BIM-assisted steel structure construction safety management early warning method, device and equipment

Abstract

The invention relates to the technical field of data processing, and discloses a BIM-assisted steel structure construction safety management early warning method, device and equipment. The method comprises the following steps: carrying out multi-source heterogeneous data acquisition and space-time alignment processing on a steel structure construction site to obtain an engineering data stream; carrying out dynamic noise reduction and feature extraction on the engineering data flow to obtain a standardized multi-modal feature vector, and carrying out time sequence correlation analysis and mode mining to obtain a potential risk index set; performing multi-level threshold calculation and context sensitivity evaluation on the potential risk index set to obtain an adaptive early warning trigger matrix; performing multi-dimensional cross validation and causal chain analysis on the self-adaptive early warning trigger matrix to obtain a graded and classified risk assessment report; and carrying out scene analysis and response scheme generation on the graded and classified risk assessment report to obtain an executable multi-step emergency plan. According to the invention, the efficiency and accuracy of BIM-assisted steel structure construction safety management early warning are improved.

Description

Technical Field

[0001] The present application relates to the field of data processing, and in particular to a BIM-assisted steel structure construction safety management early warning method, device and equipment. Background Art

[0002] Steel structures are widely used in modern buildings and infrastructure due to their advantages such as high strength, light weight and fast construction. Traditional steel structure construction safety management mainly relies on manual inspections, regular testing and empirical judgment. With the development of information technology, some advanced monitoring methods such as strain sensors and inclinometers have been introduced into steel structure construction safety monitoring. At the same time, the application of building information modeling (BIM) technology has provided new possibilities for information management of the entire life cycle of steel structures. Some studies have attempted to combine BIM with Internet of Things technology to achieve real-time monitoring and visual management of the steel structure construction process.

[0003] However, the existing safety management methods for steel structure construction still have some shortcomings. First, most methods only focus on a single or a few risk factors, and it is difficult to fully consider the comprehensive impact of multi-source heterogeneous data. Secondly, the existing early warning systems often use fixed thresholds and lack adaptability to complex construction environments. Furthermore, risk assessments mostly remain at the qualitative analysis level and it is difficult to provide accurate risk quantification results. Finally, current emergency plans are usually pre-made static plans that cannot be dynamically adjusted according to real-time conditions. These problems limit the effectiveness and accuracy of safety management for steel structure construction. Summary of the invention

[0004] The present application provides a BIM-assisted steel structure construction safety management early warning method, device and equipment, which are used to improve the efficiency and accuracy of BIM-assisted steel structure construction safety management early warning.

[0005] In the first aspect, the present application provides a BIM-assisted steel structure construction safety management and early warning method, which includes: multi-source heterogeneous data collection and spatiotemporal alignment processing on the steel structure construction site to obtain a high-dimensional synchronized engineering data stream; dynamic noise reduction and feature extraction on the high-dimensional synchronized engineering data stream to obtain a standardized multimodal feature vector; time series correlation analysis and pattern mining on the standardized multimodal feature vector to obtain a set of potential risk indicators; multi-level threshold calculation and context-sensitive evaluation on the potential risk indicator set to obtain an adaptive early warning trigger matrix; multi-dimensional cross-validation and causal chain analysis on the adaptive early warning trigger matrix to obtain a hierarchical and classified risk assessment report; scenario-based analysis and response plan generation on the hierarchical and classified risk assessment report to obtain an executable multi-step emergency plan.

[0006] In a second aspect, the present application provides a BIM-assisted steel structure construction safety management and early warning device, the BIM-assisted steel structure construction safety management and early warning device comprising:

[0007] The processing module is used to collect multi-source heterogeneous data and perform spatiotemporal alignment processing on the steel structure construction site to obtain a high-dimensional synchronized engineering data stream;

[0008] An extraction module, used for dynamically reducing noise and extracting features from the high-dimensional synchronized engineering data stream to obtain a standardized multimodal feature vector;

[0009] A mining module, used for performing time series correlation analysis and pattern mining on the standardized multimodal feature vector to obtain a set of potential risk indicators;

[0010] An evaluation module, used for performing multi-level threshold calculation and context-sensitive evaluation on the potential risk indicator set to obtain an adaptive early warning trigger matrix;

[0011] An analysis module is used to perform multi-dimensional cross-validation and causal chain analysis on the adaptive early warning trigger matrix to obtain a hierarchical and classified risk assessment report;

[0012] A generation module is used to perform scenario-based analysis and response plan generation on the hierarchical and classified risk assessment report to obtain an executable multi-step emergency plan.

[0013] The third aspect of the present application provides a computer device, wherein the memory stores machine-readable instructions executable by the processor. When the computer device is running, the processor and the memory communicate via a bus, and when the machine-readable instructions are executed by the processor, the steps of the above-mentioned BIM-assisted steel structure construction safety management and early warning method are executed.

