Building risk monitoring and evaluation method and system

Abstract

The invention relates to a building risk monitoring and evaluating method and system, and the method comprises the steps: collecting the full life cycle data of a building, carrying out the preprocessing, extracting the failure mode of the building and key impact factors, designing an orthogonal test to simulate the accelerated aging process of the building, and determining the critical state of the accelerated failure. Selecting a dynamic covariable which is closely related to a failure mechanism and changes along with time from test conditions of an equivalent acceleration failure critical state, and constructing a risk regression model according to dynamic covariable data; and inputting the real-time dynamic covariable into the risk regression model, and calculating the relative failure risk of the current building. According to the method, the accelerated aging process of the building is simulated through the orthogonal test, the critical state of accelerated failure is accurately determined, observation of the aging process is greatly accelerated, interaction among different factors is comprehensively revealed, the relative failure risk of the building can be accurately quantified and evaluated in combination with the risk evaluation model, and the risk assessment efficiency is improved. And the real-time performance and the accuracy of the evaluation result are ensured.

Description

Technical Field

[0001] The present invention belongs to the technical field of building risk assessment, and in particular relates to a monitoring and assessment method and system for building risk. Background Art

[0002] With the acceleration of urbanization and the continuous development of construction technology, the complexity and diversity of building structures are increasing, and higher requirements are also put forward for the monitoring and assessment of building failure risks. Building failure may not only cause casualties and property losses, but also have a significant impact on the surrounding environment and society. Therefore, the development of accurate, efficient and reliable building failure risk monitoring and assessment technology is of great significance to ensure building safety.

[0003] In the construction industry, the main failure risk monitoring and assessment technologies currently used include structural monitoring technologies such as vibration monitoring, displacement monitoring and strain monitoring, as well as risk assessment technologies based on risk matrix, fault tree and event tree analysis. These technologies monitor and assess the status of building structures in real time through the steps of installing sensors, collecting data, and analyzing data, so as to discover potential safety hazards and take corresponding measures.

[0004] Although traditional technologies have played an important role in building failure risk monitoring and assessment, they still have some defects. The main defects are as follows: 1. Traditional technologies can often only provide static assessment results, and cannot provide real-time risk monitoring and early warning mechanisms, making it difficult to promptly detect and respond to potential safety hazards; 2. Traditional technologies lack standardized and standardized processes and indicator systems, resulting in insufficient comparability and credibility of assessment results; 3. Traditional technologies often rely on expert subjective judgment and personal experience, and there is a certain degree of uncertainty in the accuracy and objectivity of the assessment results. Summary of the invention

[0005] The present invention provides a method and system for monitoring and evaluating building risks, aiming to solve the problems mentioned in the above background technology.

[0006] The present invention is implemented by providing a method for monitoring and evaluating building risks, the steps comprising:

[0007] Collect data from the entire life cycle of the building and perform pre-processing;

[0008] Based on the pre-processed building life cycle data, analyze and extract the building failure mode and its key influencing factors;

[0009] According to the extracted failure modes and key influencing factors of the building, an orthogonal test is designed to simulate the accelerated aging process of the building and determine the critical state of accelerated failure;

[0010] Select dynamic covariates that are closely related to the failure mechanism and change over time from the test conditions of the equivalent accelerated failure critical state, and build a risk regression model based on the dynamic covariate data;

[0011] Through building monitoring equipment, real-time dynamic covariate data related to building performance is obtained, and the real-time dynamic covariate data is input into the constructed risk regression model to calculate and output the relative failure risk value of the current building. If the relative failure risk value of the current building exceeds the preset safety threshold, a risk warning is issued.

[0012] Furthermore, the step of inputting the real-time dynamic covariate data into the constructed risk regression model to calculate and output the relative risk value of the current building includes:

[0013] The dynamic covariate data obtained by real-time monitoring are organized into a vector form to obtain a real-time dynamic covariate vector;

[0014] The real-time dynamic covariate vector is input into the constructed risk regression model. Based on the real-time dynamic covariate vector, the instantaneous failure risk of the current building is calculated. The calculation formula is:

[0015]

[0016] Among them, h(t 当前 ∣Z 实时 ) represents the real-time observation value Z at the current time t and the dynamic covariate vector 实时 Under this condition, the instantaneous failure risk of the building, h0(t 当前 ) is the baseline risk, which refers to the instantaneous risk of building failure at the current time t when there is no covariate effect, that is, when all covariates are 0. B is the covariate coefficient vector, which indicates the influence of each dynamic covariate on the failure risk. B = {β1, β2, …, β p}, Z 实时 is the real-time observation value of the covariate vector;

[0017] Based on the instantaneous failure risk and benchmark risk of the current building, the relative risk of the current building is calculated to obtain the relative risk value d of the current building.

[0018] Furthermore, the step of selecting a dynamic covariate that changes with time and is closely related to the failure mechanism from the test conditions of the equivalent accelerated failure critical state, and constructing a risk regression model based on the dynamic covariate data includes:

[0019] Selecting time-varying dynamic covariates that are closely related to the failure mechanism from the test conditions of the equivalent accelerated failure critical state, collecting the dynamic covariate data of each test sample and the corresponding failure time, and generating a dynamic covariate vector for each sample to construct a risk set, wherein each sample in the risk set contains a set of time-varying dynamic covariate values ​​and a corresponding failure time;

[0020] A risk regression model is constructed. The risk regression model is used to evaluate the impact of covariates on failure risk. The failure risk assessment formula of the model is as follows:

[0021]

[0022] Where h(t|X) represents the failure risk function value of the building at time t when the covariate vector is Z, h0(t) is the baseline risk function, which refers to the instantaneous risk of building failure at time t when there is no covariate effect, that is, when all covariates are 0, β1, β2, …, β p are model parameters, i.e., the coefficients of each covariate, which represent the influence of each dynamic covariate on the failure risk, z1, z2, …, z p is the value of each dynamic covariate in a set of dynamic covariate vectors Z in the risk set;

[0023] Substitute the dynamic covariate vector in the risk concentration into the risk regression model, and use maximum likelihood estimation to estimate the covariate coefficient vector B, B = {β1, β2, ..., β p}, a parameter-determined risk regression model is obtained to predict the building failure risk under given covariate values.

