A building structure safety monitoring method and system based on a digital twin model

Abstract

The application discloses a kind of based on digital twin model's building structure safety monitoring method and system, including obtaining the target environment parameter and target monitoring point parameter of building structure, input damage characteristic prediction model and obtain the damage rate distribution characteristic coefficient of building structure in different period;The damage distribution characteristic mode of building structure and the coefficient nonlinear combination generation target damage rate distribution matrix;Obtain the first performance index distribution matrix of building structure obtained in last monitoring period, and calculate the second performance index distribution matrix of current period in combination with target damage rate distribution matrix;Judge whether the value of any monitoring point position is exceeded the structural failure threshold value of this position, when exceeding, mark as abnormal monitoring point and trigger multi-stage early warning;Diagnose the damage type of abnormal monitoring point and match corresponding repair strategy.The application realizes the fine simulation and accurate prediction of building structure damage evolution process, and provides reliable basis for performance evaluation and matching repair strategy.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of building safety monitoring, and particularly relates to a building structure safety monitoring method and system based on a digital twin model. BACKGROUND

[0002] In the field of building structure safety monitoring, traditional methods mostly rely on real-time data collection of single sensors or periodic manual inspection. In the prior art, some solutions only judge the safety state by monitoring the real-time displacement, stress and other single-point parameters of the structure, lacking systematic analysis of the correlation between environmental factors and structural damage evolution. Some methods introduce damage prediction models, but often use linearization processing methods, which are difficult to accurately capture the nonlinear damage accumulation effect of building structures under complex environments. At the same time, the early warning mechanism of the prior art is mostly based on a single threshold trigger, without considering the performance degradation rate differences of different monitoring points, which can easily cause early warning lag or false alarm problems, and after an anomaly occurs, it is difficult to quickly locate the damage type and match the repair strategy, resulting in low efficiency of safety hazard investigation and high structure maintenance cost. SUMMARY

[0003] In order to solve at least one of the above technical problems, the present application provides a building structure safety monitoring method and system based on a digital twin model.

[0004] In a first aspect, the present application provides a building structure safety monitoring method based on a digital twin model, the method comprising:

[0005] Obtaining target environmental parameters and target monitoring point parameters of the building structure, inputting them into a damage feature prediction model to obtain damage rate distribution characteristic coefficients of the building structure at different time periods;

[0006] Monitoring the damage distribution characteristic mode of the building structure, and performing nonlinear combination of the damage distribution characteristic mode and the damage rate distribution characteristic coefficients to generate a target damage rate distribution matrix;

[0007] Obtaining a first performance index distribution matrix of the building structure obtained in the last monitoring period, and calculating a second performance index distribution matrix of the current time period according to the first performance index distribution matrix and the target damage rate distribution matrix;

[0008] Judging whether the value of any monitoring point position in the second performance index distribution matrix exceeds the structural failure threshold of the position, marking as an abnormal monitoring point and triggering a multi-level early warning when it exceeds; diagnosing the damage type of the abnormal monitoring point, and matching the corresponding repair strategy according to different damage types.

[0009] Preferably, the damage feature prediction model is a deep learning-based time series prediction model.

[0010] The target environment parameters at least include environment temperature, humidity, wind speed and peak ground acceleration;

[0011] The target monitoring point parameters at least include historical strain data, displacement data and acceleration data of the monitoring point;

[0012] The damage feature prediction model is trained by historical data and simulation data generated by the building structure digital twin model, and the model output is the damage rate distribution characteristic coefficient of each monitoring point in the prediction period.

[0013] Preferably, the damage distribution feature mode is obtained in real time by a sensor network deployed on the building structure, specifically at least one of a structural strain mode, a vibration mode or an acoustic emission mode;

[0014] The nonlinear combination is a tensor product operation of the damage distribution feature mode vector and the damage rate distribution characteristic coefficient vector to generate the target damage rate distribution matrix.

[0015] Preferably, the second performance index distribution matrix of the current period is calculated according to the first performance index distribution matrix and the target damage rate distribution matrix, comprising:

[0016] An environmental mutation factor is calculated based on the Euclidean distance between the current environment parameter vector and the historical environment mean value;

[0017] The damage rate weight coefficient is dynamically adjusted according to the environmental mutation factor;

[0018] The current damage distribution feature mode is input into the building structure digital twin model, and a simulation performance index matrix is output;

[0019] An error value of the first performance index distribution matrix and the simulation matrix is calculated, and a performance decay coefficient is adjusted according to the error value;

[0020] The second performance index distribution matrix is calculated according to the adjusted damage rate weight coefficient and the performance decay coefficient.