[0014] In the technical solution provided by the present application, a high-dimensional synchronized engineering data stream is obtained by collecting multi-source heterogeneous data and processing time and space alignment on the steel structure construction site, realizing all-round and multi-angle real-time monitoring of the construction site, and providing a comprehensive and accurate data basis for subsequent analysis. Secondly, dynamic noise reduction and feature extraction are performed on the high-dimensional synchronized engineering data stream to obtain a standardized multimodal feature vector, which effectively eliminates the noise interference in the data, extracts the most representative features, and improves the accuracy and efficiency of subsequent analysis. Thirdly, by performing time series correlation analysis and pattern mining on the standardized multimodal feature vector, a potential risk indicator set is obtained, and the time dependency and risk pattern implicit in the data are deeply mined, providing a scientific basis for risk warning. In addition, multi-level threshold calculation and context-sensitive evaluation are performed on the potential risk indicator set to obtain an adaptive early warning trigger matrix, which realizes the dynamic adjustment of the early warning mechanism and significantly improves the accuracy and adaptability of the early warning. At the same time, by performing multi-dimensional cross-validation and causal chain analysis on the adaptive early warning trigger matrix, a hierarchical and classified risk assessment report is obtained, which not only quantifies the risk level, but also reveals the causal relationship between risk factors, providing comprehensive support for risk management decisions. Finally, the hierarchical and classified risk assessment reports were analyzed and response plans were generated to obtain executable multi-step emergency plans, realizing intelligent management of the entire process from risk identification to emergency response. This method greatly improves the accuracy, real-time and operability of steel structure construction safety management, effectively reduces the probability of safety accidents, and improves the efficiency and effectiveness of emergency response. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0016] Figure 1 A schematic diagram of an embodiment of a BIM-assisted steel structure construction safety management and early warning method in an embodiment of the present application;

[0017] Figure 2 This is a schematic diagram of an embodiment of a BIM-assisted steel structure construction safety management and early warning device in an embodiment of the present application;

[0018] Figure 3 It is a schematic diagram of the structure of a computer device in an embodiment of the present application. DETAILED DESCRIPTION

[0019] The present application embodiment provides a BIM-assisted steel structure construction safety management early warning method, device and equipment. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of this application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments described here can be implemented in an order other than the content illustrated or described here. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0020] For ease of understanding, the specific process of the embodiment of the present application is described below. Figure 1 In the embodiment of the present application, an embodiment of the BIM-assisted steel structure construction safety management early warning method includes:

[0021] Step S101, performing multi-source heterogeneous data collection and spatiotemporal alignment processing on the steel structure construction site to obtain a high-dimensional synchronized engineering data stream;

[0022] Step S102, performing dynamic noise reduction and feature extraction on the high-dimensional synchronized engineering data stream to obtain a standardized multi-modal feature vector;

[0023] Step S103: performing time series correlation analysis and pattern mining on the standardized multimodal feature vector to obtain a set of potential risk indicators;

[0024] Step S104: performing multi-level threshold calculation and context-sensitive evaluation on the potential risk indicator set to obtain an adaptive warning trigger matrix;

[0025] Step S105: Perform multi-dimensional cross-validation and causal chain analysis on the adaptive early warning trigger matrix to obtain a hierarchical and classified risk assessment report;

[0026] Step S106: Perform scenario-based analysis and generate response plans for the graded and classified risk assessment reports to obtain an executable multi-step emergency plan.

[0027] It is understandable that the execution subject of the present application can be a BIM-assisted steel structure construction safety management early warning device, or a terminal or a server, which is not limited here. The present application embodiment is described by taking a server as the execution subject as an example.

[0028] Specifically, multi-source heterogeneous data collection and spatiotemporal alignment processing are performed on the steel structure construction site to obtain a high-dimensional synchronized engineering data stream. This involves the integration of multiple data sources, including point cloud data obtained by 3D laser scanning, real-time strain data collected by strain sensors, multi-dimensional environmental parameter data recorded by environmental monitoring equipment, and dynamic position information captured by the real-time positioning system. By performing spatiotemporal alignment processing on these data, the consistency of data from different sources in time and space dimensions is ensured, thereby forming a high-dimensional synchronized engineering data stream. Dynamic denoising and feature extraction are performed on the high-dimensional synchronized engineering data stream to obtain a standardized multimodal feature vector. In this step, adaptive wavelet transform is first applied to perform multi-scale decomposition of the data, and then threshold denoising is performed to eliminate the influence of noise. Subsequently, principal component analysis is used for dimensionality reduction, and high-order feature representation is obtained through nonlinear mapping. Key frequency features are extracted by time-frequency domain transformation and peak detection, and finally, standardized multimodal feature vectors are obtained after statistical moment extraction and normalization.

[0029] The standardized multimodal feature vectors are subjected to time series association analysis and pattern mining to obtain a potential risk indicator set. In this step, the feature vectors are first decomposed into time series to separate the trend, cycle and random components, and the time-dependent characteristics are obtained through autocorrelation analysis. The feature sequences are aligned using dynamic time warping technology, and then the sliding window segmentation is performed to obtain local feature fragments. Frequent pattern mining and association rule learning are used to identify repeated feature combinations and association rules between features. Finally, through graph structure modeling and community detection, highly correlated feature clusters are identified and their importance is scored to obtain a potential risk indicator set. Multi-level threshold calculation and context-sensitive evaluation are performed on the potential risk indicator set to obtain an adaptive warning trigger matrix. In this step, historical data statistical analysis is first performed on the indicator set, and multi-level quantiles are calculated to obtain the initial threshold set. The threshold range is adjusted in combination with expert knowledge, and multi-level fuzzy thresholds are obtained through fuzzy set partitioning. Dynamic time windows and context information association are applied to obtain context-sensitive thresholds. Threshold decision trees are formed through multi-dimensional cross-combination, and simplified decision rules are obtained through pruning optimization. Finally, the decision rules are mapped into a warning trigger condition matrix, and an adaptive adjustment mechanism is designed to obtain an adaptive warning trigger matrix.

