system uses a relative error exceeding 10% as a threshold to filter out feature dimensions with significant errors, forming a deviation dimension vector indicating which features in that segment exhibit systematic offsets. All deviation dimension vectors are aggregated and input into the system mapping model. The model performs logical operations according to structural coupling relationships, material parameters, and transmission paths, outputting the possible concentrated causes and paths of deviation sources. If the deviation path is consistent with the original damage propagation path, it indicates that the error source is a deviation in the prediction model parameters. In this case, the system will call the data update tool to adjust the parameter weights in the damage prediction model based on the existing deviation vector. The parameter weight update uses a weighted moving average method. The weight is calculated by multiplying the current error by the current segment's status influence factor and dividing by the sum of the total influence factors. The influence factor is set based on the segment's structural level, coupling strength, and historical reliability score. Segments with high levels have a factor value of 1.5, ordinary segments have 1, and marginal segments have 0.5. After adjustment, the system re-runs the prediction model to generate new damage assessment values ​​and compares them with the previous results. If the overall average error decreases by more than 20%, or the number of extremely high error points in the error concentration area decreases by more than 50%, the correction is considered effective. Otherwise, the system enters the structured processing stage, reclassifying and combining all segments according to their floor, component type, and connection method, and re-establishing the structured feature matrix.Principal component analysis (PCA) is performed on the structured matrix to extract the first three principal components and match them with historical damage templates. If the similarity of the principal component scores exceeds 90%, the current state is considered a typical damage pattern, and the system records its features and updates the template library. Finally, the system uses a logistic regression algorithm to classify all updated data for final performance level. The algorithm uses five-dimensional features as independent variables and known structural risk levels as labels for model training. When the classification accuracy exceeds 90% in cross-validation, the system sets the current evaluation state as the final evaluation result and feeds this result back to the database as a reference for the next round of prediction correction. Simultaneously, the system automatically updates the parameter records, error statistics table, and model configuration file during the current iteration process to ensure higher prediction accuracy and a complete feedback loop in subsequent runs.

[0039] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A prefabricated multi-tower full-life-cycle multi-hazard coupled damage early warning and intelligent monitoring system, characterized in that: include: The data acquisition module acquires the stress data and environmental difference performance data of each segment under complex conditions, and performs comprehensive processing on the stress distribution and environmental response to obtain the initial damage state distribution of each segment. The damage propagation modeling module constructs a damage propagation network diagram based on the initial damage state distribution, taking into account the effects of contact surfaces and connection points between segments, and determines potential damage propagation paths. The feature extraction module extracts features based on the potential path of damage propagation and the impact of the coupling of diffusion trends and multiple disasters, and obtains the weight of the direct impact of local damage on adjacent segments. The evolution analysis module constructs a dynamic evolution model of damage transmission between segments based on the weight of the direct impact of local damage on adjacent segments, and determines the stage distribution characteristics of damage accumulation. The risk prediction module simulates the damage transmission and diffusion trend if the phased distribution characteristics of accumulated damage exceed the preset threshold range, and obtains the preliminary risk distribution of overall performance prediction. The critical node identification module predicts the initial risk distribution based on overall performance, tracks the evolution of risk distribution in response to the coupling effect of multiple disasters, and determines the key time nodes for quantifying damage status. The assessment and calculation module constructs a multi-dimensional indicator system for quantifying damage status using data from key time points, and obtains comprehensive damage assessment results for prefabricated structures under different disaster scenarios; The feedback correction module constructs a feedback correction mechanism based on the comprehensive damage assessment results. Through iterative updates of the multi-dimensional index system, it obtains an optimized distribution of damage assessment accuracy.