[0021] In a second aspect, the application further provides a building structure safety monitoring system based on a digital twin model, comprising:

[0022] A parameter acquisition unit is configured to acquire target environment parameters and target monitoring point parameters of a building structure, and input the parameters into a damage feature prediction model to obtain damage rate distribution characteristic coefficients of the building structure in different periods;

[0023] A nonlinear combination unit is configured to monitor the damage distribution feature mode of the building structure, and perform nonlinear combination of the damage distribution feature mode and the damage rate distribution characteristic coefficient to generate a target damage rate distribution matrix;

[0024] a matrix calculation unit, configured to obtain a first performance index distribution matrix of the building structure in a previous monitoring period, and calculate a second performance index distribution matrix of a current period according to the first performance index distribution matrix and a target damage rate distribution matrix;

[0025] an abnormality early warning unit, configured to determine whether a value of any monitoring point position in the second performance index distribution matrix exceeds a structural failure threshold value of the position, and when the value exceeds the structural failure threshold value, mark the position as an abnormal monitoring point and trigger a multi-level early warning, diagnose a damage type of the abnormal monitoring point, and match a corresponding repair strategy according to different damage types.

[0026] Preferably, the damage feature prediction model is a deep learning-based time series prediction model.

[0027] The target environmental parameters at least include environmental temperature, humidity, wind speed and peak ground acceleration.

[0028] The target monitoring point parameters at least include historical strain data, displacement data and acceleration data of the monitoring point.

[0029] The damage feature prediction model is trained by historical data and simulation data generated by a building structure digital twin model, and the model output is a damage rate distribution feature coefficient of each monitoring point in a prediction period.

[0030] Preferably, the damage distribution feature mode is obtained in real time by a sensor network deployed on the building structure, and specifically at least one of a structural strain mode, a vibration mode or an acoustic emission mode.

[0031] The nonlinear combination is a tensor product operation of the damage distribution feature mode vector and the damage rate distribution feature coefficient vector to generate the target damage rate distribution matrix.

[0032] Preferably, the matrix calculation unit is further configured to:

[0033] calculate an environmental mutation factor based on an Euclidean distance between the current environmental parameter vector and a historical environmental mean value.

[0034] dynamically adjust a damage rate weight coefficient according to the environmental mutation factor.

[0035] input the current damage distribution feature mode into the building structure digital twin model to output a simulation performance index matrix.

[0036] calculate an error value of the first performance index distribution matrix and the simulation matrix, and adjust a performance decay coefficient according to the error value.

[0037] calculate the second performance index distribution matrix according to the adjusted damage rate weight coefficient and the performance decay coefficient.

[0038] In a third aspect, the present application also provides an electronic device, comprising a processor and a memory, the memory being configured to store computer program codes, the computer program codes comprising computer instructions, when the processor executes the computer instructions, the electronic device executes the method according to the first aspect and any possible implementation manner thereof.

[0039] In a fourth aspect, the present application also provides a computer readable storage medium, the computer readable storage medium storing a computer program, the computer program comprising program instructions, when the program instructions are executed by a processor of an electronic device, the processor executes the method according to the first aspect and any possible implementation manner thereof.

[0040] Compared with the prior art, the present application has the following beneficial effects:

[0041] The present application can dynamically fuse environmental factors and structural monitoring data by acquiring target environmental parameters and target monitoring point parameters and inputting the damage feature prediction model, realize fine simulation of the damage evolution process of the building structure, and realize accurate prediction of the damage rate distribution at different time periods; by nonlinearly combining the damage distribution feature mode and the damage rate distribution feature coefficient to generate a target damage rate distribution matrix, the defect that linearization processing cannot reflect the nonlinear damage accumulation of the structure is overcome, and the accuracy of damage rate calculation is improved; based on the last period performance index matrix and the target damage rate matrix, the current performance index is calculated, combined with the hierarchical early warning mechanism, the hysteresis and false alarm problems of single threshold early warning are solved, and dynamic tracking and accurate early warning of the safety state are realized; by diagnosing the damage type of the abnormal monitoring point and matching the repair strategy, the efficiency and accuracy of structure maintenance are greatly improved, and finally the service life of the building structure is prolonged and its operation safety is ensured.