[0030] The adaptive early warning trigger matrix is ​​subjected to multi-dimensional cross-validation and causal chain analysis to obtain a hierarchical risk assessment report. In this step, similar early warning scenarios are first found through historical case matching, and the result feedback analysis is performed to evaluate the accuracy of the early warning. The stability of the early warning is evaluated through multi-fold cross-validation, and sensitivity analysis is performed to identify key influencing factors. A causal inference model is constructed to obtain a potential causal relationship network, and the risk propagation chain is determined through path analysis. Multiple possible risk evolution paths are generated through scenario simulation, and probability evaluation is performed to obtain risk level classification. Finally, a multi-dimensional comprehensive score is performed on the risk level to form a risk severity matrix, and a structured description is performed to obtain a hierarchical risk assessment report. Finally, the hierarchical risk assessment report is scenario-based and response plan generation is performed to obtain an executable multi-step emergency plan. In this step, the risk assessment report is semantically parsed, key risk descriptors are extracted, and a risk scenario model is constructed. Available emergency resources are analyzed and resource allocation strategies are formulated. The response measures are decomposed into a sequence of preliminary action steps and time-series logic optimization is performed. Responsible roles are assigned to each action step, and an information transmission network is designed. Finally, key decision nodes are identified, alternative plans are generated, and an executable multi-step emergency plan is formed.

[0031] For example, in a steel structure construction project, a high-dimensional engineering data stream containing 100,000 data points was obtained through multi-source data collection. After dynamic noise reduction and feature extraction, the data dimension was reduced from the original 100 dimensions to 20 dimensions, forming a standardized multimodal feature vector. Time series association analysis and pattern mining identified 15 potential risk indicators, including key parameters such as structural stress, ambient temperature, and wind speed. Multi-level threshold calculation sets 3 levels of warning thresholds for each indicator, forming 45 warning trigger conditions. Cross-validation shows that the warning accuracy rate reaches 85%, and causal chain analysis draws out 3 main risk transmission paths. The final emergency plan contains 5 main response stages, involving 12 specific execution steps, and clarifies the division of responsibilities of 7 key roles.

[0032] In the embodiment of the present application, by collecting multi-source heterogeneous data and processing time and space alignment on the steel structure construction site, a high-dimensional synchronized engineering data stream is obtained, which realizes the real-time monitoring of the construction site from all directions and angles, and provides a comprehensive and accurate data basis for subsequent analysis. Secondly, dynamic noise reduction and feature extraction are performed on the high-dimensional synchronized engineering data stream to obtain a standardized multimodal feature vector, which effectively eliminates the noise interference in the data, extracts the most representative features, and improves the accuracy and efficiency of subsequent analysis. Thirdly, by performing time series correlation analysis and pattern mining on the standardized multimodal feature vector, a potential risk indicator set is obtained, and the time dependency and risk pattern implicit in the data are deeply mined, providing a scientific basis for risk warning. In addition, multi-level threshold calculation and context-sensitive evaluation are performed on the potential risk indicator set to obtain an adaptive early warning trigger matrix, which realizes the dynamic adjustment of the early warning mechanism and significantly improves the accuracy and adaptability of the early warning. At the same time, by performing multi-dimensional cross-validation and causal chain analysis on the adaptive early warning trigger matrix, a hierarchical and classified risk assessment report is obtained, which not only quantifies the risk level, but also reveals the causal relationship between risk factors, and provides comprehensive support for risk management decisions. Finally, the hierarchical and classified risk assessment reports were analyzed and response plans were generated to obtain executable multi-step emergency plans, realizing intelligent management of the entire process from risk identification to emergency response. This method greatly improves the accuracy, real-time and operability of steel structure construction safety management, effectively reduces the probability of safety accidents, and improves the efficiency and effectiveness of emergency response.

[0033] In a specific embodiment, the process of executing step S101 may specifically include the following steps:

[0034] (1) Perform three-dimensional laser scanning on steel structure components to obtain high-precision point cloud data, and perform noise filtering and sparse processing on the high-precision point cloud data to obtain structural geometric contour data;

[0035] (2) Feature extraction and parametric modeling are performed on the structural geometric contour data to obtain a steel structure BIM model, and attribute information is supplemented on the steel structure BIM model to obtain a semantic BIM model;

[0036] (3) Arrange strain sensors at key nodes of the steel structure to obtain real-time strain data, and perform signal denoising and outlier detection on the real-time strain data to obtain an effective strain data set;

[0037] (4) Deploy environmental monitoring equipment at the construction site to obtain multi-dimensional environmental parameter data, and perform data standardization and interpolation processing on the multi-dimensional environmental parameter data to obtain a continuous environmental state sequence;

[0038] (5) Configure a real-time positioning system for construction personnel and equipment to obtain dynamic location information, and perform trajectory smoothing and prediction analysis on the dynamic location information to obtain spatiotemporal behavior pattern data;

[0039] (6) The semantic BIM model, effective strain data set, continuous environmental state sequence and spatiotemporal behavior pattern data are timestamped to obtain a multi-source data synchronization matrix, and the multi-source data synchronization matrix is ​​spatially registered and scaled to obtain a high-dimensional synchronized engineering data stream.