2. The prefabricated multi-tower full-life-cycle multi-hazard coupled damage early warning and intelligent monitoring system according to claim 1, characterized in that: The data acquisition module obtains stress data and environmental difference data of each segment under complex conditions, comprehensively processes the stress distribution and environmental response to obtain the initial damage state distribution of each segment. This includes constructing a segment model of the prefabricated structure, acquiring stress data and environmental difference data of each segment under complex conditions, using data acquisition tools to initially process multi-source data, obtaining the original stress distribution and environmental response records of each segment; based on the original stress distribution and environmental response records, using fusion technology to comprehensively process the multi-source data, performing weighted analysis on the stress data and environmental difference data to determine the stress distribution characteristics and environmental response patterns of each segment; through the analysis of stress distribution characteristics and environmental response patterns, identifying stress concentration areas and environmentally sensitive points of each segment under complex conditions, and using a support vector machine algorithm to analyze key areas. The system classifies and identifies potential initial damage locations. If the classification results show that the stress concentration area of ​​a segment exceeds a preset threshold, a deep comparison of the environmental response pattern of that segment is performed to obtain its correlation coefficient with the environment, thus determining the range of initial damage. Based on the range of initial damage, secondary verification is performed on the stress distribution characteristics and environmental response patterns. Anomalies in multi-source data are filtered using the information entropy calculation method to obtain preliminary results of the damage state distribution of each segment. By integrating the preliminary results of the damage state distribution, comprehensive performance data of each segment under complex conditions is obtained. Data backtracking is performed on anomalies and potential damage locations to determine the final initial damage state distribution. Based on the final initial damage state distribution, data visualization tools are used to dynamically map the stress distribution and environmental response of each segment, obtaining a panoramic view of the damage distribution of each segment under complex conditions.

3. The prefabricated multi-tower full-life-cycle multi-hazard coupled damage early warning and intelligent monitoring system according to claim 1, characterized in that: The damage transmission modeling module, based on the initial damage state distribution and considering the effects of contact surfaces and connections between segments, constructs a damage transmission network diagram to determine potential damage transmission paths. This includes classifying and organizing the damage distribution data between segments using data integration tools, obtaining damage distribution characteristic data for each segment, and identifying key differences in the distribution characteristics. Based on the damage distribution characteristic data between segments, it analyzes the force characteristics of contact and connection points between segments, obtains mechanical action data for contact surfaces and connection areas, and determines the distribution range of the effects. Based on the distribution range of the effects, it constructs a damage transmission network model and uses graph analysis to analyze the interactions between segments. The interaction relationships are digitally mapped to obtain a network graph. Based on this network graph, potential damage transmission paths are analyzed, and path tracing technology is used to simulate the transmission direction, identifying the main nodes and branches of the potential paths. For the main nodes and branches of the potential paths, the influence of each node on the transmission process is analyzed. If the transmission intensity between nodes exceeds a preset threshold, the path is prioritized to obtain the transmission paths of priority. Based on the priority transmission paths, combined with state distribution and piecewise contact data, data comparison is performed on key areas along the paths to obtain the dynamic changes in damage transmission along the paths and identify risk areas in the transmission process. For risk areas in the transmission process, the support vector machine algorithm is used to classify the damage distribution characteristics within the area, obtain the risk level distribution after classification, and determine the key monitoring scope of the risk area.

4. The prefabricated multi-tower full-life-cycle multi-hazard coupled damage early warning and intelligent monitoring system according to claim 1, characterized in that: The feature extraction module extracts features based on the potential damage propagation path and the coupling effect of the diffusion trend and multiple disasters. It obtains the direct impact weight of local damage on adjacent segments, including the potential path of damage propagation. Data acquisition tools are used to monitor the diffusion process of local damage in real time, acquiring dynamic feature data to determine the initial range and direction of the diffusion. Based on the initial range and direction of the diffusion, time series analysis is used to process the dynamic feature data to obtain the fluctuation pattern and key time nodes of the diffusion trend. Based on the fluctuation pattern and key time nodes of the diffusion trend, the coupling effect of multiple disasters is analyzed to obtain the damage aggravation characteristics under the coupling effect and determine the superposition effect of damage under different disasters. By leveraging the cumulative effects of damage under different disasters, and considering the impact on adjacent segments, a data comparison tool is used to perform a stratified analysis of damage distribution, revealing the direct impact on adjacent segments. Based on the direct impact on adjacent segments, a weight allocation model is constructed to determine the impact weight value of each segment in damage transmission, identifying key areas for weight distribution. Using these key areas, and considering the correlation between local damage and adjacent segments, a logical mapping method is employed to digitize the impact paths, obtaining a priority ranking of damage transmission. Based on this priority ranking, information integration tools are used to classify and organize path data for key nodes along paths from high to low priority, determining the distribution of risk areas along the paths.

