Steel structure building construction whole process mechanical property evaluation method based on digital twinning

Abstract

The invention relates to the technical field of digital twinning, in particular to a steel structure building construction whole process mechanical property evaluation method based on digital twinning, which comprises the following steps of: acquiring multi-source heterogeneous steel structure state data from a stress sensor, BIM model component information and a displacement monitoring point, comparing timestamps from all data sources, and determining the state of the multi-source heterogeneous steel structure according to the comparison result; and calculating a transmission delay value of each data source relative to the unified reference time, correcting the original data timestamps one by one, and establishing a steel structure state data stream with aligned time sequences. According to the method, the timestamps of the multi-source heterogeneous data are subjected to transmission delay correction through unified reference time, the timestamps of the original data are corrected one by one, the steel structure state data streams with the aligned time sequences are generated, the time deviation among the sensors, the BIM model and the monitoring points is eliminated, and the synchronism and coherence of dynamic changes of mechanical parameters are guaranteed.

Description

Technical Field

[0001] The present invention relates to the technical field of digital twins, and in particular to a method for evaluating mechanical properties of a steel structure during the entire construction process based on digital twins. Background Art

[0002] Digital twin is a technical field that takes physical entities as objects, builds digital mirrors of physical entities in virtual space through multi-source data fusion, dynamic modeling and real-time interaction technology, and realizes state mapping, behavior prediction and closed-loop optimization based on data drive.

[0003] When existing digital twin technology integrates multi-source heterogeneous data in real time, it does not consider the timestamp deviation caused by transmission delays of different data sources. There is a time sequence misalignment in the data streams of sensors, BIM models and monitoring points, which may cause state misjudgment due to time asynchrony when evaluating dynamic mechanical properties. At the same time, the dependency relationship is implicit in the model, making it difficult to trace abnormal results to specific input data or intermediate links, and it is difficult to quickly locate the source of the fault. Therefore, improvements are needed. Summary of the invention

[0004] The purpose of the present invention is to solve the shortcomings existing in the prior art and propose a method for evaluating the mechanical properties of the entire process of steel structure construction based on digital twins.

[0005] In order to achieve the above purpose, the present invention adopts the following technical scheme: a method for evaluating the mechanical properties of a steel structure building during the entire construction process based on digital twins, comprising the following steps: Acquire multi-source heterogeneous steel structure status data from stress sensors, BIM model component information and displacement monitoring points, compare the timestamps of each data source with each other, calculate the transmission delay value of each data source relative to the unified reference time, correct the timestamps of the original data one by one, and establish a time-aligned steel structure status data stream; Based on the time-series aligned steel structure state data stream, extract data records describing the state parameters of the same steel structure component, assign a confidence value to each data record, generate steel structure state parameter records with confidence, calculate fusion estimation values ​​for records of the same parameter based on the steel structure state parameter records with confidence, compare the fusion estimation value differences between different records with a preset conflict judgment limit, and construct a steel structure state parameter set with fusion conflict marked; Based on the steel structure state parameter set marked by the fusion conflict, the state variables directly corresponding to the steel structure state parameter set marked by the fusion conflict in the steel structure digital twin model are retrieved, the variables are incrementally updated to obtain the steel structure model state increment to be propagated, based on the steel structure model state increment to be propagated, the influence of the state increment on the associated variables is calculated, and the steel structure twin model state and data dependency log after the update propagation is established; Retrieve the mechanical property evaluation results of the target in the state of the steel structure twin model after the update propagation, locate the corresponding record entries of the mechanical property evaluation results in the data dependency log, obtain the log entry points associated with the target results, and based on the log entry points associated with the target results, perform reverse trace retrieval in the data dependency log according to the recorded variable dependency chain information to generate a causal trace path of mechanical property anomalies.

[0006] Preferably, the steps for obtaining the time-series aligned steel structure state data stream are as follows: Extract the original timestamps of each data source from the stress sensor, BIM model component information, and displacement monitoring points. Taking the unified reference time as the standard, calculate the absolute value of the difference between the timestamp of each data source and the reference time, define the absolute value of the difference as the initial transmission delay value, and generate a set of initial transmission delay values; Based on the set of initial transmission delay values, statistically analyze the distribution of the initial transmission delay values of all data sources, take the 90th percentile of the initial transmission delay value distribution as the preset time deviation threshold, and generate a compensation factor according to the ratio of the time deviation threshold to the data source delay value. The compensation factor calculation formula is: Compensation factor = data source delay value / time deviation threshold, and generate a set of dynamic time compensation factors; Based on the set of dynamic time compensation factors, for data sources with initial transmission delay values exceeding the time deviation threshold, use the formula: Corrected timestamp = original timestamp + (data source delay value / (compensation factor + 1)) to correct the timestamps one by one and generate a time-series aligned steel structure state data stream.

[0007] Preferably, the steps for obtaining the steel structure state parameter records with confidence levels are as follows: Extract multiple state parameter data records of the same steel structure component from the time-series aligned steel structure state data stream, classify them according to the data source type, and generate a set of state parameter records without confidence levels; Based on the set of state parameter records without confidence levels, calculate the confidence level of each record. The calculation formula is: ; where is the confidence level of the th record, is the reliability score of the data source to which the th record belongs. The preset values are stress sensor , BIM model , displacement monitoring point , and is the th and the The difference between the timestamp of each record and the current system time, is the timeliness decay factor, is the total number of status parameter records of the same steel structure component, is the reliability score of the data source to which the record belongs; Based on the confidence of each record, compare the confidence with a preset confidence threshold, eliminate the records with a confidence lower than the preset confidence threshold, and sort the remaining records in descending order of confidence to generate steel structure status parameter records with confidence.

[0008] Preferably, the step of obtaining the steel structure status parameter set with marked fusion conflicts is as follows: Extract the records of the same parameter from the steel structure status parameter records with confidence, classify them according to the parameter type, and generate a set of confidence-parameter value pairs to be fused; Based on the set of confidence-parameter value pairs to be fused, calculate the fusion estimated value of the parameter, and the calculation formula is: ; where, is the fusion estimated value of the current parameter type, is the confidence of the th record, is the parameter value of the th record, is the standard deviation of the same type of parameter values, is the total number of records of the current parameter type; Based on the fusion estimated value, calculate the absolute difference between the parameter value of each record and the fusion estimated value, mark the records with an absolute difference exceeding the preset conflict determination limit value as conflict data points, and generate a steel structure status parameter set with marked fusion conflicts.