[0040] Specifically, the steel structure components are scanned by 3D laser to obtain high-precision point cloud data. 3D laser scanning technology quickly captures the 3D coordinate information of the surface of an object by emitting lasers and receiving reflected signals to form dense point cloud data. The original point cloud data is subjected to noise filtering and sparse processing to remove abnormal points and redundant information, and obtain clearer structural geometric contour data. Noise filtering usually uses a statistical outlier removal algorithm, while sparse processing uses methods such as voxel grid method or octree segmentation to reduce the amount of data while retaining key geometric features. Feature extraction and parametric modeling are performed on the structural geometric contour data to construct a steel structure BIM model. The feature extraction process includes identifying basic geometric elements such as planes, edges, corners, and structural components such as beams, columns, and nodes. Parametric modeling is based on these extracted features and uses a predefined parametric component library to quickly generate an editable 3D model. On this basis, the steel structure BIM model is supplemented with attribute information, including multi-dimensional information such as material properties, construction progress, and safety specification requirements, so as to obtain a semantic BIM model. This semantic model contains not only geometric information, but also rich semantic information, providing comprehensive data support for subsequent safety management and risk analysis.

[0041] Strain sensors are arranged at key nodes of steel structures to collect strain data in real time. Strain sensors reflect the stress state of the structure by measuring material deformation and are an important means of monitoring structural safety. Signal denoising and outlier detection are performed on the collected real-time strain data to improve data quality. Signal denoising often uses methods such as wavelet transform or Kalman filtering, while outlier detection can use methods based on statistics or machine learning, such as the 3σ criterion or isolation forest algorithm. The effective strain data set obtained after processing is more reliable and can accurately reflect the actual stress condition of the structure. At the same time, environmental monitoring equipment is deployed at the construction site to obtain multi-dimensional environmental parameter data. These data include key environmental factors that affect construction safety, such as temperature, humidity, wind speed, noise, and dust concentration. These raw data are standardized to eliminate the dimensional differences between different parameters and facilitate comprehensive analysis. In addition, through interpolation processing, such as linear interpolation or spline interpolation, missing values ​​that may occur during data collection are filled to obtain a continuous environmental state sequence, providing a complete data basis for subsequent time series analysis and risk prediction.

[0042] In order to grasp the dynamic situation of the construction site, the real-time positioning system is configured for the construction personnel and equipment to obtain dynamic location information. This can be achieved through technologies such as RFID, GPS or UWB. The acquired raw location data is subjected to trajectory smoothing to eliminate positioning errors and jumps, and obtain a more coherent motion trajectory. Then, predictive analysis is performed, such as using methods such as Kalman filtering or long short-term memory network (LSTM) to predict future location changes. Through these processes, spatiotemporal behavior pattern data reflecting the movement rules of personnel and equipment are obtained, providing an important reference for safety management. Finally, the semantic BIM model, effective strain data set, continuous environmental state sequence and spatiotemporal behavior pattern data are timestamped to ensure the consistency of data from different sources in the time dimension and form a multi-source data synchronization matrix. Then, this matrix is ​​spatially registered and scaled to solve the problems of coordinate system differences and scale inconsistency that may exist between different data sources. Spatial registration usually uses feature point matching or iterative closest point (ICP) algorithm, while scale unification is achieved through coordinate transformation. After these processes, a high-dimensional synchronized engineering data stream is finally obtained, providing a comprehensive and consistent data foundation for subsequent safety risk analysis.

[0043] For example, in a steel structure project, 3D laser scanning was used to obtain raw point cloud data containing 100 million points. After noise filtering and sparse processing, the data volume was reduced to 20 million points, while retaining key structural geometry information. Based on these data, a steel structure BIM model containing 500 components was constructed, and 2,000 attribute information was supplemented to form a semantic BIM model. 50 strain sensors were arranged at key nodes, collecting data 100 times per second. After denoising and outlier detection, 95% of the valid data was retained. The environmental monitoring equipment recorded 10 environmental parameters per minute, and a continuous 24-hour environmental state sequence was generated through standardization and interpolation processing. The real-time positioning system tracked the location information of 100 workers and 20 equipment, and obtained 8 hours of spatiotemporal behavior pattern data after trajectory smoothing and predictive analysis. Finally, these multi-source heterogeneous data were integrated into a high-dimensional synchronized engineering data stream containing 10,000 time points and 1,000 feature dimensions, providing a comprehensive and unified data foundation for subsequent safety risk analysis. Through the collection, processing and integration of multi-source data, all-round and real-time monitoring of the steel structure construction site is achieved.

[0044] In a specific embodiment, the process of executing step S102 may specifically include the following steps:

[0045] (1) Adaptive wavelet transform is performed on the high-dimensional synchronized engineering data stream to obtain a multi-scale decomposition result, and threshold denoising is performed on the multi-scale decomposition result to obtain a smoothed data stream;

[0046] (2) Perform principal component analysis on the smoothed data stream to obtain a reduced-dimensionality feature set, and perform nonlinear mapping on the reduced-dimensionality feature set to obtain a high-order feature representation;

[0047] (3) Performing time-frequency domain transformation on the high-order feature representation to obtain a time-frequency feature matrix, and performing peak detection and spectrum analysis on the time-frequency feature matrix to obtain key frequency features;

[0048] (4) extracting statistical moments from key frequency features to obtain statistical feature vectors, and normalizing the statistical feature vectors to obtain standardized feature sets;

[0049] (5) Perform multimodal fusion on the standardized feature set to obtain a fused feature tensor, and perform tensor decomposition on the fused feature tensor to obtain a standardized multimodal feature vector.