[0024] Furthermore, the step of substituting the dynamic covariate vector in the risk set into the risk regression model and estimating the covariate coefficient vector B by using maximum likelihood estimation includes:

[0025] The partial likelihood function is used to estimate the dynamic covariate coefficient vector B. The partial likelihood function formula is:

[0026]

[0027] The entire partial likelihood function formula is a product that only includes those observations for which events actually occurred, i.e., δ i = 1, the numerator Represents the i-th observation object in the dynamic covariate vector Z i and the covariate coefficient vector B, the denominator Indicates that at time t iThe sum of the risks of all the observations that have not yet experienced an event, that is, those that are still at risk. The denominator is to normalize all the observations that are still at risk to ensure that the value of the partial likelihood function is between 0 and 1. i is an event indicator. If the i-th observation object is at t i When an event occurs at time, i.e., building failure, then δ i =1, if the i-th observation object is at t i If the time is censored, that is, no failure is observed, then δ i =0, R(t i ) is at time t i The set of all observed objects that have not experienced any events before is the risk set; Z i and Z j Represent the covariate vectors of the i-th and j-th observations respectively;

[0028] The partial likelihood function L(B) is maximized by numerical optimization to obtain the estimated value of the covariate coefficient vector B, where B = {β1, β2, ..., β p}.

[0029] Furthermore, the step of maximizing the partial likelihood function L(B) by a numerical optimization method to obtain an estimated value of the covariate coefficient vector B includes:

[0030] Randomly select an initial covariate coefficient vector B0;

[0031] Calculate the gradient vector of the partial likelihood function L(B) with respect to the covariate coefficient vector B Among them, the gradient vector contains the direction information where the partial likelihood function value changes fastest in each parameter direction;

[0032] Use the optimization algorithm to iteratively update the value of the covariate coefficient vector B until the gradient is small enough. In each iteration, a new value of the covariate coefficient vector B is calculated based on the current value of the covariate coefficient vector B and the gradient information. The update formula is: Among them, α is the learning rate;

[0033] Output the estimated value of the final covariate coefficient vector B, where B = {β1, β2, ..., β p}.

[0034] Furthermore, the steps of analyzing and extracting the failure modes and key influencing factors of the building based on the pre-processed life cycle data of the building include:

[0035] Identify all events related to building failure from the building's full life cycle data and set them as failure events. Failure events include crack generation, structural deformation, and equipment failure.

[0036] Clustering algorithms are used to cluster failure events and group similar failure events together to identify different failure modes. The specific steps are as follows:

[0037] Construct a data set of failure events {x1, x2, ..., x n}, where each data point x i is a multidimensional vector that contains multiple features describing the failure event, that is, it contains the observed values ​​of each factor, expressed as x i ={F1(i), F2(i), ..., F m (i)}, m is the number of factors;

[0038] Run the clustering algorithm to divide the data set into K clusters. The objective function of the clustering algorithm is used to minimize the sum of squares of the data points in the cluster. The objective function of the clustering algorithm is:

[0039]

[0040] Among them, C k is the set of data points of the kth cluster, μ k is the centroid of the kth cluster, i.e., the mean vector;

[0041] Based on the characteristics of each cluster data point, K different failure modes of the building are defined;

[0042] Calculate the normalized difference ratio NDP for each data point using the formula:

[0043] Among them, NDP i is the normalized difference ratio of data point i, which belongs to failure mode k, F j (i) is the factor F in data point i j The observed value, y k is the factor F in failure mode k j The representative value of all data points in failure mode k is the factor F j The mean of

[0044] Calculate the factor F j The factor F is obtained by taking the average normalized difference ratio of failure mode k. j The influence degree on failure mode k is calculated as follows:

[0045] Among them, Indices k is the number of data points in failure mode k;

[0046] By analogy, the influence of each factor on each failure mode is calculated;

[0047] For each failure mode, factors with high influence are extracted as their key influencing factors.

[0048] Furthermore, the steps of designing an orthogonal test to simulate the accelerated aging process of the building based on the extracted failure mode and key influencing factors of the building and determining the critical state of accelerated failure include:

[0049] Based on the extracted key influencing factors, an orthogonal test scheme was designed, and the accelerated aging process of the building under different conditions was simulated by finite element method;

[0050] Carry out the test according to each test scheme condition in the orthogonal table and record the changes in building performance under different conditions;

[0051] Through test data, analyze the performance change trend of the building under different failure modes, and determine the critical state of accelerated failure under each failure mode. The critical state of accelerated failure means that when the performance of the building drops to a certain level, it is considered that the building has reached or is close to failure.

[0052] Compare the accelerated failure critical states obtained in the simulation with the failure states in the actual building;

[0053] When a simulated accelerated failure critical state consistent with the actual failure state is found, the test conditions represented by the orthogonal test scheme are determined to be accelerated test conditions equivalent to the actual service conditions.

[0054] The present invention also provides a building risk monitoring and assessment system for executing a building risk monitoring and assessment method, comprising:

[0055] Data collection module: used to collect data on the entire life cycle of a building and perform preprocessing;

[0056] Extraction module: used to analyze and extract the failure modes and key influencing factors of buildings based on the pre-processed life cycle data of buildings;

[0057] Orthogonal test module: used to design orthogonal tests to simulate the accelerated aging process of buildings and determine the critical state of accelerated failure based on the extracted failure modes and key influencing factors of the buildings;

[0058] Model building module: used to select dynamic covariates that vary with time and are closely related to the failure mechanism from the test conditions of the equivalent accelerated failure critical state, and to build a risk regression model based on the dynamic covariate data;

[0059] Risk prediction module: used to obtain real-time dynamic covariate data related to building performance through building monitoring equipment, input the real-time dynamic covariate data into the constructed risk regression model, calculate and output the relative failure risk value of the current building, and issue a risk warning if the relative failure risk value of the current building exceeds the preset safety threshold.

[0060] The present invention brings the following beneficial effects: the monitoring and assessment method of building risk provided by the present invention collects the data of the whole life cycle of the building, and after preprocessing, is used to analyze and accurately identify the failure mode of the building and its key influencing factors. By designing orthogonal experiments to simulate the accelerated aging process of the building, it can not only efficiently explore the interaction between different factors, but also accurately determine the critical state of accelerated failure, which is crucial for understanding and predicting the performance changes of the building in actual use. Dynamic covariates closely related to the failure mechanism are screened out from the test conditions. These variables can effectively reflect the changing trend of building performance and are key factors in assessing building risks. Based on these dynamic covariate data, a risk regression model is constructed. The constructed risk regression model not only considers the comprehensive effect of multiple influencing factors, but also considers the interaction between these factors. It can accurately reflect the complex relationship between building performance and failure risk, and quantitatively assess the relative failure risk of the building, thereby improving the accuracy of risk assessment. Through the data obtained by the real-time monitoring device, the model can dynamically update the risk value. Once the preset threshold is exceeded, the system will automatically issue a risk warning, thereby greatly improving the efficiency and effect of building risk management. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 It is a flow chart of a method for monitoring and assessing building risks provided by an embodiment of the present invention;

[0062] Figure 2 It is a system block diagram of a building risk monitoring and assessment system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0063] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0064] To facilitate understanding of this embodiment, a monitoring and assessment method for building risks disclosed in an embodiment of the present invention is first introduced in detail. Figure 1 A flow chart of a method for monitoring and assessing building risks provided by an embodiment of the present invention is shown. Figure 1 As shown, the method includes steps S101 to S105:

[0065] S101, collect the building’s full life cycle data and perform pre-processing.