[0042] It should be understood that the above general description and the following detailed description are only exemplary and explanatory, but not limiting the present disclosure. BRIEF DESCRIPTION OF DRAWINGS

[0043] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the background art, the drawings needed to be used in the embodiments of the present application or the background art will be described below.

[0044] The drawings herein are incorporated into the specification and form part of the specification, which show embodiments consistent with the present disclosure, and together with the specification, serve to explain the technical solutions of the present disclosure.

[0045] Figure 1 A flowchart of a building structure safety monitoring method based on a digital twin model provided by the embodiments of the present application is shown in the figure.

[0046] Figure 2A structural schematic diagram of a building structure safety monitoring system based on a digital twin model is provided for an embodiment of the present application. DETAILED DESCRIPTION

[0047] In order for those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0048] The terms "first", "second", and the like in the specification and claims of the present application and the above-mentioned drawings are used to distinguish different objects, and are not used to describe a specific order. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but can optionally include steps or units not listed, or can optionally include other steps or units inherent to the process, method, product or device.

[0049] Reference herein to "embodiments" means that the particular features, structures, or characteristics described in connection with the embodiments can be included in at least one embodiment of the present application. The appearance of the phrase in various places in the specification does not necessarily all refer to the same embodiment, nor does it necessarily refer to a separate or alternative embodiment in isolation from other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0050] Please refer to Figure 1 , Figure 1 A flowchart of a building structure safety monitoring method based on a digital twin model is provided for an embodiment of the present application. As shown in Figure 1 , the method comprises:

[0051] S10, obtaining target environment parameters and target monitoring point parameters of the building structure, inputting the damage characteristic prediction model to obtain damage rate distribution characteristic coefficients of the building structure at different time periods;

[0052] The damage feature prediction model is a time series prediction model fusing environmental factors and structural response data, which is based on a deep learning algorithm and can learn the nonlinear correlation between parameters and damage evolution law. By inputting real-time collected environmental parameters and monitoring point parameters, the damage development trend of the building structure at different future time periods can be predicted, thereby providing a quantitative basis for subsequent safety assessment. The damage rate distribution characteristic coefficient is a quantitative index for describing the damage development degree of each monitoring point of the building structure per unit time, which is a matrix corresponding to the spatial distribution of the monitoring points, for example, an n x m matrix, n is the number of monitoring points, and m is the number of prediction periods. The larger the value is, the faster the damage develops. The coefficient can intuitively reflect the spatial distribution difference and time evolution law of structural damage, thereby providing parameter basis for subsequent calculation of the target damage rate distribution matrix.

[0053] In one embodiment, the damage feature prediction model is a time series prediction model based on deep learning.

[0054] The target environmental parameters at least include environmental temperature, humidity, wind speed and peak ground acceleration.

[0055] The target monitoring point parameters at least include historical strain data, displacement data and acceleration data of the monitoring points.

[0056] The damage feature prediction model is trained by historical data and simulation data generated by the building structure digital twin model, and the model output is the damage rate distribution characteristic coefficient of each monitoring point at the prediction period.

[0057] At present, in the process of building structure safety monitoring, only a single environmental factor is usually focused on, such as only temperature or only earthquake, and the coupling effect of multiple environmental parameters is ignored; the environmental parameters are usually regarded as “static constants”, and the influence of dynamic time series characteristics such as random fluctuation of wind speed and diurnal periodicity of temperature on damage accumulation is not considered, resulting in deviation of damage prediction results from the actual situation. As for the monitoring points, static parameters (such as displacement) are focused on and dynamic parameters (such as acceleration) are ignored, which is difficult to capture early stiffness damage of the structure; only current data is collected and historical time series data is ignored, which cannot trace the “causal relationship” of damage evolution, and damage identification has a lag, which often only warns after obvious deformation or cracks appear in the structure, missing the early repair opportunity; at the same time, due to the lack of historical data support, it is difficult to distinguish between “normal deformation” and “damage deformation”. Therefore, in the present embodiment, multiple aspects are considered for environmental parameters and monitoring point parameters, thereby avoiding the problem of low prediction accuracy caused by considering only a single dimension in the prior art.