[0009] Preferably, the step of obtaining the incremental state of the steel structure model to be propagated is as follows: Extract all the data points marked as conflicts from the steel structure status parameter set with marked fusion conflicts, traverse the list of state variable names and identifiers in the steel structure digital twin model, match the parameter names of the conflict data points with the twin model variable names one by one, and generate a corresponding list of state variable names and identifiers to be updated; Based on the corresponding list of state variable names and identifiers to be updated, read the current values of the state variables in the twin model one by one, synchronously extract the corresponding parameter values in the conflict data points, calculate the absolute difference between the current value and the parameter value, and generate a set of incremental state variable parameters; Based on the set of incremental state variable parameters, perform incremental superposition on the matching variables in the steel structure digital twin model, update the variable values one by one, and generate the incremental state of the steel structure model to be propagated.

[0010] Preferably, the steps for obtaining the state of the steel structure twin model and the data dependency log after update propagation are as follows: Extract the incremental values of each main variable from the set of incremental states of the steel structure model to be propagated, traverse the pre-defined index of mechanical transfer and geometric association relationships between various components of the steel structure, match the component identifiers directly associated with the current main variable, and generate a main variable-associated component mapping list; Based on the main variable-associated component mapping list, calculate the propagation influence value of the main variable increment on each associated component. The calculation formula is: ; Where is the propagation influence value of the th associated component, is the incremental value of the main variable, is the th geometric distance between the associated component and the component where the main variable is located, is the number of mechanical transfer path nodes from the main variable to the th associated component, is the distance attenuation coefficient, is the path complexity coefficient; Based on the propagation influence values of all associated components, construct complete dependency entries, write them into the database log table item by item, and generate the state of the steel structure twin model and the data dependency log after update propagation.

[0011] Preferably, the steps for obtaining the log entry point associated with the target result are as follows: Extract the mechanical property evaluation results of all steel structure components from the state of the steel structure twin model after update propagation, traverse the stress, strain, and displacement parameter values of each component, and generate a set of mechanical property evaluation results; Based on the set of mechanical property evaluation results, parse the component identifier, parameter name, and timestamp fields item by item, and perform a full-text matching search in the state of the steel structure twin model and the data dependency log after update propagation using the component identifier and parameter name as the combined key to generate a list of matching log entries; Based on the list of matching log entries, extract the log storage path, data update timestamp, and associated variable identifier fields in each log to generate the log entry point associated with the target result.

[0012] Preferably, the steps for obtaining the causal traceability path of abnormal mechanical properties are as follows: Extract the log storage path of each entry point from the log entry points associated with the target result. Traverse each variable dependency chain record in the updated and propagated steel structure twin model state and data dependency log. Starting from the target result identifier, recursively resolve the upstream variable identifiers in the dependency chain layer by layer to generate a set of reverse trace variable chains; Based on the set of reverse trace variable chains, match the original data records in the time-series aligned steel structure state data stream, BIM model version library, and raw displacement monitoring point database according to the variable identifier and timestamp. Extract the sensor device number, BIM model version number, displacement monitoring point coordinates, and original values to generate a set of original input data identifiers; Merge the set of original input data identifiers and the set of reverse trace variable chains in chronological order and dependency relationship levels to construct a complete link of original data identifiers, intermediate calculation variable identifiers, and target result identifiers, and generate a causal trace path for mechanical property anomalies.

[0013] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In the present invention, the timestamps of multi-source heterogeneous data are corrected for transmission delay through a unified reference time. The original data timestamps are corrected one by one to generate a time-series aligned steel structure state data stream, eliminating the time deviation between sensors, BIM models, and monitoring points, and ensuring the synchronization and coherence of the dynamic changes of mechanical parameters. For conflicting data records of the same component state parameters, a fusion conflict marking mechanism is constructed through confidence value assignment and comparison of fusion estimate differences to distinguish between reliable data and abnormal data, avoiding the problem of error amplification caused by direct fusion without marking conflicts in traditional methods, and improving the reliability and accuracy of the input data of the twin model. Based on the incremental update strategy, only the model variables corresponding to the conflict-marked parameters are locally corrected, reducing the computational resource consumption caused by full-scale updates. At the same time, the mechanical transfer relationship between variables is recorded through the data dependency log to achieve the traceability of abnormal results. Combining log entry points and reverse dependency chain retrieval, the original input data is retroactively located from the abnormal result of the target mechanical property, establishing a causal trace path, clarifying the cause and propagation path of the anomaly, and providing an accurate decision-making basis for construction process optimization. Brief Description of the Drawings

[0014] Figure 1 It is a schematic diagram of the steps of the present invention. Detailed Embodiment

[0015] In order to make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to 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 used to limit the present invention.

[0016] See also Figure 1 The present invention provides a technical solution, a method for evaluating the mechanical properties of a steel structure building during the entire construction process based on digital twins, comprising the following steps: Acquire multi-source heterogeneous steel structure status data from stress sensors, BIM model component information and displacement monitoring points, compare the timestamps of each data source with each other, calculate the transmission delay value of each data source relative to the unified reference time, correct the timestamps of the original data one by one, and establish a time-aligned steel structure status data stream; Based on the time-series aligned steel structure state data stream, data records describing the state parameters of the same steel structure component are extracted, and a confidence value is assigned to each data record to generate steel structure state parameter records with confidence. Based on the steel structure state parameter records with confidence, the fusion estimation value is calculated for the records of the same parameter, and the difference in the fusion estimation values ​​between different records is compared with the preset conflict judgment limit to construct a steel structure state parameter set with fusion conflicts marked. Based on the steel structure state parameter set marked by the fusion conflict, the state variables directly corresponding to the steel structure state parameter set marked by the fusion conflict in the steel structure digital twin model are retrieved, and the variables are incrementally updated to obtain the state increment of the steel structure model to be propagated. Based on the state increment of the steel structure model to be propagated, the influence of the state increment on the associated variables is calculated, and the state and data dependency log of the steel structure twin model after the update propagation is established; Retrieve the target mechanical properties evaluation results in the steel structure twin model state after update propagation, locate the corresponding record entries of the mechanical properties evaluation results in the data dependency log, obtain the log entry point associated with the target result, and perform reverse tracing retrieval in the data dependency log based on the recorded variable dependency chain information based on the log entry point associated with the target result to generate a causal tracing path for mechanical property anomalies.