[0050] Specifically, adaptive wavelet transform can automatically adjust the wavelet basis function according to the local characteristics of the data to achieve multi-scale decomposition of the signal. This process decomposes the original data into sub-signals at different frequency levels, effectively separating noise and useful information. The multi-scale decomposition results are subjected to threshold denoising. By setting an appropriate threshold, significant coefficients are retained while suppressing small-amplitude coefficients, thereby obtaining a smoothed data stream. This method effectively removes high-frequency noise while retaining the main features of the signal. Principal component analysis (PCA) is performed on the smoothed data stream to achieve data dimensionality reduction. PCA projects high-dimensional data into a low-dimensional space by finding the main direction of change of the data to obtain a reduced-dimensional feature set. This step not only reduces the redundancy of the data, but also extracts the most representative features in the data. In order to capture the nonlinear relationship in the data, the reduced-dimensional feature set is nonlinearly mapped to obtain a high-order feature representation. Commonly used methods for nonlinear mapping include deep learning techniques such as kernel principal component analysis (KPCA) or autoencoders, which can better express the intrinsic structure and complex patterns of the data.

[0051] Subsequently, the high-order feature representation is transformed into the time-frequency domain to obtain the time-frequency feature matrix. Time-frequency domain transformations, such as short-time Fourier transform (STFT) or wavelet transform, can simultaneously analyze the characteristics of the signal in the time and frequency domains, providing a more comprehensive signal representation. Peak detection and spectral analysis are performed on the time-frequency feature matrix to identify the significant frequency components and energy distribution in the signal, thereby obtaining the key frequency features. These features reflect the response characteristics of the steel structure at different frequencies and are of great significance for identifying structural anomalies and potential risks. Statistical moments are extracted from the key frequency features, and statistical quantities such as mean, variance, skewness, and kurtosis are calculated to obtain statistical feature vectors. These statistical features provide important information about the data distribution and help capture the overall characteristics and abnormal patterns of the data. In order to eliminate the dimensional differences between different features, the statistical feature vectors are normalized to obtain a standardized feature set. Normalization ensures that different features have the same weight in subsequent analysis and improves the effectiveness of feature fusion.

[0052] Finally, the standardized feature set is multimodally fused to integrate features from different data sources (such as BIM models, strain data, environmental data, etc.) to obtain a fused feature tensor. Multimodal fusion makes full use of the complementarity of different types of data and provides a more comprehensive information representation. The fused feature tensor is subjected to tensor decomposition, such as Tucker decomposition or CP decomposition, to further extract the core structure and correlation pattern of multimodal data, and finally obtain a standardized multimodal feature vector. This feature vector integrates the information of multi-source data and provides high-quality input for subsequent safety risk analysis.

[0053] For example, in a steel structure project, a raw data stream containing 1000 time points and 100 sensors was collected. Through adaptive wavelet transform, the data was decomposed into 4 scale levels, each scale level containing 250 sub-signals at time points. Threshold denoising reduced the noise amplitude by 80% to obtain a smoothed data stream. Principal component analysis reduced the 100-dimensional features to 20 dimensions, retaining 95% of the variance information. Nonlinear mapping mapped the 20-dimensional features to a 50-dimensional high-order feature space through a 3-layer autoencoder. The time-frequency domain transform used STFT with a window size of 64 to obtain a 64×250 time-frequency feature matrix. Peak detection identified 10 key frequency features. Statistical moment extraction calculated 4 statistics to obtain a 40-dimensional statistical feature vector. Normalization scaled all features to the range of [-1,1]. Multimodal fusion integrates the 10-dimensional geometric features of the BIM model, the 20-dimensional material features, and the 5-dimensional features of the environmental data to form a 75-dimensional fusion feature tensor. The tensor decomposition finally obtains a 50-dimensional standardized multimodal feature vector, which integrates multiple information such as structural geometry, material properties, dynamic response and environmental factors, providing a comprehensive and accurate data basis for the safety management of steel structure construction.

[0054] In a specific embodiment, the process of executing step S103 may specifically include the following steps:

[0055] (1) Perform time series decomposition on the standardized multimodal feature vector to obtain trend, cycle and random components, and perform autocorrelation analysis on the trend, cycle and random components to obtain time-dependent characteristics;

[0056] (2) Dynamically time warping the time-dependent features to obtain an aligned feature sequence, and then perform sliding window segmentation on the aligned feature sequence to obtain local feature fragments;

[0057] (3) Perform frequent pattern mining on local feature fragments to obtain repeated feature combinations, and perform association rule learning on the repeated feature combinations to obtain association rules between features;

[0058] (4) Model the association rules between features using graph structures to obtain a feature association network, and perform community detection on the feature association network to obtain highly correlated feature clusters;

[0059] (5) The importance of highly correlated feature clusters is scored to obtain key feature subsets, and the key feature subsets are semantically interpreted to obtain a set of potential risk indicators.