[0066] Collect data from the entire life cycle of the building, including construction drawings in the design phase, progress reports in the construction phase, monitoring data such as temperature, humidity and stress in the use phase, and repair records in the maintenance phase. The pre-processing phase involves data cleaning, sorting, classification and standardization, cleaning out invalid and duplicate entries in the data, integrating data from different sources into a unified database, and performing standardization, such as standardizing temperature data to degrees Celsius.

[0067] S102, based on the pre-processed building life cycle data, analyze and extract the building failure mode and its key influencing factors.

[0068] After obtaining comprehensive and high-quality data, the data needs to be analyzed in depth to identify the potential failure modes of the building (such as structural cracking, aging of waterproof materials, etc.) and their key influencing factors (such as temperature, humidity, load, etc.). These analyses help to understand the causes and mechanisms of building performance degradation.

[0069] For example, an in-depth analysis of the pre-processed building data revealed that the main failure modes of the building were cracking of the concrete structure and aging of the waterproofing materials. Further analysis revealed that temperature fluctuations, humidity changes, and long-term loads were the key influencing factors of these failure modes.

[0070] S103, based on the extracted failure modes of the building and their key influencing factors, design an orthogonal test to simulate the accelerated aging process of the building and determine the critical state of accelerated failure.

[0071] In order to study the failure mechanism of buildings in more depth and predict their long-term performance, orthogonal tests are designed to simulate the accelerated aging process of buildings. In the simulation test, by controlling the changes in key influencing factors, the performance degradation of the building can be accelerated and failure phenomena can be observed in a shorter time. The critical state of accelerated failure is determined by experiments, indicating that the performance of the building has degraded to an unacceptable level. This step simulates the accelerated aging process of the building through orthogonal experiments, which can comprehensively consider the impact of multiple key influencing factors on the failure mode, thereby more accurately predicting the failure risk of the building in actual use. This not only helps to take preventive measures in the design stage to reduce the possibility of failure in the future, but also provides an important basis for formulating reasonable maintenance plans and ensuring the safety of buildings.

[0072] Orthogonal experimental design is an efficient experimental design method that can evaluate the impact of multiple factors and their interactions on system performance within a limited number of experiments. In the process of simulating accelerated aging of buildings, appropriate orthogonal tables can be selected and corresponding experimental schemes can be designed according to the number and level of key influencing factors. Each experimental scheme represents a specific combination of influencing factor levels, which is used to simulate different aging conditions. According to the designed orthogonal experimental scheme, accelerated aging simulation is implemented in the simulated environment. This usually involves adjusting environmental parameters (such as temperature, humidity, light, etc.) to accelerate the natural aging process of building materials. During the accelerated aging simulation process, the performance indicators of the building, such as structural strength, material properties, waterproofing effect, etc., are regularly evaluated. Based on the results of the performance evaluation, the data can be further analyzed to determine the critical state of accelerated failure of the building. This usually involves identifying the thresholds or inflection points of the performance indicators. When the performance indicators reach or exceed these thresholds, the building is considered to be close to or has reached the failure state. By comparing the changes in performance indicators under different experimental schemes, the key factor combinations and their critical levels that lead to accelerated failure can be determined.

[0073] For example, in the test, the accelerated aging process of the building in different environments was simulated by controlling the changes in temperature, humidity and load. After many tests, it was determined that when the cracking of the concrete structure reached a certain degree or the waterproof material was seriously aged, it was the critical state of accelerated failure.

[0074] Assume that the main failure mode of the building has been identified as the reduced durability of the concrete structure, and its key influencing factors include temperature (T), humidity (H) and the number of cyclic loads (L). In order to study the impact of these three factors on the durability of the concrete structure, a three-factor three-level orthogonal test is designed. Use the L9 (3X4) orthogonal table (where "L" represents the orthogonal table, "9" represents the number of tests, and "3X4" means that this is an orthogonal table with 4 columns, but only the first 3 columns are used in this example), as follows:

[0075] Experiment No. Temperature (T, °C) Humidity (H,%) Cyclic load times (L, times) 1 20 50 1000 2 40 60 2000 3 60 70 3000 4 20 60 2000 5 40 70 3000 6 60 50 1000 7 20 70 3000 8 40 50 1000 9 60 60 2000

[0076] Table 1 Orthogonal test table example

[0077] According to the test combination in the orthogonal table, the simulation environment is adjusted so that the test conditions correspond to the temperature, humidity and number of cyclic loads under each test number. Under each test condition, the concrete test block is subjected to accelerated aging simulation.

[0078] After the accelerated aging simulation is completed, the concrete specimens are evaluated for performance, such as by measuring their compressive strength, durability index and other indicators. The performance indicator data under each test number is recorded. And based on the performance evaluation results, the critical state of accelerated failure of the concrete specimen durability is determined. This is usually the threshold of one or more performance indicators. When the performance indicators reach or exceed these thresholds, the concrete specimen is considered to be close to or has reached the failure state.

[0079] S104, selecting a dynamic covariate that changes with time and is closely related to the failure mechanism from the test conditions of the equivalent accelerated failure critical state, and constructing a risk regression model based on the dynamic covariate data.

[0080] In orthogonal experiments, multiple dynamic covariates that change over time can be observed. For example, dynamic covariates that are closely related to concrete structure cracking and waterproof material aging, such as real-time data of temperature, humidity, and load, are extracted from the data of orthogonal experiments. These covariates are closely related to the failure mechanism of the building and can be used to construct a risk regression model, so that the constructed model can predict the relative failure risk of the building under different conditions.

[0081] Selecting dynamic covariates that are closely related to the failure mechanism is the basis for ensuring the accuracy and reliability of the model. These covariates can change over time and reflect key indicators of building performance. By carefully selecting these covariates, the core characteristics of the failure process can be more accurately captured, providing strong support for risk prediction.

[0082] By building a risk regression model based on dynamic covariate data, we can quantitatively analyze the impact of covariates on failure risk. This model can reveal the relationship between covariates and failure risk, thereby more accurately predicting the failure risk of a building under given conditions. Through the analysis of the risk regression model, we can also understand which factors are the key factors affecting the failure risk, and formulate effective preventive measures and response strategies accordingly.

[0083] In addition, the risk regression model is flexible and interpretable. By adjusting the model parameters and considering different combinations of covariates, it can adapt to different test conditions and failure mechanisms. At the same time, the model results are clearly interpretable, which can help us better understand the failure process and the results of risk prediction.