[0017] The steps for obtaining the time-aligned steel structure status data stream are as follows: Extract the original timestamp of each data source from the stress sensor, BIM model component information, and displacement monitoring point, calculate the absolute value of the difference between the timestamp of each data source and the reference time based on the unified reference time, define the absolute value of the difference as the initial transmission delay value, and generate the initial transmission delay value set; Based on the initial transmission delay value set, the initial transmission delay value distribution of all data sources is counted, and the 90% quantile of the initial transmission delay value distribution is taken as the preset time deviation threshold. The compensation factor is generated according to the ratio of the time deviation threshold to the data source delay value. The compensation factor calculation formula is: compensation factor = data source delay value / time deviation threshold, and a dynamic time compensation factor set is generated; Based on the set of dynamic time compensation factors, for data sources where the initial transmission delay value exceeds the time deviation threshold, the formula is used: corrected timestamp = original timestamp + (data source delay value / (compensation factor + 1)), and the timestamps are corrected one by one to generate a steel structure status data stream with time series alignment.

[0018] Specifically, based on the original timestamps of each data source extracted from stress sensors, BIM model component information, and displacement monitoring points. For example, the original timestamp of stress sensor S1 is 2025-04-15 10:00:01.500, the information timestamp of component B-101 in the BIM model is 2025-04-15 10:00:00.800, and the timestamp of displacement monitoring point D3 is 2025-04-15 10:00:02.100. Set a unified reference time, which can be the startup time of the data acquisition system or a preset synchronization time point, for example, set as 2025-04-15 10:00:00.000. Next, calculate the absolute value of the difference between the original timestamp of each data source and this reference time (10:00:00.000). For stress sensor S1, the absolute value of the difference is |10:00:01.500 - 10:00:00.000| = 1.500 seconds. For BIM model information, the absolute value of the difference is |10:00:00.800 - 10:00:00.000| = 0.800 seconds. For displacement monitoring point D3, the absolute value of the difference is |10:00:02.100 - 10:00:00.000| = 2.100 seconds. Define these calculated absolute values of the differences as the initial transmission delay values of their respective data sources. For example, the initial transmission delay of S1 is 1.500 seconds, BIM is 0.800 seconds, and D3 is 2.100 seconds. Collect the initial transmission delay values of all participating data sources to form an initial transmission delay value set. For example, if there is also a sensor S2 with a delay of 3.000 seconds and a displacement point D4 with a delay of 1.200 seconds, then the set is {1.500, 0.800, 2.100, 3.000, 1.200}. Based on this set {1.500, 0.800, 2.100, 3.000, 1.200} containing the initial transmission delay values of multiple data sources, statistically analyze the distribution of these values. For example, calculate its cumulative distribution function and determine the 90th percentile of the distribution. First, sort the set: {0.800, 1.200, 1.500, 2.100, 3.000}. For a set with N = 5 data points, the position of the 90th percentile is P = 90 / 100 * (N + 1) = 0.9 * 6 = 5.4. Since the position 5.4 is between the 5th value (3.000) and, for example, the 6th value, usually interpolation or taking the nearest rank is used. In this simple example or for large data sets, the value at a specific position can be directly taken or calculated using a standard statistical library. For example, the 90th percentile obtained through standard calculation is 2.500 seconds. Take this 90th percentile (2.Set the time deviation threshold (2.500 seconds) to a preset value, which is set with reference to the transmission delay levels of most (90%) of the data in the dataset. The 90th percentile is selected to tolerate a certain degree of network fluctuations while identifying significant delay anomalies. Next, generate respective compensation factors based on the ratio of the time deviation threshold (2.500 seconds) to the actual delay values of each data source. The calculation formula for the compensation factor is: Compensation factor = Data source delay value / Time deviation threshold. Calculate the compensation factors for each data source. The compensation factor of S1 = 1.500 / 2.500 = 0.6, the compensation factor of BIM = 0.800 / 2.500 = 0.32, the compensation factor of D3 = 2.100 / 2.500 = 0.84, the compensation factor of S2 = 3.000 / 2.500 = 1.2, the compensation factor of D4 = 1.200 / 2.500 = 0.48. Pool all the calculated compensation factors to form a dynamic time compensation factor set {0.6, 0.32, 0.84, 1.2, 0.48}. Based on this dynamic time compensation factor set, check whether the initial transmission delay value of each data source exceeds the preset time deviation threshold (2.500 seconds). It is found that S1 (1.500s), BIM (0.800s), D3 (2.100s), and D4 (1.200s) do not exceed the threshold, while the delay of S2, which is 3.000 seconds, exceeds the threshold. Therefore, only the timestamp of S2 is corrected. Use the correction formula: Corrected timestamp = Original timestamp + (Data source delay value / (Compensation factor + 1)). Apply this formula to S2. Its original timestamp is, for example, 10:00:03.000, the delay value is 3.000 seconds, and the compensation factor is 1.2. Then the corrected timestamp = 10:00:03.000 + (3.000 / (1.2 + 1)) = 10:00:03.000 + (3.000 / 2.2) = 10:00:03.000 + 1.364 seconds = 10:00:04.364. Do not correct the timestamps of other data sources (S1, BIM, D3, D4) that do not exceed the threshold, and keep their original timestamps. Process all data sources one by one, and merge the corrected timestamps (such as 10:00:04.364 of S2) with the uncorrected timestamps (such as 10:00:01.500 of S1, 10:00:00.800 of BIM, 10:00:02.100 of D3, 10:00:01.200 of D4) to generate a time-series aligned steel structure status data stream.

[0019] The steps for obtaining steel structure status parameter records with confidence are as follows: Extract multiple status parameter data records of the same steel structure component from the time-series aligned steel structure status data stream, classify them according to the data source type, and generate a set of status parameter records without confidence assignment; Based on the set of state parameter records without confidence, the confidence of each record is calculated using the following formula: ; in, For the The confidence level of the records, For the The reliability score of the data source to which the record belongs. The default value is the stress sensor. , BIM model , displacement monitoring point , and For the Article and The difference between the timestamp of the record and the current system time, is the time-dependent attenuation factor, The total number of state parameter records for the same steel structure component, For the The reliability score of the data source to which the record belongs; Based on the confidence of each record, the confidence is compared with the preset confidence threshold, and the records with a confidence level less than the preset confidence threshold are eliminated. The retained records are sorted in descending order according to the confidence level to generate steel structure state parameter records with confidence levels.