[0058] In building structures, temperature changes cause thermal expansion and contraction of building materials such as concrete and steel. Long-term repeated temperature stresses can cause structural cracks, loose connections, and other damage. Extreme high or low temperatures can also accelerate material aging. Temperature fluctuations in different climate zones have a significant impact on structural damage, and temperature parameters are needed to quantify this long-term cumulative effect. High humidity environments can exacerbate chemical damage such as steel corrosion in concrete, wood decay, and steel structure coating peeling. Humidity and temperature coupling can also cause freeze-thaw cycles, causing internal pore water in concrete to freeze and expand, leading to material cracking. Humidity is a key factor affecting structural durability damage, especially for underground structures, coastal buildings, and other humid environments. Wind load is the main dynamic load for high-rise buildings and long-span bridges. Long-term wind-induced vibration can cause structural component fatigue damage such as bolt loosening and support wear. Strong winds can cause instantaneous structural deformation to exceed limits and even cause local damage. Wind speed, direction, and timing characteristics directly determine the distribution and effect of wind loads, which need to be captured through real-time monitoring to capture their impact on structural dynamic response. Earthquakes are the main factor causing sudden structural damage. Peak ground acceleration directly reflects the strength of seismic action and determines whether the structure will experience plastic deformation, component yield, or even collapse. Even in non-seismic areas, small earthquakes or distant earthquakes can accumulate structural damage, and this parameter is needed to evaluate the instantaneous impact and time effect of seismic loads on structures.

[0059] The core basis for selecting historical strain data, displacement data, and acceleration data from monitoring points is that these parameters directly reflect the mechanical response state of the structure and are the "direct signals" of damage occurrence. Strain is a direct reflection of the internal stress state of structural components. Abnormal strain, such as sudden increases in local strain and long-term cumulative strain exceeding limits, often indicates component cracking, cross-section weakening, or material yield. Historical strain data can be used to trace the timing changes in structural stress and identify the initiation and development of damage. Displacement is a direct reflection of the overall or local deformation of the structure. Excessive relative displacement, such as beam-column joint displacement and foundation settlement difference, can cause structural stiffness degradation, load transfer path changes, and even trigger chain damage. Displacement data can effectively evaluate the stability of the structure and the performance of component connections. Acceleration reflects the vibration response of the structure under dynamic loads. Changes in vibration frequency and amplitude can identify structural stiffness damage. Dynamic response data are more sensitive to early damage and are key indicators for damage warning.

[0060] In one embodiment, data collection is achieved through environmental sensors deployed around the building structure and in key areas, including:

[0061] Temperature sensor: Real-time collection of structure surface and surrounding environment temperature, such as sampling frequency 1 / min;

[0062] Humidity sensor: monitors the relative humidity of the air and the surface humidity of the structure, such as sampling frequency 1 / min;

[0063] Wind speed sensor: installed on the top of the building or the windward surface, collecting instantaneous wind speed and direction, such as sampling frequency 10 / s;

[0064] Seismic motion sensor: arranged at the structure foundation and key floors, collecting peak ground acceleration, such as sampling frequency 100Hz.

[0065] Target monitoring point parameter collection: realized by attaching structure sensors to key monitoring points of the building structure, including beam-column joints, load-bearing walls, foundations, etc., including:

[0066] Strain gauge, fiber optic sensor: collect strain data of monitoring points, such as sampling frequency 10Hz;

[0067] Displacement meter: measures the relative displacement or absolute displacement of the monitoring point, such as sampling frequency 1Hz;

[0068] Acceleration sensor: records the vibration acceleration of the monitoring point, such as sampling frequency 50Hz.

[0069] After collecting data, data processing is performed first, including removing abnormal values caused by sensor failure, smoothing high-frequency noise by moving average method; unify environmental parameters and monitoring point parameters to the same time scale to ensure time consistency; extract statistical features such as daily average temperature, maximum wind speed, and humidity fluctuation amplitude from environmental parameters; calculate derived indicators such as strain cumulative value, displacement change rate, and vibration frequency spectrum characteristics from monitoring point parameters; map all parameters to the [0, 1] interval to avoid interference of different orders of magnitude data on model training.