[0084] S105, obtain real-time dynamic covariate data related to building performance through building monitoring equipment, input the real-time dynamic covariate data into the constructed risk regression model, calculate and output the relative failure risk value of the current building, and issue a risk warning if the relative failure risk value of the current building exceeds the preset safety threshold.

[0085] By acquiring dynamic covariate data closely related to building performance in real time through building monitoring equipment and inputting it into the constructed risk regression model, the relative failure risk value of the current building can be quickly calculated and output. This step realizes continuous monitoring and dynamic risk assessment of the building status. Once the relative failure risk value exceeds the preset safety threshold, the system immediately issues a risk warning so that corresponding maintenance and preventive measures can be taken in time to ensure the safe and reliable operation of the building.

[0086] Based on steps 101 to 105, by collecting the data of the entire life cycle of the building, after preprocessing, it is used to analyze and accurately identify the failure mode of the building and its key influencing factors. By designing an orthogonal test to simulate the accelerated aging process of the building, it can not only efficiently explore the interaction between different factors, but also accurately determine the critical state of accelerated failure, which is crucial for understanding and predicting the performance changes of the building in actual use. Dynamic covariates closely related to the failure mechanism are selected from the test conditions. These variables can effectively reflect the changing trend of building performance and are key factors in assessing building risks. Based on these dynamic covariate data, a risk regression model is constructed. The constructed risk regression model not only considers the comprehensive effect of multiple influencing factors, but also considers the interaction between these factors. It can accurately reflect the complex relationship between building performance and failure risk, and quantitatively evaluate the relative failure risk of the building, thereby improving the accuracy of risk assessment. Through the data obtained by the real-time monitoring equipment, the model can dynamically update the risk value. Once the preset threshold is exceeded, the system will automatically issue a risk warning, thereby greatly improving the efficiency and effectiveness of building risk management.

[0087] Further optionally, in step 102, the step of analyzing and extracting the failure modes and key influencing factors of the building based on the pre-processed full life cycle data of the building includes:

[0088] Identify all events related to building failure from the building's full life cycle data and set them as failure events;

[0089] Clustering algorithms are used to cluster failure events and group similar failure events together to identify different failure modes. The specific steps are as follows:

[0090] Construct a data set of failure events {x1, x2, ..., x n}, where each data point x i is a multidimensional vector that contains multiple features describing the failure event, that is, it contains the observed values ​​of each factor, expressed as x i ={F1(i), F2(i), ..., F m (i)}, m is the number of factors;

[0091] Run the clustering algorithm to divide the data set into K clusters. The objective function of the clustering algorithm is used to minimize the sum of squares of the data points in the cluster. The objective function of the clustering algorithm is:

[0092]

[0093] Among them, C k is the set of data points of the kth cluster, μ k is the centroid of the kth cluster, i.e., the mean vector;

[0094] Based on the characteristics of each cluster data point, K different failure modes of the building are defined;

[0095] Calculate the normalized difference ratio NDP for each data point using the formula:

[0096] Among them, NDP i is the normalized difference ratio of data point i, which belongs to failure mode k, F j (i) is the factor F in data point i j The observed value, y k is the factor F in failure mode k j The representative value of all data points in failure mode k is the factor F j The mean of

[0097] Calculate the factor F j The factor F is obtained by taking the average normalized difference ratio of failure mode k. j The influence degree on failure mode k is calculated as follows:

[0098] Among them, Indices k is the number of data points in failure mode k;

[0099] By analogy, the influence of each factor on each failure mode is calculated;

[0100] For each failure mode, factors with high influence are extracted as their key influencing factors.

[0101] Based on this step, when analyzing the full life cycle data of a building, the first task is to identify events related to building failures, which are classified as failure events. In order to deeply understand these failure events and identify different failure modes, this embodiment adopts a clustering algorithm. Specifically, a failure event data set containing multiple features (i.e., the observed values ​​of each factor) is first constructed, and a clustering algorithm is used to divide it into several clusters. During the clustering process, the objective function is optimized to minimize the sum of squares of the data points within the cluster, thereby ensuring that similar failure events are classified into the same cluster. Each cluster represents a unique failure mode, and these failure modes of the building are defined according to the characteristics of the data points in the cluster.

[0102] Next, to quantify the degree of influence of each factor on a specific failure mode, the Normalized Difference Proportion (NDP) is calculated for each data point, which is a measure of the difference between the observed value of each factor in the data point and the representative value of the factor in the corresponding failure mode. Then, the average Normalized Difference Proportion for each factor in each failure mode is calculated, providing a direct measure of the influence of each factor on the failure mode.

[0103] Once the various failure modes and their key influencing factors are identified by analyzing the building's full life cycle data, this information can be used to design targeted orthogonal experiments. Orthogonal experiments are an effective experimental design method that allows the systematic study of the effects of multiple variables (i.e., key influencing factors) on failure modes while controlling the interference of other factors. In orthogonal experiments, different test conditions are set according to the extracted key influencing factors, and the accelerated aging process of the building under these conditions is simulated. By performing these experiments, the changes in building performance under different conditions can be observed and recorded, thereby gaining a deeper understanding of the specific effects of these key influencing factors on failure modes.

[0104] Further optionally, in step 103, the step of designing an orthogonal test to simulate the accelerated aging process of the building based on the extracted failure mode and key influencing factors of the building, and determining the critical state of accelerated failure includes:

[0105] Based on the extracted key influencing factors, an orthogonal test scheme was designed, and the accelerated aging process of the building under different conditions was simulated by finite element method;

[0106] Carry out the test according to each test scheme condition in the orthogonal table and record the changes in building performance under different conditions;

[0107] Through test data, analyze the performance change trend of the building under different failure modes, and determine the critical state of accelerated failure under each failure mode. The critical state of accelerated failure means that when the performance of the building drops to a certain level, it is considered that the building has reached or is close to failure.

[0108] Compare the accelerated failure critical states obtained in the simulation with the failure states in the actual building;

[0109] When a simulated accelerated failure critical state consistent with the actual failure state is found, the test conditions represented by the orthogonal test scheme are determined to be accelerated test conditions equivalent to the actual service conditions.

[0110] Based on this step, when evaluating the durability of a building, an orthogonal test scheme is first designed based on the extracted key influencing factors. This step aims to simulate the accelerated aging process of the building under different conditions through the finite element simulation method. Subsequently, the tests are performed one by one according to the test scheme conditions set in the orthogonal table, and the change data of the building performance under different conditions are recorded in detail. Next, the test data is deeply analyzed to reveal the performance change trend of the building under different failure modes, and the critical state of accelerated failure under each failure mode is determined accordingly. The critical state of accelerated failure here is a sign when the building performance drops to a certain degree, indicating that the building has reached or is close to a failure state.