[0020] Specifically, from the time-aligned steel structure state data stream generated in the previous step, multiple state parameter data records for the same steel structure component (for example, the component identifier is "beam B-101") are extracted. These records may come from different sensors or data sources. For example, four records about the stress state of "beam B-101" are extracted: record 1 (source: stress sensor S1, parameter value: 150MPa, aligned timestamp: 10:00:01.500), record 2 (source: BIM model, parameter value: 145MPa, aligned timestamp: 10:00:00.800), Record 3 (source: estimated stress of displacement monitoring point D3, parameter value: 155MPa, timestamp after alignment: 10:00:02.100), record 4 (source: stress sensor S2, parameter value: 152MPa, timestamp after alignment: 10:00:04.364), these extracted records are classified and sorted according to their data source type (stress sensor, BIM model, displacement monitoring point), forming an unconfident state parameter record set {record 1, record 2, record 3, record 4}, based on this unconfident state parameter record set, the confidence of each record is calculated one by one , and its calculation formula is: ; here, Representative The confidence of a record is a value between 0 and 1, indicating the credibility of the record. It represents the inherent reliability score of the data source to which the record belongs. This is a preset value that reflects the general accuracy of different types of data sources. According to the preset, the reliability score of stress sensors (such as S1, S2) is 0.9, and that of BIM model information is 0.7, and that of the calculation results of displacement monitoring points (such as D3) is 0.8. Therefore, , , , , and respectively represent the absolute value of the difference between the timestamp of the record and the current system time at the time of evaluation, representing the freshness of the data. For example, if the current system time is 10:00:05.000, then seconds, seconds, seconds, seconds, seconds, seconds, is the timeliness decay factor, which controls the rate at which the data confidence decreases over time. The preset is s. This value is set based on the fact that structural health monitoring data usually still has high reference value within seconds to minutes. A decay factor of 0.1 means that for every 10-second increase in the time difference, the time weight factor will decrease by approximately 63%. is the total number of state parameter records of the same steel structure component (beam B-101) being processed currently. Here , is the reliability score of the data source corresponding to the th record in the summation term. The logic of the formula is to calculate the weight of a single record through the product of the data source reliability and the timeliness factor , and then normalize it by dividing by the sum of the weights of all records to obtain the relative confidence of each record. First, calculate the denominator, which is the sum of the weights of all records : The first term ( ): ; The second term ( ): ; The third term ( ): ; The fourth term ( ) ; Denominator = ; Then calculate the confidence of each record : ; ; ; ; The advantage of the formula is that by combining the inherent reliability of the data source and the timeliness weight that exponentially decays over time , and performing normalization processing, it is possible to dynamically evaluate the relative credibility of each data record at the current moment, providing a quantitative basis for subsequent data fusion and conflict detection. Based on the calculated confidence of each record {0.2500, 0.1813, 0.2359, 0.3329}, compare these confidence values with a preset confidence threshold. Set the preset confidence threshold to 0.15. This threshold is set to filter out those records with very low comprehensive confidence due to low source reliability or overly old data. The reference standard is to retain the meaningful data whose confidence accounts for more than 15% of the total (this ratio can be adjusted according to the application). Compare the confidence of each record with 0.15 and find that the confidence of all records (0.2500, 0.1813, 0.2359, 0.3329) is greater than 0.15. Therefore, all records are retained. Sort the retained records {Record 1, Record 2, Record 3, Record 4} in descending order according to their confidence . The sorting result is: Record 4 ( ), Record 1 ( ), Record 3 ( ), Record 2 ( ), generating a record of steel structure state parameters with confidence. This result (for example ) indicates that Record 4 is considered the most credible stress data at the current evaluation time point because it comes from a highly reliable stress sensor and has the latest timestamp, while Record 2 has the lowest confidence because its source is a BIM model with relatively low reliability and an earlier timestamp. These confidence values will be used as weights in the next step of data fusion.

[0021] The steps to obtain the set of steel structure state parameters with marked fusion conflicts are as follows: Extract the records of the same parameter from the records of steel structure state parameters with confidence, classify them according to the parameter type, and generate a set of confidence-parameter value pairs to be fused; Based on the set of confidence - parameter value pairs to be fused, calculate the fused estimated value of the parameter. The calculation formula is as follows: ; Wherein, is the fused estimated value of the current parameter type, is the confidence of the th record, is the parameter value of the th record, is the standard deviation of the same - type parameter values, is the total number of records of the current parameter type; Based on the fused estimated value, calculate the absolute difference between the parameter value of each record and the fused estimated value. Mark the records with absolute differences exceeding the preset conflict determination limit value as conflict data points, and generate a set of steel structure state parameters with fused conflicts marked.

[0022] Specifically, from the steel structure state parameter records with confidence generated in the previous step, extract all records for the same specific parameter (for example, the maximum Y - direction stress of beam B - 101), classify them according to the parameter type (here it is "maximum Y - direction stress"), and organize these records into a set of confidence - parameter value pairs to be fused. Using the sorted result: {(Record 4: , MPa), (Record 1: , MPa), (Record 3: , MPa), (Record 2: , MPa)}, which contains records. Based on this set of confidence - parameter value pairs to be fused, calculate the fused estimated value of this parameter (maximum Y - direction stress). The calculation formula is as follows: ; Here, is the fused estimated value of the current parameter type (maximum Y - direction stress), is the confidence of the th record, is the parameter value of the th record (unit: MPa), is the standard deviation of the same - type parameter values (i.e., {152, 150, 155, 145} MPa), is the total number of records of the current parameter type. Here . The logic of this formula is to calculate the sum of confidence - weighted parameter values through the numerator and perform normalization through the denominator. The denominator takes into account the sum of the squares of the confidences and the dispersion degree of the parameter values themselves (variance ), first calculate the standard deviation of the parameter values : The set of parameter values is {152, 150, 155, 145}, and the average value MPa, calculate the sample variance , the standard deviation MPa, next calculate the numerator in the formula : Item 1( ): ; Item 2( ): ; Item 3( ): ; Item 4( ): ; Numerator = ; Then calculate the inside the square root of the denominator in the formula Item 1( ): ; Item 2( ): ; Item 3( ): ; Item 4( ): ; ; Now calculate the denominator : Denominator = ; Finally, calculate the fusion estimated value : MPa; The advantage of the formula is that its structural design attempts to combine the confidence distribution and the consistency of the data itself to obtain the fusion result. The term in the denominator makes the fusion result be adjusted when the original data has a large dispersion, while the term reflects the distribution of the confidence itself. Based on the calculated fusion estimated value MPa, calculate the absolute difference between the parameter value of each record and the fusion estimated value : : Record 4: MPa; Record 1: MPa; Record 3: MPa; Record 2: MPa; Compare these absolute differences with a preset conflict determination limit value. Set the conflict determination limit value. For example, set it to , and the setting of this limit value refers to the multiple of the standard deviation commonly used in statistics to judge outliers. Here, taking 2 times the standard deviation is intended to identify data points with large deviations from the fusion result. The limit value = MPa. Compare each absolute difference with the limit value of 8.406 MPa: 116.352 > 8.406, 114.352 > 8.406, 119.352 > 8.406, 109.352 > 8.406. The absolute differences between the parameter values of all records and the fusion estimated values have exceeded the conflict determination limit value. Therefore, mark Record 1, Record 2, Record 3, and Record 4 as conflict data points. This result ( MPa, and all points are marked as conflicts) indicates that there are significant inconsistencies within the current data set according to the specific fusion formula and conflict determination rules used, or the fusion formula produces estimated values that deviate from the data center in this scenario. All original records are considered to conflict with this fusion result, generating a set of steel structure state parameters with marked fusion conflicts, which includes the original record information and their conflict marks.