[0060] Specifically, time series decomposition of standardized multimodal feature vectors is a key step to deeply analyze data characteristics. Decomposing the original time series into trend, cycle and random components can reveal the intrinsic structure of the data. The trend component reflects the long-term trend of the data, the cycle component captures the periodic fluctuations of the data, and the random component represents irregular fluctuations and noise. Autocorrelation analysis is performed on these three components respectively, and the autocorrelation coefficients of different time lags are calculated to obtain time-dependent characteristics. Autocorrelation analysis can reveal the correlation of data at different time scales and provide an important basis for subsequent risk prediction. Dynamic Time Warping (DTW) is performed on the time-dependent characteristics. The DTW algorithm can effectively process time series of different rates or lengths by nonlinearly aligning time series, so that similar patterns can be aligned. This step obtains the aligned feature sequence, which lays the foundation for subsequent pattern recognition. Then, the aligned feature sequence is segmented by sliding windows to obtain local feature fragments. The sliding window technology divides a long time series into multiple short time fragments by moving a fixed-size window on the time series, which is convenient for capturing local patterns and features.

[0061] Frequent pattern mining is performed on local feature fragments to identify repeated feature combinations. Frequent pattern mining algorithms, such as Apriori or FP-Growth, can discover sets of features that often appear at the same time from a large amount of data. These repeated feature combinations may represent typical states or potential risk patterns in the steel structure construction process. Subsequently, association rule learning is performed on these repeated feature combinations to obtain association rules between features. Association rule learning can not only discover strong correlations between features, but also quantify the confidence and support of these associations, providing a quantitative basis for risk assessment. The association rules between features are modeled in a graph structure to construct a feature association network. In this network, nodes represent different features, and edges represent the strength of association between features. This graph structure representation intuitively shows the complex relationship between features, which is convenient for subsequent analysis. Community detection is performed on the constructed feature association network to identify highly correlated feature clusters. Community detection algorithms, such as the Louvain method or spectral clustering, can group closely connected nodes together to form feature clusters. These feature clusters represent groups of interrelated risk factors in the safety management of steel structure construction.

[0062] Finally, the importance of highly correlated feature clusters is scored to obtain a subset of key features. Importance scoring can be based on a variety of indicators, such as node degree centrality, betweenness centrality, or eigenvector centrality. These indicators measure the importance of features in the entire network from different perspectives. According to the scoring results, the most important feature subsets are selected, and these features are semantically interpreted. The data analysis results are matched with the actual engineering concepts, and finally a set of potential risk indicators is obtained. These risk indicators directly reflect the potential safety hazards in the steel structure construction process and provide a clear direction for management decisions.

[0063] For example, in a large steel structure project, standardized multimodal feature vectors containing 100 features and 10,000 time points were collected. Through time series decomposition, each feature sequence was decomposed into three components: trend, cycle, and random. Autocorrelation analysis of these components showed that 50% of the features had significant autocorrelation in the 24-hour cycle, reflecting the cyclical impact of daily construction activities. Dynamic time warping aligned different feature sequences, and the average alignment error was reduced from 2.5 hours to 0.5 hours. Using a 12-hour sliding window, about 2,000 local feature fragments were obtained. Frequent pattern mining identified 50 frequent feature combinations, including typical combinations such as "steel beam stress-temperature change-wind speed". Association rule learning generated 200 rules, such as "When the temperature rises by 5°C and the wind speed exceeds 10m / s, the probability of the steel beam stress increasing by 20% is 0.8". The feature association network contains 100 nodes and 500 edges, and the community detection algorithm identifies 10 highly correlated feature clusters. The importance score is based on the PageRank algorithm, and the top 20 most important features are selected as the key feature subset. Semantic interpretation maps these features to actual engineering concepts, such as "main beam deformation rate", "node stress concentration", "ambient temperature fluctuation", etc., and finally forms 15 specific potential risk indicators. These indicators cover multiple aspects such as structural stress, environmental factors, construction progress, etc., providing a comprehensive and accurate risk assessment basis for steel structure construction safety management.

[0064] In a specific embodiment, the process of executing step S104 may specifically include the following steps:

[0065] (1) Perform historical data statistical analysis on the potential risk indicator set to obtain the indicator distribution characteristics, and perform multi-level quantile calculation on the indicator distribution characteristics to obtain the initial threshold set;

[0066] (2) Perform expert knowledge fusion on the initial threshold set to obtain an adjusted threshold range, and perform fuzzy set partitioning on the adjusted threshold range to obtain a multi-level fuzzy threshold;

[0067] (3) Applying dynamic time windows to the multi-level fuzzy thresholds to obtain a time-varying threshold sequence, and associating the time-varying threshold sequence with context information to obtain a context-sensitive threshold;

[0068] (4) Perform multi-dimensional cross-combination on the context-sensitive thresholds to obtain a threshold decision tree, and perform pruning optimization on the threshold decision tree to obtain a simplified decision rule;

[0069] (5) Perform matrix mapping on the simplified decision rules to obtain the warning trigger condition matrix, and design an adaptive adjustment mechanism for the warning trigger condition matrix to obtain an adaptive warning trigger matrix.