[0111] To verify the accuracy of the simulation results, the critical state of accelerated failure obtained in the simulation is compared with the failure state observed in the actual building. Once the critical state of simulated accelerated failure consistent with the actual failure state is found, it can be determined that the test conditions represented by the orthogonal test scheme are accelerated test conditions equivalent to the actual service conditions. This step not only verifies the effectiveness of the simulation method, but also provides an important reference for subsequent building risk prediction and assessment.

[0112] The finite element simulation in this step is used to simulate the accelerated aging process of buildings under different conditions. Finite element simulation is a powerful numerical analysis tool that can handle complex physical phenomena and structural responses, especially in building durability assessment. The specific analysis steps are as follows:

[0113] First, according to the different test conditions in the orthogonal test scheme, the corresponding parameters in the finite element model are set. These parameters reflect the material properties, boundary conditions, loading conditions, etc. of the building under different conditions, which can directly affect the accuracy and reliability of the simulation.

[0114] Next, the finite element analysis software is used to construct a finite element model of the building. The model needs to accurately reflect the structural characteristics, material properties and geometric shape of the building to ensure the accuracy of the simulation results. At the same time, in order to simulate the accelerated aging process, it is also necessary to introduce a time-related material degradation model or damage accumulation model into the model.

[0115] After the model is built, the finite element simulation begins. During the simulation, the software calculates the response of the building at different time points, including stress, strain, displacement, etc., according to the set parameters and conditions. By continuously simulating the time history, the degradation process of the building performance over time can be observed, thereby simulating the accelerated aging process.

[0116] After the simulation is finished, the simulation results can be extracted and analyzed. These results include the performance trend of the building under different conditions, the occurrence time of failure modes, and the critical state of accelerated failure. By analyzing this data, the durability of the building under specific conditions can be evaluated, the risk of failure can be predicted, and even its remaining life can be predicted.

[0117] Further optionally, in step 104, the step of selecting a dynamic covariate that varies with time and is closely related to the failure mechanism from the test conditions of the equivalent accelerated failure critical state, and constructing a risk regression model based on the dynamic covariate data includes:

[0118] Selecting time-varying dynamic covariates that are closely related to the failure mechanism from the test conditions of the equivalent accelerated failure critical state, collecting the dynamic covariate data of each test sample and the corresponding failure time, and generating a dynamic covariate vector for each sample to construct a risk set, wherein each sample in the risk set contains a set of time-varying dynamic covariate values ​​and a corresponding failure time;

[0119] A risk regression model is constructed. The risk regression model is used to evaluate the impact of covariates on failure risk. The failure risk assessment formula of the model is as follows:

[0120]

[0121] Where h(t|X) represents the failure risk function value of the building at time t when the covariate vector is Z, h0(t) is the baseline risk function, which refers to the instantaneous risk of building failure at time t when there is no covariate effect, that is, when all covariates are 0, β1, β2, …, β p are model parameters, i.e., the coefficients of each covariate, which represent the influence of each dynamic covariate on the failure risk, z1, z2, …, z p is the value of each dynamic covariate in a set of dynamic covariate vectors Z in the risk set;

[0122] Substitute the dynamic covariate vector in the risk concentration into the risk regression model, and use maximum likelihood estimation to estimate the covariate coefficient vector B, B = {β1, β2, ..., β p}, a parameter-determined risk regression model is obtained to predict the building failure risk under given covariate values.

[0123] Based on this step, it is crucial to accurately select dynamic covariates that are closely related to the failure mechanism and vary with time from the test conditions of the critical state of equivalent accelerated failure. By collecting the dynamic covariate data of each test sample and its corresponding failure time, and constructing the dynamic covariate vector of each sample to form a risk set, the correlation between each covariate value and the failure time can be effectively captured. Then, these data are used to construct a risk regression model, which evaluates the impact of each covariate on the failure risk through the formula h(t|X). The dynamic covariate vector in the risk set is substituted into the model, and the covariate coefficient vector B is estimated by the maximum likelihood estimation method, thereby obtaining a risk regression model with determined parameters. The constructed risk regression model not only provides a reliable tool for predicting the building failure risk under a given covariate value, but also greatly improves the accuracy and reliability of failure risk assessment, which has a significant beneficial effect on guiding practical engineering applications and reducing risks.

[0124] Further optionally, the step of substituting the dynamic covariate vector in the risk set into the risk regression model and estimating the covariate coefficient vector B by using maximum likelihood estimation includes:

[0125] The partial likelihood function is used to estimate the dynamic covariate coefficient vector B. The partial likelihood function formula is:

[0126]

[0127] The entire partial likelihood function formula is a product that only includes those observations for which events actually occurred, i.e., δ i = 1, the numerator Represents the i-th observation object in the dynamic covariate vector Z i and the covariate coefficient vector B, the denominator Indicates that at time t i The sum of the risks of all the observations that have not yet experienced an event, that is, those that are still at risk. The denominator is to normalize all the observations that are still at risk to ensure that the value of the partial likelihood function is between 0 and 1. i is an event indicator. If the i-th observation object is at t i When an event occurs at time, i.e., building failure, then δ i =1, if the i-th observation object is at t i If the time is censored, that is, no failure is observed, then δ i =0, R(t i ) is at time t i The set of all observed objects that have not experienced any events before is the risk set; Z i and Z j Represent the covariate vectors of the i-th and j-th observations respectively;

[0128] The partial likelihood function L(B) is maximized by numerical optimization to obtain the estimated value of the covariate coefficient vector B, where B = {β1, β2, ..., β p}.

[0129] Based on this step, the dynamic covariate vector in the risk set is substituted into the risk regression model, and the maximum likelihood estimation method is used to estimate the covariate coefficient vector B. First, by constructing a partial likelihood function, which only contains the observed objects of the actual event (i.e., δ i =1), ensuring that the model focuses on information that directly affects risk prediction. The partial likelihood function is maximized using numerical optimization methods to obtain an estimate of the covariate coefficient vector B. This step not only considers the impact of covariates on failure risk, but also finds the optimal combination of covariate coefficients by maximizing the likelihood function. This optimization process improves the predictive ability of the model, enabling the model to more accurately predict building failure risks under given covariate values.