[0023] The steps to obtain the state increment of the steel structure model to be propagated are as follows: Extract all data points marked as conflicts from the set of steel structure state parameters with marked fusion conflicts. Traverse the list of state variable names and identifiers in the steel structure digital twin model, and match the parameter names of the conflict data points with the variable names in the twin model one by one to generate a corresponding list of state variable names and identifiers to be updated; Based on the corresponding list of state variable names and identifiers to be updated, read the current values of the state variables in the twin model one by one, synchronously extract the corresponding parameter values in the conflict data points, calculate the absolute differences between the current values and the parameter values, and generate a set of state variable increment parameters; Based on the set of state variable increment parameters, perform incremental superposition on the matching variables in the steel structure digital twin model, update the variable values one by one, and generate the state increment of the steel structure model to be propagated.

[0024] Specifically, from the set of steel structure state parameters with marked fusion conflicts obtained in the previous stage, all the data points marked as conflicts are extracted. According to the records of the maximum Y-direction stress parameter of beam B-101 (records 1, 2, 3, 4), all of them are marked as conflicts. These conflict data points are: (record 4: 152 MPa, marked: conflict), (record 1: 150 MPa, marked: conflict), (record 3: 155 MPa, marked: conflict), (record 2: 145 MPa, marked: conflict). Next, traverse the list of predefined state variable names and identifiers in the digital twin model of the steel structure. This list maintains the correspondence between physical world parameters and internal variables of the twin model. For example, the list contains an entry {state variable name: "maximum Y-direction stress", identifier: "DT_B101_StressY", belonging component: "B-101"}. Match the parameter name ("maximum Y-direction stress") of the conflict data point with the variable name in the twin model ("maximum Y-direction stress") item by item. When both the parameter name and the belonging component (implied in the record source, for example, all belong to B-101) match, establish a correspondence and generate a corresponding list of state variable names and identifiers to be updated. In this example, the list is {("maximum Y-direction stress", "DT_B101_StressY")}. Based on this corresponding list of state variable names and identifiers to be updated {("maximum Y-direction stress", "DT_B101_StressY")}, read the current values of the corresponding state variables in the twin model item by item, access the digital twin model database or memory, query the variable with the identifier "DT_B101_StressY", and obtain its currently stored value. For example, the current value is 148 MPa. Synchronously, extract the parameter values corresponding to this variable from the set of conflict data points. Since there are multiple conflict points (152, 150, 155, 145 MPa), a rule is needed to select the value used for calculating the increment. The rule can be to select the value of the conflict point with the highest confidence (152 MPa), or the value of the latest conflict point (depending on the timestamp, for example, record 4 is the latest, value is 152 MPa), or the average value of all conflict point values ((152 + 150 + 155 + 145) / 4 = 150.5 MPa), or use the fusion value F calculated in the previous step (although all points are in conflict, F itself may still be used as the target value, that is, 35.648 MPa), for example, using the value of the conflict point with the highest confidence level, i.e., 152 MPa of record 4, calculate the absolute difference between the current value (148 MPa) in the twin model and the parameter value of the selected conflict data point (152 MPa). The absolute difference is |148 - 152| = 4 MPa, or calculate the signed difference as an increment: increment = conflict value - current value = 152 - 148 = +4 MPa. Generate a set of state variable increment parameters, which contains the adjustment amounts required for the twin model variables. For example, {("DT_B101_StressY", +4 MPa)}. Based on this set of state variable increment parameters {("DT_B101_StressY", +4 MPa)}, perform an incremental superposition operation on the matching variables in the steel structure digital twin model. Access the variable with the identifier "DT_B101_StressY", add the calculated increment (+4 MPa) to its current value (148 MPa), and the updated variable value is 148 + 4 = 152 MPa. Update the values of all variables that need to be adjusted one by one. After the update is completed, a set of state increments of the steel structure model to be propagated is formed. In this example, the state increment of the variable "DT_B101_StressY" is +4 MPa, and its new state is 152 MPa.

[0025] The steps for obtaining the updated and propagated state of the steel structure twin model and the data dependency log are as follows: Extract the increment values of each main variable from the set of state increments of the steel structure model to be propagated. Traverse the pre-defined mechanical transfer and geometric association relationship index between each component of the steel structure, match the component identifiers directly associated with the current main variable, and generate a main variable - associated component mapping list; Based on the main variable - associated component mapping list, calculate the propagation influence value of the main variable increment on each associated component. The calculation formula is: ; where, is the propagation influence value of the th associated component, is the increment value of the main variable, is the th geometric distance between the associated component and the component where the main variable is located, is the number of mechanical transfer path nodes from the main variable to the th associated component, is the distance attenuation coefficient, is the path complexity coefficient; Based on the propagation influence values of all associated components, construct complete dependency entries, write them into the database log table one by one, and generate the updated and propagated state of the steel structure twin model and the data dependency log.