[0070] Specifically, the probability distribution characteristics of each indicator are calculated, including statistics such as mean, standard deviation, skewness and kurtosis. Then, multi-level quantile calculations are performed on these indicator distribution characteristics to obtain the initial threshold set. The multi-level quantile calculation can be expressed by the following formula:

[0071] Q p =F -1 (p)

[0072] Among them, Q p is the p-quantile, F -1 is the inverse function of the cumulative distribution function, and p is the probability value (such as 0.95, 0.99, etc.). This method can set multiple initial thresholds for warning levels for each risk indicator. Next, the initial threshold set is fused with expert knowledge to obtain the adjusted threshold range. This step uses methods such as the Delphi method or the analytic hierarchy process (AHP) to adjust the statistically obtained thresholds in combination with expert experience. The adjusted threshold range is more in line with actual engineering experience and safety management needs. Subsequently, the adjusted threshold range is divided into fuzzy sets to obtain multi-level fuzzy thresholds. Fuzzy set theory can better deal with the uncertainty and transition of thresholds, and the fuzzy threshold can be expressed by a membership function:

[0073]

[0074] Among them, μ A (x) is the membership of element x to the fuzzy set A, and a and b are the lower and upper thresholds, respectively. Dynamic time windows are applied to the multi-level fuzzy thresholds to obtain a time-varying threshold sequence. The dynamic time window takes into account the impact of different time periods (such as seasons, day and night) on the threshold, so that the threshold is adjusted dynamically over time. This can be achieved through sliding window technology, and the window size is determined according to the specific application scenario. Then, the time-varying threshold sequence is associated with context information to obtain a context-sensitive threshold. Context information includes factors such as weather conditions, construction stage, and surrounding environment. These factors are taken into consideration through a weighted adjustment method to make the threshold more in line with the actual situation.

[0075] The context-sensitive thresholds are cross-combined in multiple dimensions to obtain a threshold decision tree. Each node of the decision tree represents the threshold judgment of a risk indicator, and the path represents the combined judgment condition of multiple indicators. In order to avoid the decision tree being too complicated, the threshold decision tree is pruned and optimized to obtain simplified decision rules. The pruning process can use indicators such as error rate or complexity penalty to improve decision efficiency while maintaining the accuracy of the warning. Finally, the simplified decision rules are matrix mapped to obtain the warning trigger condition matrix. This matrix clearly represents the warning level corresponding to different risk indicator combinations. In order to enable the early warning system to adapt to changes in the actual situation of the project, an adaptive adjustment mechanism is designed for the early warning trigger condition matrix to obtain an adaptive early warning trigger matrix. The adaptive adjustment mechanism can dynamically update the trigger conditions in the matrix based on the feedback of the early warning effect, thereby continuously optimizing the early warning performance.

[0076] For example, in a large steel structure project, 10 key risk indicators were selected, including main beam deflection, node stress, ambient temperature, etc. The distribution characteristics of each indicator were calculated through statistical analysis of three years of historical data. Taking the main beam deflection as an example, its mean is 20mm and the standard deviation is 2mm. Using multi-level quantile calculation, three levels of warning thresholds are obtained: mild warning (95% quantile, 24mm), moderate warning (99% quantile, 26mm) and severe warning (99.9% quantile, 28mm). In the expert knowledge fusion stage, five senior engineers adjusted the initial thresholds through the Delphi method. For example, considering the safety margin, the severe warning threshold of the main beam deflection was adjusted to 27mm. Fuzzy set partitioning converts this threshold into a fuzzy interval, such as [26mm, 28mm]. In this interval, the warning level gradually increases with the increase of deflection.

[0077] The dynamic time window application takes into account the impact of the temperature difference between day and night on structural deformation, and slightly relaxes the threshold at night (such as increasing by 1mm). Contextual information association further considers the impact of wind load, and lowers the threshold by 5% in windy weather (wind speed exceeds 10m / s). The decision tree formed by multi-dimensional cross-combination contains more than 100 nodes, which are reduced to 50 key nodes through pruning optimization to form a simplified decision rule. The final warning trigger condition matrix is ​​a 10×3 matrix (10 indicators, 3 warning levels), and the matrix elements are the specific thresholds of each indicator.

[0078] The adaptive adjustment mechanism has designed a feedback loop to evaluate the warning effect every week. If the false alarm rate is found to be too high (such as more than 10%), the threshold of the relevant indicator will be appropriately relaxed; if the false alarm rate is too high (such as more than 1%), the threshold will be tightened accordingly. In this way, the early warning system can continuously optimize itself, adapt to changes in the actual situation of the project, and improve the accuracy and reliability of the early warning.

[0079] In a specific embodiment, the process of executing step S105 may specifically include the following steps:

[0080] (1) Match historical cases on the adaptive warning trigger matrix to obtain a set of similar warning scenarios, and perform result feedback analysis on the similar warning scenario set to obtain an evaluation of the warning accuracy;

[0081] (2) Perform multi-fold cross-validation on the early warning accuracy assessment to obtain stability indicators, and perform sensitivity analysis on the stability indicators to obtain key influencing factors;

[0082] (3) Construct a causal inference model for key influencing factors to obtain a potential causal relationship network, and perform path analysis on the potential causal relationship network to obtain the risk transmission chain;

[0083] (4) Conduct scenario simulation on the risk transmission chain to obtain multiple possible risk evolution paths, and conduct probability assessment on multiple possible risk evolution paths to obtain risk level classification;

[0084] (5) Perform a multi-dimensional comprehensive scoring on the risk level classification to obtain a risk severity matrix, and then perform a structured description of the risk severity matrix to obtain a graded and classified risk assessment report.