[0130] Further optionally, the step of maximizing the partial likelihood function L(B) by a numerical optimization method to obtain an estimated value of the covariate coefficient vector B includes:

[0131] Randomly select an initial covariate coefficient vector B0;

[0132] Calculate the gradient vector of the partial likelihood function L(B) with respect to the covariate coefficient vector B Among them, the gradient vector contains the direction information where the partial likelihood function value changes fastest in each parameter direction;

[0133] Use the optimization algorithm to iteratively update the value of the covariate coefficient vector B until the gradient is small enough. In each iteration, a new value of the covariate coefficient vector B is calculated based on the current value of the covariate coefficient vector B and the gradient information. The update formula is: Among them, α is the learning rate, which is used to control the update step size;

[0134] Output the estimated value of the final covariate coefficient vector B, where B = {β1, β2, ..., β p}.

[0135] Based on this step, when constructing the risk regression model, the partial likelihood function L(B) is maximized by a numerical optimization method to estimate the covariate coefficient vector B. First, an initial covariate coefficient vector B0 is randomly selected as the starting point. Then, the gradient vector of the partial likelihood function L(B) with respect to the covariate coefficient vector B is calculated. Next, the optimization algorithm is used to iteratively update the value of the covariate coefficient vector B. In each iteration, according to the current value of the covariate coefficient vector B and the gradient information, the formula is updated To calculate the value of the new covariate coefficient vector B. This process continues until the gradient is small enough, that is, the convergence condition of the optimization algorithm is reached. Finally, the estimated value of the final covariate coefficient vector B is output, B = {β1, β2, ..., βp}. These estimated values ​​reflect the influence of each dynamic covariate on the failure risk, providing an important basis for the prediction of building failure risk.

[0136] Further optionally, in step 105, the step of inputting the real-time dynamic covariate data into the constructed risk regression model to calculate and output the relative risk value of the current building includes:

[0137] The dynamic covariate data obtained by real-time monitoring are organized into a vector form to obtain a real-time dynamic covariate vector;

[0138] The real-time dynamic covariate vector is input into the constructed risk regression model. Based on the real-time dynamic covariate vector, the instantaneous failure risk of the current building is calculated. The calculation formula is:

[0139]

[0140] Among them, h(t 当前 ∣Z 实时 ) represents the real-time observation value Z at the current time t and the dynamic covariate vector 实时 Under this condition, the instantaneous failure risk of the building, h0(t 当前 ) is the baseline risk, which refers to the instantaneous risk of building failure at the current time t when there is no covariate effect, that is, when all covariates are 0. B is the covariate coefficient vector, which indicates the influence of each dynamic covariate on the failure risk. B = {β1, β2, …, β p}, Z 实时 is the real-time observation value of the covariate vector;

[0141] Based on the instantaneous failure risk and benchmark risk of the current building, the relative risk of the current building is calculated to obtain the relative risk value d of the current building.

[0142] Based on this step, the dynamic covariate data obtained in real time by the building monitoring equipment is organized into a vector form to obtain a real-time dynamic covariate vector, which is then input into the risk regression model, and the risk regression model is used to calculate the instantaneous failure risk of the current building. Based on the instantaneous failure risk and the baseline risk of the current building, the relative risk value d of the current building is calculated. This value represents the relative size of the failure risk of the building relative to the baseline risk under the current conditions. This process not only provides an accurate assessment of the current risk of the building, but also provides strong support for subsequent risk management and maintenance decisions.

[0143] Furthermore, an embodiment of the present invention also provides a monitoring and assessment system for building risks. Figure 2 FIG. 4 shows a system block diagram of a building risk monitoring and assessment system provided by an embodiment of the present invention, such as Figure 2 As shown, the system includes:

[0144] Data collection module: used to collect data on the entire life cycle of a building and perform preprocessing;

[0145] Extraction module: used to analyze and extract the failure modes and key influencing factors of buildings based on the pre-processed life cycle data of buildings;

[0146] Orthogonal test module: used to design orthogonal tests to simulate the accelerated aging process of buildings and determine the critical state of accelerated failure based on the extracted failure modes and key influencing factors of the buildings;

[0147] Model building module: used to select dynamic covariates that vary with time and are closely related to the failure mechanism from the test conditions of the equivalent accelerated failure critical state, and to build a risk regression model based on the dynamic covariate data;

[0148] Risk prediction module: used to obtain real-time dynamic covariate data related to building performance through building monitoring equipment, input the real-time dynamic covariate data into the constructed risk regression model, calculate and output the relative failure risk value of the current building, and issue a risk warning if the relative failure risk value of the current building exceeds the preset safety threshold.

[0149] Further optionally, the extraction module is further used for:

[0150] Identify all events related to building failure from the building's full life cycle data and set them as failure events. Failure events include crack generation, structural deformation, and equipment failure.

[0151] Clustering algorithms are used to cluster failure events and group similar failure events together to identify different failure modes. The specific steps are as follows:

[0152] Construct a data set of failure events {x1, x2, ..., x n}, where each data point x i is a multidimensional vector that contains multiple features describing the failure event, that is, it contains the observed values ​​of each factor, expressed as x i ={F1(i), F2(i), ..., F m (i)}, m is the number of factors;

[0153] Run the clustering algorithm to divide the data set into K clusters. The objective function of the clustering algorithm is used to minimize the sum of squares of the data points in the cluster. The objective function of the clustering algorithm is:

[0154]

[0155] Among them, C k is the set of data points of the kth cluster, μ k is the centroid of the kth cluster, i.e., the mean vector;

[0156] Based on the characteristics of each cluster data point, K different failure modes of the building are defined;

[0157] Calculate the normalized difference ratio NDP for each data point using the formula:

[0158] Among them, NDP i is the normalized difference ratio of data point i, which belongs to failure mode k, F j (i) is the factor F in data point i j The observed value, y k is the factor F in failure mode k j The representative value of all data points in failure mode k is the factor F j The mean of

[0159] Calculate the factor F j The factor F is obtained by taking the average normalized difference ratio of failure mode k. j The influence degree on failure mode k is calculated as follows:

[0160] Among them, Indices k is the number of data points in failure mode k;

[0161] By analogy, the influence of each factor on each failure mode is calculated;

[0162] For each failure mode, factors with high influence are extracted as their key influencing factors.

[0163] Optionally, the orthogonal test module is also used for:

[0164] Based on the extracted key influencing factors, an orthogonal test scheme was designed, and the accelerated aging process of the building under different conditions was simulated by finite element method;

[0165] Carry out the test according to each test scheme condition in the orthogonal table and record the changes in building performance under different conditions;

[0166] Through test data, analyze the performance change trend of the building under different failure modes, and determine the critical state of accelerated failure under each failure mode. The critical state of accelerated failure means that when the performance of the building drops to a certain level, it is considered that the building has reached or is close to failure.

[0167] Compare the accelerated failure critical states obtained in the simulation with the failure states in the actual building;

[0168] When a simulated accelerated failure critical state consistent with the actual failure state is found, the test conditions represented by the orthogonal test scheme are determined to be accelerated test conditions equivalent to the actual service conditions.