[0026] Specifically, from the set of state increments of the steel structure model to be propagated generated in the previous step, extract the increment values of each main variable (i.e., the variable that has undergone direct update). In this example, the main variable is , and its increment value is MPa. Next, traverse the pre-defined index of the mechanical transfer and geometric association relationships between the various components of the steel structure. This index library (e.g., stored in the form of a graph database or a relational table) describes the connection methods, distances, and mechanical influence paths between components. For example, the index library indicates that component B-101 (the component where the main variable is located) has direct mechanical associations and geometric proximity relationships with component C-05 (a column) and component B-102 (an adjacent beam). According to the index, match the component identifiers that are directly associated with the current main variable (located in B-101), identify the associated components as C-05 and B-102, and generate a main variable-associated component mapping list. For example: . Based on this main variable-associated component mapping list, calculate the increment of the main variable for the propagation influence value of each associated component ( ). The calculation formula is: ; Here, is the propagation influence intensity generated by the increment of the main variable on the rd associated component, is the increment value of the main variable, is the geometric distance between the th associated component and the component where the main variable is located, which needs to be queried from the BIM model or geometric database. For example, the central distance from B-101 to C-05 is meters (e.g., directly connected), and the distance from B-101 to B-102 is meters. is the number of nodes (including connection points or intermediate components) passed through on the main mechanical transfer path from the component where the main variable is located to the th associated component, which needs to be determined through structural model analysis. For example, for the direct force transfer from B-101 to C-05, the number of path nodes is , and for the transfer from B-101 to B-102 through a certain connection node, the number of path nodes is . is the distance attenuation coefficient, which controls the attenuation rate of the influence as the distance increases. Set m. The setting of this coefficient is based on engineering experience or simulation results, indicating that the influence intensity decreases exponentially with distance. A value of 0.5 means that for every 2-meter increase in distance, the influence factor decreases by approximately 63%. is the path complexity coefficient, reflecting the effect that the more complex the mechanical transmission path is (the more nodes it passes through), the greater the attenuation of the influence. It is set , and the setting of this coefficient is also based on experience or calibration. A value of 0.2 means that for each additional transmission node, the denominator increases by 0.2, thereby reducing the influence value. The logic of the formula is: the influence value is proportional to the original increment, decays exponentially with distance, and decays algebraically with the increase in path complexity. Calculate the propagation influence value on the associated component C-05 : ; Calculate the propagation influence value on the associated component B-102 : ; The benefit of the formula is that it quantifies the propagation effect of state changes within the structure based on geometric distance and mechanical path, simulates the locality and attenuation of the influence, and provides a basis for understanding the chain reaction and conducting more accurate global state assessment. Based on the calculated propagation influence values of all associated components {C-05: +3.171, B-102: +0.637}, a complete dependency entry is constructed for each influence propagation. The entry content should include: source variable identifier ("DT_B101_StressY"), source increment value (+4), target component identifier ("C-05" or "B-102"), calculated influence value (+3.171 or +0.637), parameters used in the calculation ( , , , ), as well as the timestamp and the storage path of the log itself. These dependency entries are written one by one into the specified database log table, for example, into a table named "DependencyLog", to generate the state and data dependency log of the steel structure twin model after the update propagation. This result (for example ) represents the influence strength of the stress (or other related states) that is expected to be generated on C-05 after the stress of B-101 increases by 4 MPa, which is approximately +3.171 units. These logs record the propagation path and strength of the state changes, which are key information for subsequent problem tracing.

[0027] The steps to obtain the log entry point associated with the target result are as follows: Extract the mechanical property evaluation results of all steel structure components from the state of the steel structure twin model after the update propagation, traverse the stress, strain, and displacement parameter values of each component, and generate a set of mechanical property evaluation results; Based on the set of mechanical property evaluation results, parse the component identifier, parameter name, and timestamp fields one by one. In the updated and propagated steel structure twin model state and data dependency log, perform a full-text matching search using the component identifier and parameter name as the combined key to generate a list of matching log entries; Based on the list of matching log entries, extract the log storage path, data update timestamp, and associated variable identifier fields from each log to generate the log entry point associated with the target result.

[0028] Specifically, extract the mechanical property evaluation results of all steel structure components from the updated and propagated steel structure twin model state. This evaluation may be based on the state values after model update (e.g., the stress of B-101 is updated to 152 MPa, and the states of C-05 and B-102 may also be based on the influence values It is carried out with corresponding adjustments or markings. Traverse the key performance parameter values such as stress, strain, and displacement of each key component (such as B-101, C-05, B-102, etc.). For example, the evaluation system detects that the maximum Y-direction stress value of component B-101 is 152 MPa, the timestamp is 2025-04-15 10:00:15.000, and this value exceeds the preset safety threshold or is in the warning range. Mark it as a mechanical property anomaly (or potential anomaly) result. Collect the evaluation results of all components to form a mechanical property evaluation result set, which may contain {component: "B-101", parameter: "maximum Y-direction stress", value: 152 MPa, timestamp: "2025-04-15 10:00:15.000", status: "warning"}, {component: "C-05", parameter: "axial stress", value: 98 MPa, timestamp: "2025-04-15 10:00:15.000", status: "normal"},..., Based on this mechanical property evaluation result set, focus on the results marked as abnormal or requiring attention. For example, the "maximum Y-direction stress" warning result of "B-101". Analyze the component identifier ("B-101"), parameter name ("maximum Y-direction stress"), and timestamp ("2025-04-15 10:00:15.000") fields in these target result records one by one. Use these parsed fields to perform a retrieval in the "Updated Propagation of Steel Structure Twin Model Status and Data Dependency Log" (stored in a database table such as "DependencyLog"). Use the component identifier ("B-101") and parameter name ("maximum Y-direction stress") as the combined query key, and possibly limit the range in combination with the timestamp to perform a matching retrieval operation. This retrieval aims to find the relevant log entries that record the cause or influence of the change in the "maximum Y-direction stress" status on "B-101". For example, by querying WHERE TargetComponent='B-101' AND TargetParameter='maximum Y-direction stress' or WHERE SourceVariable LIKE '%B101_StressY%' (depending on the log table structure) and the timestamp is close to "2025-04-15 10:00:15.000", the log entries that record the update of this variable and the entries that record the influence propagated from this variable can be found. For example, find the following relevant log entries: Table 1: Matching Log Entry Table:

[0029] As shown in Table 1, the table lists the log entry information related to the target result (B-101 maximum Y-direction stress warning) found through retrieval. Based on this list of matching log entries (as shown in Table 1), extract the key fields from each log record: the log storage path (e.g., " / logs / update / 20250415_100010_B101.log"), the data update timestamp (e.g., "2025-04-15 10:00:10.000"), and the associated variable identifier (e.g., "DT_B101_StressY"). Combine this extracted information to form an entry point pointing to a specific log record. These entry points are the starting points for subsequent causal tracing, generating a set of log entry points associated with the target result, such as {(LogPath: " / logs / update / 20250415_100010_B101.log", Timestamp: "10:00:10.000", VariableID: "DT_B101_StressY")}. This entry point directly points to the log event where the target abnormal state variable (DT_B101_StressY) was updated.