[0070] When training the model, a deep learning-based time series prediction model is adopted, which has the core advantages of capturing long-term and short-term dependencies between parameters and nonlinear mapping rules. For the above-processed data, the damage record is quantized as a "damage rate distribution feature coefficient" as the model output label; the data is divided into training set, validation set and test set in the ratio of 7:2:1 to ensure consistent data distribution. The input is the time series sequence of the processed environmental parameters and monitoring point parameters (such as the past 24 hours of data), and the output is the damage rate distribution feature coefficient in the future period; the mean square error is combined with the structural damage physical constraints such as non-negativity of damage rate to ensure that the prediction result conforms to the engineering practice; the model parameters are iteratively optimized by the back propagation algorithm to prevent overfitting, and the model weight with the best performance in the validation set is finally saved. In order to obtain the damage rate distribution feature coefficient in different periods, the prediction period granularity is set, the model is used for multi-step time series prediction, and the damage rate distribution feature coefficient in the future period is output; combined with the digital twin model of the building structure, the prediction result is physically consistent, such as excluding abnormal values that do not meet the material mechanics characteristics, and finally generating the damage rate distribution feature coefficient matrix of each monitoring point in different periods.

[0071] S20, monitoring the damage distribution feature mode of the building structure, and performing nonlinear combination on the damage distribution feature mode and the damage rate distribution feature coefficient to generate a target damage rate distribution matrix;

[0072] Preferably, the damage distribution feature mode is obtained in real time by a sensor network deployed on the building structure, specifically at least one of a structural strain mode, a vibration mode or an acoustic emission mode;

[0073] The nonlinear combination is a tensor product operation of the damage distribution feature mode vector and the damage rate distribution feature coefficient vector to generate the target damage rate distribution matrix.

[0074] In this embodiment, sensor data is collected in real time, and the sampling frequency is dynamically adjusted according to the mode type, for example, 10 Hz for strain mode, 100 Hz for vibration mode, and 1 kHz for acoustic emission mode; modal features are extracted by Fourier transform, wavelet decomposition and other methods, which are converted into standardized vectors, such as vectors with dimension n, n is the number of monitoring points, and each element in the vector corresponds to the modal feature value of a monitoring point, such as strain amplitude, vibration displacement and acoustic emission signal energy.

[0075] Further, the damage distribution characteristic mode and the damage rate distribution characteristic coefficient are combined non-linearly, that is, a target damage rate distribution matrix is obtained through tensor product operation, each element in the matrix represents the coupling effect of the modal characteristic and the damage rate coefficient of the monitoring point, and reflects the correlation of structural damage in space. Then, the matrix is modified in combination with the physical constraints of the building structure digital twin model, such as material mechanical properties and structural topological relationship. For example, the elements beyond the physical reasonable range, such as abnormal values caused by sensor noise, are smoothed; and the coupling terms of the non-connected regions of the structure, such as independent components, are given decay coefficients to weaken the influence of elements without actual mechanical correlation. The finally generated target damage rate distribution matrix contains two layers of core information: the diagonal elements reflect the direct coupling effect of the “modal characteristic-damage rate” of a single monitoring point, and quantify the absolute speed of the current damage development of the point; and the non-diagonal elements reflect the damage transferability between different monitoring points, and quantify the spatial diffusion effect of structural damage. Through tensor product, the non-linear combination of modal characteristics and damage rate is realized, which breaks through the limitations of the traditional linear superposition method, and can accurately depict the nonlinear evolution law from structural modal anomaly to damage acceleration, from local damage to overall diffusion, and the like.

[0076] S30, obtaining a first performance index distribution matrix of the building structure obtained in a previous monitoring period, and calculating a second performance index distribution matrix of a current time period according to the first performance index distribution matrix and the target damage rate distribution matrix;

[0077] Preferably, the calculation of the second performance index distribution matrix of the current time period according to the first performance index distribution matrix and the target damage rate distribution matrix comprises:

[0078] An environmental mutation factor is calculated based on the Euclidean distance between the current environmental parameter vector and the historical environmental mean value.

[0079] The damage rate weight coefficient is dynamically adjusted according to the environmental mutation factor.

[0080] The current damage distribution characteristic mode is input into the building structure digital twin model, and a simulation performance index matrix is output.

[0081] An error value of the first performance index distribution matrix and the simulation matrix is calculated, and the performance decay coefficient is adjusted according to the error value.

[0082] The second performance index distribution matrix is calculated according to the adjusted damage rate weight coefficient and the performance decay coefficient.