[0169] Optionally, the model building module is further used to:

[0170] Selecting time-varying dynamic covariates that are closely related to the failure mechanism from the test conditions of the equivalent accelerated failure critical state, collecting the dynamic covariate data of each test sample and the corresponding failure time, and generating a dynamic covariate vector for each sample to construct a risk set, wherein each sample in the risk set contains a set of time-varying dynamic covariate values ​​and a corresponding failure time;

[0171] A risk regression model is constructed. The risk regression model is used to evaluate the impact of covariates on failure risk. The failure risk assessment formula of the model is as follows:

[0172]

[0173] Where h(t|X) represents the failure risk function value of the building at time t when the covariate vector is Z, h0(t) is the baseline risk function, which refers to the instantaneous risk of building failure at time t when there is no covariate effect, that is, when all covariates are 0, β1, β2, …, β p are model parameters, i.e., the coefficients of each covariate, which represent the influence of each dynamic covariate on the failure risk, z1, z2, …, z p is the value of each dynamic covariate in a set of dynamic covariate vectors Z in the risk set;

[0174] Substitute the dynamic covariate vector in the risk concentration into the risk regression model, and use maximum likelihood estimation to estimate the covariate coefficient vector B, B = {β1, β2, ..., β p}, a parameter-determined risk regression model is obtained to predict the building failure risk under given covariate values.

[0175] Optionally, the model building module is further used to:

[0176] The partial likelihood function is used to estimate the dynamic covariate coefficient vector B. The partial likelihood function formula is:

[0177]

[0178] The entire partial likelihood function formula is a product that only includes those observations for which events actually occurred, i.e., δ i = 1, the numerator Represents the i-th observation object in the dynamic covariate vector Z i and the covariate coefficient vector B, the denominator Indicates that at time t i The sum of the risks of all the observations that have not yet experienced an event, that is, those that are still at risk. The denominator is to normalize all the observations that are still at risk to ensure that the value of the partial likelihood function is between 0 and 1. i is an event indicator. If the i-th observation object is at t i When an event occurs at time, i.e., building failure, then δ i =1, if the i-th observation object is at t i If the time is censored, that is, no failure is observed, then δ i =0, R(t i ) is at time t i The set of all observed objects that have not experienced any events before is the risk set; Z i and Z j Represent the covariate vectors of the i-th and j-th observations respectively;

[0179] The partial likelihood function L(B) is maximized by numerical optimization to obtain the estimated value of the covariate coefficient vector B, where B = {β1, β2, ..., β p}.

[0180] Optionally, the model building module is further used to:

[0181] Randomly select an initial covariate coefficient vector B0;

[0182] Calculate the gradient vector of the partial likelihood function L(B) with respect to the covariate coefficient vector B Among them, the gradient vector contains the direction information where the partial likelihood function value changes fastest in each parameter direction;

[0183] Use the optimization algorithm to iteratively update the value of the covariate coefficient vector B until the gradient is small enough. In each iteration, a new value of the covariate coefficient vector B is calculated based on the current value of the covariate coefficient vector B and the gradient information. The update formula is: Among them, α is the learning rate;

[0184] Output the estimated value of the final covariate coefficient vector B, where B = {β1, β2, ..., β p}.

[0185] Further optionally, the risk prediction module is also used to:

[0186] The dynamic covariate data obtained by real-time monitoring are organized into a vector form to obtain a real-time dynamic covariate vector;

[0187] The real-time dynamic covariate vector is input into the constructed risk regression model. Based on the real-time dynamic covariate vector, the instantaneous failure risk of the current building is calculated. The calculation formula is:

[0188]

[0189] Among them, h(t 当前 ∣Z 实时 ) represents the real-time observation value Z at the current time t and the dynamic covariate vector 实时 Under this condition, the instantaneous failure risk of the building, h0(t 当前 ) is the baseline risk, which refers to the instantaneous risk of building failure at the current time t when there is no covariate effect, that is, when all covariates are 0. B is the covariate coefficient vector, which indicates the influence of each dynamic covariate on the failure risk. B = {β1, β2, …, β p}, Z 实时 is the real-time observation value of the covariate vector;

[0190] Based on the instantaneous failure risk and benchmark risk of the current building, the relative risk of the current building is calculated to obtain the relative risk value d of the current building.

[0191] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for monitoring and assessing building risks, characterized in that the steps include: Collect data from the entire life cycle of the building and perform pre-processing; Based on the pre-processed building life cycle data, analyze and extract the building failure mode and its key influencing factors; According to the extracted failure modes and key influencing factors of the building, an orthogonal test is designed to simulate the accelerated aging process of the building and determine the critical state of accelerated failure; Select dynamic covariates that are closely related to the failure mechanism and change over time from the test conditions of the equivalent accelerated failure critical state, and build a risk regression model based on the dynamic covariate data; Through building monitoring equipment, real-time dynamic covariate data related to building performance is obtained, and the real-time dynamic covariate data is input into the constructed risk regression model to calculate and output the relative failure risk value of the current building. If the relative failure risk value of the current building exceeds the preset safety threshold, a risk warning is issued.

2. The method for monitoring and assessing building risks according to claim 1, characterized in that: The step of inputting the real-time dynamic covariate data into the constructed risk regression model to calculate and output the relative risk value of the current building includes: The dynamic covariate data obtained by real-time monitoring are organized into a vector form to obtain a real-time dynamic covariate vector; The real-time dynamic covariate vector is input into the constructed risk regression model. Based on the real-time dynamic covariate vector, the instantaneous failure risk of the current building is calculated. The calculation formula is: Among them, h(t 当前 ∣Z 实时 ) represents the real-time observation value Z at the current time t and the dynamic covariate vector 实时 Under this condition, the instantaneous failure risk of the building, h0(t 当前 ) is the baseline risk, which refers to the instantaneous risk of building failure at the current time t when there is no covariate effect, that is, when all covariates are 0. B is the covariate coefficient vector, which indicates the influence of each dynamic covariate on the failure risk. B = {β1, β2, …, β p }, Z 实时 is the real-time observation value of the covariate vector; Based on the instantaneous failure risk and benchmark risk of the current building, the relative risk of the current building is calculated to obtain the relative risk value d of the current building.