5. The prefabricated multi-tower full-life-cycle multi-hazard coupled damage early warning and intelligent monitoring system according to claim 1, characterized in that: The evolutionary analysis module constructs a dynamic evolutionary model of damage transmission between segments based on the direct impact weights of local damage on adjacent segments. It determines the phased distribution characteristics of damage accumulation, including the correlation between local damage and adjacent segments. Data acquisition tools are used to monitor damage distribution data in real time, obtaining preliminary impact data of local damage on adjacent segments and determining the initial boundaries of the impact range. Based on the initial boundary data, a preliminary framework for weight allocation is constructed to determine the impact weight distribution of each segment in damage transmission. Based on the impact weight distribution and relevant information of the accumulation path, a path tracing tool is used to analyze the potential directions of damage transmission and determine the distribution of key nodes in the transmission dynamics. Based on the distribution of key nodes and the changing trends of the transmission dynamics, the framework structure of the evolutionary model is constructed to obtain the evolution data of damage accumulation at different time periods. For the evolution data, the stage characteristics of damage accumulation are analyzed. If the change in stage characteristics exceeds a preset threshold, the data is stratified to determine the significant difference regions in the stage distribution. Based on the significant difference regions, information integration tools are used to classify and organize the data for the characteristic changes in the stage distribution, resulting in detailed classification results of the distribution characteristics. Based on the classification results of the distribution characteristics, combined with the logical mapping of feature judgment, the distribution characteristics of damage accumulation are digitized to determine the priority ranking of the final distribution characteristics in different stages.

6. The prefabricated multi-tower full-life-cycle multi-hazard coupled damage early warning and intelligent monitoring system according to claim 1, characterized in that: If the phased distribution characteristics of damage accumulation exceed a preset threshold range, the risk prediction module simulates the damage transmission and diffusion trend to obtain the preliminary risk distribution of overall performance prediction. This includes continuously monitoring the phased distribution of damage accumulation through a dynamic evolution model and obtaining preliminary simulation data of potential directions for cases where the characteristics exceed the threshold. Based on preliminary simulation data, a path tracing tool is used to analyze the changing characteristics of potential directions and identify key areas of risk distribution, targeting the transmission and diffusion trends. Based on the distribution of these key areas and combined with overall performance prediction data, a multi-stage analysis framework for damage transmission is constructed to obtain dynamic changes in stage distribution. If the dynamic changes in stage distribution exceed a preset threshold, information integration tools are used to stratify the changes and determine priority areas for damage accumulation. Based on the distribution characteristics of priority areas and the simulation results of the dynamic model, data mapping technology is used to perform secondary calibration on the potential directions of transmission and diffusion, obtaining calibrated risk distribution data. Using the calibrated risk distribution data and the overall performance prediction data, a multi-dimensional analysis matrix is ​​constructed to determine the distribution weights of damage accumulation at different stages. Based on the analysis results of the distribution weights and combined with the output data of the evolution simulation, the stage distribution of damage accumulation is continuously updated to obtain the final prediction data adjustment scheme.

7. The prefabricated multi-tower full-life-cycle multi-hazard coupled damage early warning and intelligent monitoring system according to claim 1, characterized in that: The key node identification module predicts the initial risk distribution based on overall performance, tracks the evolution of risk distribution in response to the coupling effect of multiple disasters, and determines the key time nodes for quantifying damage status. This includes continuously monitoring the distribution pattern of the initial risk using time series analysis technology to obtain the changing characteristics of the distribution pattern in different time periods and determine the initial evolution trend. Based on the preliminary evolution trend, and combined with the coupling effect of complex conditions and multiple disasters, the evolution trend is decomposed in multiple dimensions using a pre-established disaster impact model to obtain risk fluctuation data under the coupling effect. For risk fluctuation data, if the fluctuation range exceeds the preset threshold range, the fluctuation data is processed in layers through information integration tools to determine the potential impact of multiple disasters on the damage status. Based on the potential impact of the damage state, and combined with the correlation between quantitative nodes and key times, data mapping technology is used to locate the impact on the time axis and obtain the distribution characteristics of key time nodes. For the distribution characteristics of key time nodes, and combining the correlation between evolution patterns and distribution patterns, the distribution characteristics are dynamically updated using data comparison tools to determine the priority ranking of evolution patterns at different stages. Based on the priority ranking results, and considering the correlation between overall performance and damage state, regression analysis technology is used to perform a secondary calibration of the ranking results to obtain the calibrated risk distribution weights. Based on the calibrated risk distribution weights, and combined with the correlation between key times and quantification nodes, the weight data is comprehensively processed using information fusion tools to determine the quantification results of the overall performance damage status at different time nodes.