[0030] The steps to obtain the causal tracing path of mechanical property anomalies are as follows: Extract the log storage path of each entry point from the log entry points associated with the target result. Traverse each variable dependency chain record in the state of the steel structure twin model and the data dependency log after update propagation. Starting from the target result identifier, reverse-resolve the upstream variable identifiers in the dependency chain layer by layer to generate a set of reverse-tracing variable chains; Based on the set of reverse-tracing variable chains, match the original data records in the time-series aligned steel structure state data stream, BIM model version library, and displacement monitoring point original database according to the variable identifier and timestamp, and extract the sensor device number, BIM model version number, displacement monitoring point coordinates, and original values to generate a set of original input data identifiers; Merge the set of original input data identifiers and the set of reverse-tracing variable chains in chronological order and dependency relationship levels to construct a complete link of original data identifiers, intermediate calculation variable identifiers, and target result identifiers, generating the causal tracing path of mechanical property anomalies.

[0031] Specifically, from the set of log entry points associated with the target result obtained in the previous step, extract the log storage path of each entry point. For example, obtain the path " / logs / update / 20250415_100010_B101.log" in the entry point {LogPath: " / logs / update / 20250415_100010_B101.log", Timestamp: "10:00:10.000", VariableID: "DT_B101_StressY"}, access and parse the content of this log file, or query the corresponding log record in the database, which details the update event of the variable "DT_B101_StressY" at time "10:00:10.000", including its upstream dependencies. For example, the log shows that this update is an increment of "+4MPa" calculated based on the conflicting data points "Rec3" and "Rec2". Next, traverse all the variable dependency chain records stored in the "Steel Structure Twin Model State and Data Dependency Log after Update Propagation", starting from the target result identifier "DT_B101_StressY" and its associated update event, and begin to reverse-parse the dependency relationship from the log " / logs / update / 20250415_100010_B101.In "log", it is recognized that the upstream dependencies are data records "Rec3" and "Rec2". Continuing to trace the sources of "Rec3" and "Rec2", this requires querying the logs or metadata of the record processing process, or reverse looking up the previous steps based on the record ID. For example, it is found that "Rec3" comes from the confidence calculation step, and its input is the displacement sensor D3 data after time series alignment, while "Rec2" comes from the confidence calculation, and its input is the BIM model data after time series alignment. Continuing to trace backward, the input of the time series alignment step is the raw data. Through this process, the upstream variable identifiers or data record IDs in the dependency chain are parsed layer by layer in reverse, forming a set of variable chains that trace back from the target result to the intermediate calculation steps and the original inputs. For example, a reverse trace chain is generated: {"DT_B101_StressY"@10:00:10 <- ["Rec3", "Rec2"]@~10:00:08 <- [AlignedD3data, AlignedBIMdata]@~10:00:04 <- [RawD3data, RawBIMdata]}. Based on the raw data links (RawD3data, RawBIMdata) identified in this set of reverse trace variable chains, and their corresponding approximate timestamps or processing record IDs, query the corresponding raw databases, including the steel structure status data stream before time series alignment (including the original sensor readings), the BIM model version library, and the raw database of the displacement monitoring points. Match according to the variable identifiers (such as sensor ID "D3", component ID "B-101") and the timestamp range to find the corresponding raw data records. For example, query the data near time 10:00:02.100 in the displacement monitoring point D3 database and find the record {Sensor device number: "D3", Coordinates: (10.5, 20.1, 15.0), Original timestamp: "10:00:02.050", Original value: 5.1mm}. Query the information about component B-101 near time 10:00:00.800 in the BIM model version library and find the record {BIM model version number: "v2.1", Component ID: "B-101", Attribute: "Calculated stress", Original value: 145MPa, Original timestamp: "10:00:00.750"}. Extract the key identification information of these raw data records, such as the sensor device number, BIM model version number, displacement monitoring point coordinates, and original value, to form a set of original input data identifiers {"SensorID: D3, Time: 10:00:02.050, Value: 5.1mm", "BIMVersion: v2.1, Component: B-101, Property: Stress, Value: 145MPa, Time: 10:00:00.Merge and organize the original input data identifier set {"SensorID: D3...", "BIMVersion: v2.1..."} with the previously constructed reverse traceability variable chain set {"DT_B101_StressY" <- ["Rec3", "Rec2"] <-...} according to time and dependency hierarchy levels, and construct a complete link from the original data input, through intermediate data processing (time series alignment, confidence calculation, fusion, conflict detection, incremental calculation, update propagation) to the final target result ("DT_B101_StressY" status abnormal). This link clearly shows the data flow and calculation dependency relationship, and generates the causal traceability path of mechanical property abnormality.

[0032] The above are only the preferred embodiments of the present invention, and do not limit the present invention in other forms. Any person skilled in the art may use the disclosed technical content to make changes or modifications into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change, and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution content of the present invention still fall within the protection scope of the technical solution of the present invention.

Claims

1. A method for evaluating the mechanical properties of the whole process of steel structure building construction based on digital twin, characterized in that, The following steps are involved: Acquire multi-source heterogeneous steel structure status data from stress sensors, BIM model component information and displacement monitoring points, compare the timestamps of each data source with each other, calculate the transmission delay value of each data source relative to the unified reference time, correct the timestamps of the original data one by one, and establish a time-aligned steel structure status data stream; Based on the time-series aligned steel structure state data stream, extract data records describing the state parameters of the same steel structure component, assign a confidence value to each data record, generate steel structure state parameter records with confidence, calculate fusion estimation values ​​for records of the same parameter based on the steel structure state parameter records with confidence, compare the fusion estimation value differences between different records with a preset conflict judgment limit, and construct a steel structure state parameter set with fusion conflict marked; Based on the steel structure state parameter set marked by the fusion conflict, the state variables directly corresponding to the steel structure state parameter set marked by the fusion conflict in the steel structure digital twin model are retrieved, the variables are incrementally updated to obtain the steel structure model state increment to be propagated, based on the steel structure model state increment to be propagated, the influence of the state increment on the associated variables is calculated, and the steel structure twin model state and data dependency log after the update propagation is established; The target mechanical property evaluation result is retrieved in the steel structure twin model state after the update propagation, the corresponding record entry of the mechanical property evaluation result in the data dependency log is located, and the log entry point associated with the target result is obtained. Based on the log entry point associated with the target result, a reverse traceability search is performed in the data dependency log according to the recorded variable dependency chain information to generate a causal traceability path for the mechanical property anomaly.