[0083] In the embodiment, the environmental parameters of the past monitoring periods are extracted from the historical database of the building structure digital twin model, the mean value vector of each parameter is calculated, the Euclidean distance between the current environmental vector and the historical mean value vector is calculated,

[0084] ;

[0085] wherein, is the Euclidean distance, is the historical mean vector, is the current environment vector; based on distance define the environment mutation factor :

[0086] ;

[0087] wherein, is the preset environment sensitivity coefficient, is the threshold value, for example, 1.5 times the historical distance mean, is a natural constant, the value range is [0, 1], the greater the value, the more significant the environment mutation, such as extreme weather, sudden earthquake.

[0088] Then, according to the environment mutation factor, the damage rate weight coefficient is dynamically adjusted:

[0089] ;

[0090] wherein, is the dynamically adjusted damage rate weight coefficient, is the reference value of the coefficient;

[0091] calculate the error value of the first performance index distribution matrix and the simulation matrix :

[0092] ;

[0093] wherein, are the first performance index distribution matrix and the simulation matrix respectively, is the Frobenius norm; according to the error value, adjust the performance decay coefficient :

[0094] ;

[0095] wherein, is the reference value of the performance decay coefficient, is the error threshold value.

[0096] According to the adjusted damage rate weight coefficient and the performance decay coefficient, the second performance index distribution matrix is calculated:

[0097] ;

[0098] wherein, is the second performance index distribution matrix, A target damage rate distribution matrix is generated.

[0099] The second performance index distribution matrix generated by fusing historical performance data, current damage rate, environmental mutation influence and digital twin simulation results can accurately reflect the dynamic evolution of the structural performance, and not only retains the coherence of historical trends, but also reflects the immediate influence of real-time damage.

[0100] In S40, it is judged whether the value of any monitoring point position in the second performance index distribution matrix exceeds the structural failure threshold value of the position, and when exceeding, the position is marked as an abnormal monitoring point and a multi-level early warning is triggered; the damage type of the abnormal monitoring point is diagnosed, and a corresponding repair strategy is matched according to different damage types.

[0101] Specifically, the values of the monitoring points in the second performance index distribution matrix are compared with the preset structural failure threshold value one by one, and when the value of a certain monitoring point exceeds the failure threshold value, the monitoring point is immediately marked as an abnormal monitoring point. According to the number, position and performance index attenuation amplitude of the abnormal monitoring point, a multi-level early warning is triggered, such as a first-level early warning prompting attention, a second-level early warning starting inspection, and a third-level early warning emergency treatment. At the same time, combined with the damage feature database and the historical repair cases, the damage type of the abnormal monitoring point is diagnosed through a machine learning algorithm, including but not limited to cracks, deformation, material aging, etc., and a corresponding technical solution is matched from the repair strategy library, such as reinforcement method, material replacement, construction process, etc. In this way, timely identification and hierarchical response of structural safety risk can be realized, and the pertinence and efficiency of safety control can be improved; through accurate matching of repair strategies, scientific guidance can be provided for structural maintenance, repair costs can be reduced, and repair effects can be guaranteed.

[0102] To sum up, the safety monitoring method provided by the embodiment of the application can dynamically fuse environmental factors and structural monitoring data by acquiring target environmental parameters and target monitoring point parameters and inputting them into a damage feature prediction model, realize fine simulation of the damage evolution process of the building structure, and realize accurate prediction of the damage rate distribution in different periods. The target damage rate distribution matrix is generated by nonlinearly combining the damage distribution feature mode and the damage rate distribution feature coefficient, which overcomes the defect that linearization processing cannot reflect the nonlinear damage accumulation of the structure, and improves the accuracy of damage rate calculation. Based on the last period performance index matrix and the target damage rate matrix, the current performance index is calculated, and combined with the hierarchical early warning mechanism, the hysteresis and false alarm problems of single threshold early warning are solved, and dynamic tracking and accurate early warning of the safety state are realized. By diagnosing the damage type of the abnormal monitoring point and matching the repair strategy, the efficiency and accuracy of structural maintenance are greatly improved, the service life of the building structure is ultimately prolonged, and the operation safety is guaranteed.