8. The prefabricated multi-tower full-life-cycle multi-hazard coupled damage early warning and intelligent monitoring system according to claim 1, characterized in that: The assessment and calculation module constructs a multi-dimensional indicator system for quantifying damage status using data from key time nodes. This system obtains comprehensive damage assessment results for prefabricated structures under different disaster scenarios. This includes classifying information at each time node using data integration tools, obtaining the categorized time node feature distribution, and determining the variation patterns of the feature distribution across different time periods. Based on these variation patterns, and considering segmented transmission and dynamic characteristics, a pre-established feature mapping model is used to perform hierarchical analysis of the variation patterns, resulting in hierarchical dynamic characteristic data. For the hierarchical dynamic characteristic data, information fusion tools are used to perform multi-dimensional comparisons of the dynamic characteristic data, combining damage status and quantitative indicators, to determine the quantitative distribution of damage status across different dimensions. If the quantitative distribution exceeds a preset threshold, the distribution is structured using data filtering tools, combining the multi-dimensional system and the prefabricated structure, to obtain structured damage distribution characteristics. Based on the structured damage distribution characteristics, combined with disaster scenarios and comprehensive assessments, logistic regression analysis is used to match the feature data to the scenarios, determining the matched scenario damage assessment results. For the matched scenario damage assessment results, combined with damage data and status analysis, the assessment results are deeply analyzed through an information integration platform to obtain the final quantitative distribution of damage status. Based on the final quantitative distribution of damage status, combined with key time and node data, a data update tool is used to dynamically adjust the quantitative distribution, judging the adaptability of the adjusted distribution characteristics at the time nodes.

9. The prefabricated multi-tower full-life-cycle multi-hazard coupled damage early warning and intelligent monitoring system according to claim 1, characterized in that: The feedback correction module constructs a feedback correction mechanism based on the comprehensive damage assessment results. Through iterative updates of the multi-dimensional indicator system, it obtains an optimized distribution of damage assessment accuracy. This includes: using the comprehensive assessment results, employing data integration tools to perform preliminary classification of the damage results and obtain the classified damage distribution characteristics; based on the classified damage distribution characteristics, combined with the feedback mechanism, using a pre-established mapping model to perform deviation analysis on the distribution characteristics and determine the deviation distribution range; for the deviation distribution range, combining overall performance and prediction deviation, if the deviation range exceeds a preset threshold, then using information comparison tools to perform multi-dimensional decomposition of the deviation data and obtain the decomposed deviation characteristics; based on the decomposed deviation characteristics, combined with multi-dimensional indicators and system iteration, using data update tools to adjust the deviation characteristics layer by layer and obtain the adjusted indicator distribution.

10. The prefabricated multi-tower full-life-cycle multi-hazard coupled damage early warning and intelligent monitoring system according to claim 9, characterized in that: The feedback correction module constructs a feedback correction mechanism based on the comprehensive damage assessment results. Through iterative updates of the multi-dimensional indicator system, it obtains an optimized distribution of damage assessment accuracy. This also includes, for the adjusted indicator distribution, combining the update operation and assessment accuracy, performing a deep comparison of the distribution data through an information fusion platform to determine the accuracy distribution after comparison. Based on the accuracy distribution after comparison, combined with the optimized distribution and performance analysis, if the accuracy distribution does not meet the preset standard, the distribution is structured using a data filtering tool to obtain the structured optimized distribution characteristics. Based on the optimized distribution characteristics after structuring, logistic regression analysis is used to perform final calibration of the feature data to determine the calibrated comprehensive performance distribution.

Images