2. The method for evaluating the mechanical properties of the whole process of steel structure building construction based on digital twin according to claim 1, wherein The steps for acquiring the time-aligned steel structure status data stream are: Extract the original timestamp of each data source from the stress sensor, BIM model component information, and displacement monitoring point, calculate the absolute value of the difference between the timestamp of each data source and the reference time based on the unified reference time, define the absolute value of the difference as the initial transmission delay value, and generate the initial transmission delay value set; Based on the initial transmission delay value set, the initial transmission delay value distribution of all data sources is counted, and the 90% quantile of the initial transmission delay value distribution is taken as the preset time deviation threshold. The compensation factor is generated according to the ratio of the time deviation threshold to the data source delay value. The compensation factor calculation formula is: compensation factor = data source delay value / time deviation threshold, and a dynamic time compensation factor set is generated; Based on the dynamic time compensation factor set, for the data source whose initial transmission delay value exceeds the time deviation threshold, the formula: corrected timestamp = original timestamp + (data source delay value / (compensation factor + 1)) is used to correct the timestamps one by one and generate a time-aligned steel structure status data stream.

3. The method for evaluating the mechanical properties of the whole process of steel structure building construction based on digital twin according to claim 1, characterized in that, The steps for obtaining the steel structure state parameter record with confidence are: Extracting multiple state parameter data records of the same steel structure component from the time-series aligned steel structure state data stream, classifying them according to data source type, and generating a set of unconfident state parameter records; Based on the set of state parameter records without confidence assigned, calculate the confidence of each record. The calculation formula is as follows: ； wherein, is the confidence level of the th record, is the reliability score of the data source to which the th record belongs, and the preset value is the stress sensor , the BIM model , the displacement monitoring point , and are the differences between the timestamps of the th and th records and the current system time, is the timeliness decay factor, is the total number of status parameter records of the same steel structure component, is the reliability score of the data source to which the th record belongs; Based on the confidence of each record, compare the confidence with a preset confidence threshold, eliminate the records with a confidence lower than the preset confidence threshold, and sort the remaining records in descending order of confidence to generate a set of steel structure state parameter records with confidence.

4. The method for evaluating the mechanical properties of the whole process of steel structure building construction based on digital twin according to claim 1, wherein The steps for obtaining the set of steel structure state parameters with marked fusion conflicts are as follows: Extract the records of the same parameter from the set of steel structure state parameter records with confidence, classify them according to the parameter type, and generate a set of confidence-parameter value pairs to be fused. Based on the set of confidence-parameter value pairs to be fused, calculate the fusion estimated value of the parameter. The calculation formula is as follows: ； Among them, is the fusion estimated value of the current parameter type, is the confidence level of the th record, is the parameter value of the th record, is the standard deviation of the parameter values of the same type, is the total number of records of the current parameter type; Based on the fusion estimated value, calculate the absolute difference between the parameter value of each record and the fusion estimated value, mark the records with an absolute difference exceeding the preset conflict determination limit as conflict data points, and generate a set of steel structure state parameters with marked fusion conflicts.

5. The method for evaluating the mechanical properties of the whole process of steel structure building construction based on digital twin according to claim 1, characterized in that The steps for obtaining the increment of the steel structure model state to be propagated are as follows: Extract all the data points marked as conflicts from the set of steel structure state parameters with marked fusion conflicts, traverse the list of state variable names and identifiers in the steel structure digital twin model, match the parameter names of the conflict data points with the variable names in the twin model one by one, and generate a corresponding list of state variable names to be updated and identifiers. Based on the corresponding list of state variable names to be updated and identifiers, read the current values of the state variables in the twin model one by one, synchronously extract the corresponding parameter values in the conflict data points, calculate the absolute difference between the current value and the parameter value, and generate a set of state variable increment parameters. Based on the set of state variable increment parameters, perform incremental superposition on the matching variables in the steel structure digital twin model, update the variable values one by one, and generate the increment of the steel structure model state to be propagated.

6. The method for evaluating the mechanical properties of the whole process of steel structure building construction based on digital twin according to claim 1, wherein The steps for obtaining the state and data dependency log of the steel structure twin model after update propagation are as follows: Extract the incremental values of each main variable from the set of increments of the steel structure model state to be propagated, traverse the predefined index of mechanical transfer and geometric correlation relationships among various components of the steel structure, match the component identifiers directly associated with the current main variable, and generate a mapping list of main variable-associated components. Based on the mapping list of main variable-associated components, calculate the propagation influence value of the main variable increment on each associated component. The calculation formula is as follows: ； Among them, is the propagation influence value of the th associated component, is the incremental value of the main variable, is the th geometric distance between the associated component and the component where the main variable is located, is the number of mechanical transmission path nodes from the main variable to the th associated component, is the distance attenuation coefficient, is the path complexity coefficient; Based on the propagation influence values of all associated components, construct complete dependency entries, write them into the database log table one by one, and generate the state and data dependency log of the steel structure twin model after update propagation.

7. The method for evaluating the mechanical properties of the whole process of steel structure building construction based on digital twin according to claim 1, characterized in that, The steps for obtaining the log entry point associated with the target result are as follows: Extract the mechanical property evaluation results of all steel structure components from the state of the steel structure twin model after update propagation, traverse the stress, strain, and displacement parameter values of each component, and generate a set of mechanical property evaluation results. Based on the set of mechanical property evaluation results, parse the component identifier, parameter name, and timestamp fields one by one, and perform a full-text matching search in the state and data dependency log of the steel structure twin model after update propagation using the component identifier and parameter name as the combined key to generate a list of matching log entries. Based on the list of the matching log entries, extract the log storage path, data update timestamp, and associated variable identifier field from each log, and generate a log entry point associated with the target result.

8. The method for evaluating the mechanical properties of the whole process of steel structure building construction based on digital twin according to claim 1, characterized in that, The steps for obtaining the causal traceability path of the abnormal mechanical properties are as follows: Extract the log storage path of each entry point from the log entry point associated with the target result, traverse each variable dependency chain record in the state of the steel structure twin model and the data dependency log after the update propagation, and starting from the target result identifier, reverse-resolve the upstream variable identifiers in the dependency chain layer by layer to generate a set of reverse-traceable variable chains; Based on the set of reverse-traceable variable chains, match the original data records from the time-series aligned steel structure state data stream, BIM model version library, and raw displacement monitoring point database according to the variable identifier and timestamp, and extract the sensor device number, BIM model version number, displacement monitoring point coordinates, and original values to generate a set of original input data identifiers; Merge the set of original input data identifiers and the set of reverse-traceable variable chains in chronological order and dependency relationship levels, construct a complete link of the original data identifier, intermediate calculation variable identifier, and target result identifier, and generate a causal traceability path for the abnormal mechanical properties.

Images