3. The method for monitoring and assessing building risks according to claim 1, characterized in that: The steps of selecting a dynamic covariate that changes with time and is closely related to the failure mechanism from the test conditions of the equivalent accelerated failure critical state, and constructing a risk regression model based on the dynamic covariate data include: Selecting time-varying dynamic covariates that are closely related to the failure mechanism from the test conditions of the equivalent accelerated failure critical state, collecting the dynamic covariate data of each test sample and the corresponding failure time, and generating a dynamic covariate vector for each sample to construct a risk set, wherein each sample in the risk set contains a set of time-varying dynamic covariate values ​​and a corresponding failure time; A risk regression model is constructed. The risk regression model is used to evaluate the impact of covariates on failure risk. The failure risk assessment formula of the model is as follows: Where h(t|X) represents the failure risk function value of the building at time t when the covariate vector is Z, h0(t) is the baseline risk function, which refers to the instantaneous risk of building failure at time t when there is no covariate effect, that is, when all covariates are 0, β1, β2, …, β p are model parameters, i.e., the coefficients of each covariate, which represent the influence of each dynamic covariate on the failure risk, z1, z2, …, z p is the value of each dynamic covariate in a set of dynamic covariate vectors Z in the risk set; Substitute the dynamic covariate vector in the risk concentration into the risk regression model, and use maximum likelihood estimation to estimate the covariate coefficient vector B, B = {β1, β2, ..., β p }, a parameter-determined risk regression model is obtained to predict the building failure risk under given covariate values.

4. The method for monitoring and assessing building risks according to claim 3, characterized in that: The step of substituting the dynamic covariate vector in the risk set into the risk regression model and estimating the covariate coefficient vector B by using maximum likelihood estimation includes: The partial likelihood function is used to estimate the dynamic covariate coefficient vector B. The partial likelihood function formula is: The entire partial likelihood function formula is a product that only includes those observations for which events actually occurred, i.e., δ i = 1, the numerator Represents the i-th observation object in the dynamic covariate vector Z i and the covariate coefficient vector B, the denominator Indicates that at time t i The sum of the risks of all the observations that have not yet experienced an event, that is, those that are still at risk. The denominator is to normalize all the observations that are still at risk to ensure that the value of the partial likelihood function is between 0 and 1. i is an event indicator. If the i-th observation object is at t i When an event occurs at time, i.e., building failure, then δ i =1, if the i-th observation object is at t i If the time is censored, that is, no failure is observed, then δ i =0, R(t i ) is at time t i The set of all observed objects that have not experienced any events before is the risk set; Z i and Z j Represent the covariate vectors of the i-th and j-th observations respectively; The partial likelihood function L(B) is maximized by numerical optimization to obtain the estimated value of the covariate coefficient vector B, where B = {β1, β2, ..., β p }.

5. The method for monitoring and assessing building risks according to claim 4, characterized in that: The step of maximizing the partial likelihood function L(B) by a numerical optimization method to obtain an estimated value of the covariate coefficient vector B comprises: Randomly select an initial covariate coefficient vector B0; Calculate the gradient vector of the partial likelihood function L(B) with respect to the covariate coefficient vector B Among them, the gradient vector contains the direction information where the partial likelihood function value changes fastest in each parameter direction; Use the optimization algorithm to iteratively update the value of the covariate coefficient vector B until the gradient is small enough. In each iteration, a new value of the covariate coefficient vector B is calculated based on the current value of the covariate coefficient vector B and the gradient information. The update formula is: Among them, α is the learning rate; Output the estimated value of the final covariate coefficient vector B, where B = {β1, β2, ..., β p }.

6. The method for monitoring and assessing building risks according to claim 1, characterized in that: The steps of analyzing and extracting the failure modes and key influencing factors of the building based on the pre-processed full life cycle data of the building include: Identify all events related to building failure from the building's full life cycle data and set them as failure events. Failure events include crack generation, structural deformation, and equipment failure. Clustering algorithms are used to cluster failure events and group similar failure events together to identify different failure modes. The specific steps are as follows: Construct a data set of failure events {x1, x2, ..., x n }, where each data point x i is a multidimensional vector that contains multiple features describing the failure event, that is, it contains the observed values ​​of each factor, expressed as x i ={F1(i), F2(i), ..., F m (i)}, m is the number of factors; Run the clustering algorithm to divide the data set into K clusters. The objective function of the clustering algorithm is used to minimize the sum of squares of the data points in the cluster. The objective function of the clustering algorithm is: Among them, C k is the set of data points of the kth cluster, μ k is the centroid of the kth cluster, i.e., the mean vector; Based on the characteristics of each cluster data point, K different failure modes of the building are defined; Calculate the normalized difference ratio NDP for each data point using the formula: Among them, NDP i is the normalized difference ratio of data point i, which belongs to failure mode k, F j (i) is the factor F in data point i j The observed value, y k is the factor F in failure mode k j The representative value of all data points in failure mode k is the factor F j The mean of Calculate the factor F j The factor F is obtained by taking the average normalized difference ratio of failure mode k. j The influence degree on failure mode k is calculated as follows: Among them, Indices k is the number of data points in failure mode k; By analogy, the influence of each factor on each failure mode is calculated; For each failure mode, factors with high influence are extracted as their key influencing factors.

7. The method for monitoring and assessing building risks according to claim 1, characterized in that: The steps of designing an orthogonal test to simulate the accelerated aging process of the building based on the extracted failure mode and key influencing factors of the building and determining the critical state of accelerated failure include: Based on the extracted key influencing factors, an orthogonal test scheme was designed, and the accelerated aging process of the building under different conditions was simulated by finite element method; Carry out the test according to each test scheme condition in the orthogonal table and record the changes in building performance under different conditions; Through test data, analyze the performance change trend of the building under different failure modes, and determine the critical state of accelerated failure under each failure mode. The critical state of accelerated failure means that when the performance of the building drops to a certain level, it is considered that the building has reached or is close to failure. Compare the accelerated failure critical states obtained in the simulation with the failure states in the actual building; When a simulated accelerated failure critical state consistent with the actual failure state is found, the test conditions represented by the orthogonal test scheme are determined to be accelerated test conditions equivalent to the actual service conditions.

8. A building risk monitoring and assessment system, used to implement the building risk monitoring and assessment method according to any one of claims 1 to 7, characterized in that: include: Data collection module: used to collect data on the entire life cycle of a building and perform preprocessing; Extraction module: used to analyze and extract the failure modes and key influencing factors of buildings based on the pre-processed life cycle data of buildings; Orthogonal test module: used to design orthogonal tests to simulate the accelerated aging process of buildings and determine the critical state of accelerated failure based on the extracted failure modes and key influencing factors of the buildings; Model building module: used to select dynamic covariates that vary with time and are closely related to the failure mechanism from the test conditions of the equivalent accelerated failure critical state, and to build a risk regression model based on the dynamic covariate data; Risk prediction module: used to obtain real-time dynamic covariate data related to building performance through building monitoring equipment, input the real-time dynamic covariate data into the constructed risk regression model, calculate and output the relative failure risk value of the current building, and issue a risk warning if the relative failure risk value of the current building exceeds the preset safety threshold.

Images