[0103] Referring to Figure 2 The application further provides a building structure safety monitoring system based on a digital twin model, which comprises:

[0104] The parameter acquisition unit 100 is configured to acquire target environment parameters and target monitoring point parameters of the building structure, and input the damage characteristic prediction model to obtain damage rate distribution characteristic coefficients of the building structure in different time periods;

[0105] The nonlinear combination unit 200 is configured to monitor damage distribution characteristic modes of the building structure, and perform nonlinear combination on the damage distribution characteristic modes and the damage rate distribution characteristic coefficients to generate a target damage rate distribution matrix;

[0106] The matrix calculation unit 300 is configured to acquire a first performance index distribution matrix of the building structure obtained in a previous monitoring period, and calculate a second performance index distribution matrix of a current time period according to the first performance index distribution matrix and the target damage rate distribution matrix;

[0107] The abnormal early warning unit 400 is configured to judge whether a value of any monitoring point position in the second performance index distribution matrix exceeds a structure failure threshold value of the position, and when the value exceeds the structure failure threshold value, mark the position as an abnormal monitoring point and trigger a multi-level early warning; diagnose a damage type of the abnormal monitoring point, and match a corresponding repair strategy according to different damage types.

[0108] In one embodiment, the damage characteristic prediction model is a deep learning-based time series prediction model;

[0109] The target environment parameters at least include environmental temperature, humidity, wind speed and peak ground acceleration;

[0110] The target monitoring point parameters at least include historical strain data, displacement data and acceleration data of the monitoring point;

[0111] The damage characteristic prediction model is trained by historical data and simulation data generated by a building structure digital twin model, and the model output is damage rate distribution characteristic coefficients of each monitoring point in a prediction time period.

[0112] In one embodiment, the damage distribution characteristic modes are acquired in real time by a sensor network deployed on the building structure, and specifically at least one of a structure strain mode, a vibration mode or an acoustic emission mode;

[0113] The nonlinear combination is a tensor product operation on the damage distribution characteristic mode vector and the damage rate distribution characteristic coefficient vector to generate the target damage rate distribution matrix.

[0114] In one embodiment, the matrix calculation unit 300 is further configured to:

[0115] Calculate an environmental mutation factor based on an Euclidean distance between a current environment parameter vector and a historical environment mean value;

[0116] The damage rate weight coefficient is dynamically adjusted according to environmental mutation factors;

[0117] The current damage distribution feature mode is input into the building structure digital twin model, and a simulation performance index matrix is output.

[0118] The error value of the first performance index distribution matrix and the simulation matrix is calculated, and the performance attenuation coefficient is adjusted according to the error value.

[0119] The second performance index distribution matrix is calculated according to the adjusted damage rate weight coefficient and the performance attenuation coefficient.

[0120] It can be understood that the system provided by the embodiment has functions or contains modules that can be used to execute the method described in the above method embodiment, and the specific implementation can refer to the description of the above method embodiment.

[0121] The application also provides an electronic device, including a processor and a memory, the memory is used for storing computer program code, the computer program code includes computer instructions, when the processor executes the computer instructions, the electronic device executes the method of any one of the above possible implementations.

[0122] The application also provides a computer readable storage medium, the computer readable storage medium stores a computer program, the computer program includes program instructions, the program instructions are executed by the processor of the electronic device, and the processor executes the method of any one of the above possible implementations.

[0123] Those skilled in the art can realize that the units and algorithm steps of the examples described in combination with the embodiments disclosed herein can be realized by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are realized in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to realize the described functions for each specific application, but such implementation should not be considered beyond the scope of the application.

[0124] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the above-described system, device and unit can refer to the corresponding process in the foregoing method embodiments, which will not be repeated here. Those skilled in the art can also clearly understand that each embodiment of the application describes different focuses, and for the convenience and brevity of description, the same or similar parts may not be described in different embodiments. Therefore, the parts not described or not described in detail in a certain embodiment can refer to the description of other embodiments.

Claims

1. A method for monitoring the safety of building structures based on a digital twin model, characterized in that, The method comprises: obtaining target environmental parameters and target monitoring point parameters of the building structure, inputting the damage characteristic prediction model to obtain damage rate distribution characteristic coefficients of the building structure at different time periods; monitoring damage distribution characteristic modes of the building structure, and performing nonlinear combination on the damage distribution characteristic modes and the damage rate distribution characteristic coefficients to generate a target damage rate distribution matrix; obtaining a first performance index distribution matrix of the building structure obtained in a previous monitoring period, and calculating a second performance index distribution matrix of a current time period according to the first performance index distribution matrix and the target damage rate distribution matrix, comprising: calculating an environmental mutation factor based on the Euclidean distance between the current environmental parameter vector and the historical environmental mean value; dynamically adjusting a damage rate weight coefficient according to the environmental mutation factor; inputting the current damage distribution characteristic mode into a building structure digital twin model to output a simulation performance index matrix; calculating an error value of the first performance index distribution matrix and the simulation matrix, and adjusting a performance attenuation coefficient according to the error value; calculating the second performance index distribution matrix according to the adjusted damage rate weight coefficient and the performance attenuation coefficient; judging whether the value of any monitoring point position in the second performance index distribution matrix exceeds the structural failure threshold value of the position, marking as an abnormal monitoring point and triggering a multi-level early warning when exceeding; diagnosing the damage type of the abnormal monitoring point, and matching a corresponding repair strategy according to different damage types. 2.The building structure safety monitoring method based on a digital twin model according to claim 1, wherein, The damage characteristic prediction model is a deep learning-based time series prediction model; The target environmental parameters at least include environmental temperature, humidity, wind speed and peak ground acceleration; The target monitoring point parameters at least include historical strain data, displacement data and acceleration data of the monitoring point; The damage characteristic prediction model is trained by historical data and simulation data generated by the building structure digital twin model, and the model output is the damage rate distribution characteristic coefficient of each monitoring point at the prediction time period. 3.The building structure safety monitoring method based on digital twin model according to claim 1, wherein, The damage distribution characteristic mode is obtained in real time by a sensor network deployed on the building structure, specifically at least one of a structural strain mode, a vibration mode or an acoustic emission mode; The nonlinear combination is a tensor product operation on the damage distribution characteristic mode vector and the damage rate distribution characteristic coefficient vector to generate the target damage rate distribution matrix. 4.A building structure safety monitoring system based on a digital twin model, characterized by, The system comprises: a parameter acquisition unit configured to obtain target environmental parameters and target monitoring point parameters of the building structure, input the damage characteristic prediction model to obtain damage rate distribution characteristic coefficients of the building structure at different time periods; a nonlinear combination unit configured to monitor damage distribution characteristic modes of the building structure, and perform nonlinear combination on the damage distribution characteristic modes and the damage rate distribution characteristic coefficients to generate a target damage rate distribution matrix; a matrix calculation unit configured to obtain a first performance index distribution matrix of the building structure obtained in a previous monitoring period, and calculate a second performance index distribution matrix of a current time period according to the first performance index distribution matrix and the target damage rate distribution matrix, comprising: calculating an environmental mutation factor based on the Euclidean distance between the current environmental parameter vector and the historical environmental mean value; dynamically adjusting a damage rate weight coefficient according to the environmental mutation factor; Input the current damage distribution feature mode into the building structure digital twin model, and output a simulated performance index matrix; Calculate the error value of the first performance index distribution matrix and the simulated matrix, and adjust the performance decay coefficient according to the error value; According to the adjusted damage rate weight coefficient and the performance decay coefficient, a second performance index distribution matrix is calculated; The abnormal early warning unit is used to judge whether the value of any monitoring point position in the second performance index distribution matrix exceeds the structure failure threshold of the position, and when it exceeds, it is marked as an abnormal monitoring point and triggers a multi-level warning; diagnose the damage type of the abnormal monitoring point, and match the corresponding repair strategy according to different damage types.

5. The building structure safety monitoring system based on digital twin model according to claim 4, characterized in that, The damage feature prediction model is a deep learning-based time series prediction model; The target environmental parameters at least include environmental temperature, humidity, wind speed and peak ground acceleration; The target monitoring point parameters at least include historical strain data, displacement data and acceleration data of the monitoring point; The damage feature prediction model is trained by historical data and simulation data generated by the building structure digital twin model, and the model output is the damage rate distribution feature coefficient of each monitoring point in the prediction period. 6.The building structure safety monitoring system based on digital twin model according to claim 4, wherein, The damage distribution feature mode is obtained in real time by a sensor network deployed on the building structure, specifically at least one of the structure strain mode, vibration mode or acoustic emission mode; The nonlinear combination is a tensor product operation of the damage distribution feature mode vector and the damage rate distribution feature coefficient vector to generate the target damage rate distribution matrix.

7. An electronic device, comprising: It comprises: A processor and a memory for storing computer program code, the computer program code comprising computer instructions, when the processor executes the computer instructions, the electronic device executes the building structure safety monitoring method based on the digital twin model as claimed in any one of claims 1 to 3.

8. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a computer program, the computer program includes program instructions, the program instructions are executed by the processor of the electronic device, and the processor executes the building structure safety monitoring method based on the digital twin model as claimed in any one of claims 1 to 3.

Images