[0085] Specifically, using similarity algorithms such as cosine similarity or Euclidean distance, the current warning situation is compared with the cases in the historical database, and the cases with similarity higher than the preset threshold are screened out to form a similar warning scene set. Subsequently, the result feedback analysis of these similar warning scene sets is carried out, the warning triggering situation is compared with the actual security incident, and the accuracy, recall rate and F1 score of the warning are calculated to obtain the warning accuracy evaluation. In order to ensure the reliability of the evaluation results, a multi-fold cross-validation is performed on the warning accuracy evaluation. Specifically, the data set is divided into K subsets (usually K = 5 or 10), and K-1 subsets are used as training sets each time, and the remaining subset is used as the test set. The cycle is repeated K times to obtain K evaluation results. The average value of these results is used as the final stability index, which reflects the performance stability of the warning system under different data distributions. Next, a sensitivity analysis of the stability index is performed. By changing the values ​​of different parameters (such as thresholds, time window sizes, etc.), the changes in warning performance are observed, so as to identify the key factors that have the greatest impact on the warning results.

[0086] The causal inference model is constructed for the identified key influencing factors. This step aims to reveal the intrinsic relationship between risk factors. Common causal inference methods include structural equation model (SEM) or Bayesian network, which can handle complex multivariate relationships and consider potential confounding factors. Through these models, a potential causal relationship network describing the mutual influence of various risk factors is obtained. Subsequently, path analysis is performed on this network to calculate direct and indirect effects, identify the transmission paths between risk factors, and thus obtain the risk transmission chain. The obtained risk transmission chain is simulated by scenario simulation. This step uses methods such as Monte Carlo simulation or agent-based modeling to simulate the evolution of risks under different initial conditions. Through multiple simulations, multiple possible risk evolution paths are obtained, which reflect the development trend and possible results of risks under different circumstances. These risk evolution paths are probabilistically evaluated, the probability of occurrence of each path is calculated, and the risks are divided into different levels, such as low risk, medium risk and high risk, according to the combination of risk level and probability of occurrence, so as to obtain risk level classification.

[0087] Finally, the risk level classification is scored in multiple dimensions. This process considers multiple factors, such as the probability of risk occurrence, potential loss level, and impact range, and uses weighted summation or fuzzy comprehensive evaluation methods to calculate the comprehensive score of each risk. These scores constitute a risk severity matrix, and each element of the matrix represents the severity of a specific risk type under specific conditions. This risk severity matrix is ​​structured and the numerical results are converted into easy-to-understand text descriptions, including risk type, level, possible impact, recommended measures, etc., to finally form a hierarchical and classified risk assessment report. For example, in a large steel structure building project, the early warning system sets an adaptive early warning trigger matrix based on 10 key indicators (such as main beam stress, node deformation, ambient temperature, etc.). By matching with 200 historical cases in the past five years, 50 cases with a similarity of more than 0.8 were selected as similar early warning scenario sets. The feedback analysis of the results shows that the system has an early warning accuracy of 85%, a recall rate of 90%, and an F1 score of 0.87. The stability of these results was evaluated using 5-fold cross validation, and the average F1 score was 0.86 and the standard deviation was 0.03, indicating that the system performance was relatively stable. Sensitivity analysis found that the ambient temperature threshold and the main beam stress change rate were the two most critical factors affecting the accuracy of the warning.

[0088] Based on these key factors, a Bayesian network with 15 nodes was constructed as a causal inference model. Path analysis revealed a major risk propagation chain: increased ambient temperature → thermal expansion of steel → increased node stress → increased deformation of main beams → decreased overall structural stability.

[0089] Through 1000 Monte Carlo simulations, 100 different risk evolution paths were generated. The probability assessment showed that 20% of the paths belonged to the high risk category, 30% to medium risk, and 50% to low risk. The multi-dimensional comprehensive score took into account three aspects: risk probability (weight 0.3), potential economic loss (weight 0.4), and safety impact (weight 0.3). The scoring results formed a 10×3 risk severity matrix (10 risk types, 3 severity levels). The final risk assessment report describes in detail the characteristics, possible consequences, and recommended countermeasures of each risk, providing comprehensive risk management guidance for the project management team.

[0090] In a specific embodiment, the process of executing step S106 may specifically include the following steps:

[0091] (1) Perform semantic analysis on the hierarchical and classified risk assessment reports to obtain key risk descriptors, and extract scenario elements from the key risk descriptors to obtain a risk scenario model;

[0092] (2) Conduct resource constraint analysis on the risk scenario model to obtain a list of available emergency resources, prioritize the list of available emergency resources, and obtain a resource allocation strategy;

[0093] (3) Decompose the resource allocation strategy into tasks to obtain a preliminary action step sequence, and optimize the preliminary action step sequence through temporal logic to obtain an optimized action process;

[0094] (4) Assign roles to the optimized action process to obtain a responsibility matrix, and design communication paths for the responsibility matrix to obtain an information transmission network;

[0095] (5) Decision points are identified in the information transmission network to obtain a set of key decision nodes, and alternative plans are generated for the set of key decision nodes to obtain an executable multi-step emergency plan.

[0096] Specifically, natural language processing techniques, such as named entity recognition (NER) and keyword extraction, are used to identify key risk descriptors from risk assessment reports. These descriptors may include information such as risk type, severity, and scope of impact. Subsequently, scenario elements are extracted from these key risk descriptors, and the abstract risk descriptions are converted into specific scenario elements, such as affected structural components, dangerous areas, potential chain reactions, etc., using knowledge graphs or ontology models, so as to construct a structured risk scenario model. Resource constraint analysis of the risk scenario model is the next key step. This process first generates a list of currently available emergency resources, including manpower, equipment, materials, etc., based on the BIM model and resource management system. Then, these resources are prioritized using multi-criteria decision analysis (MCDA) methods, such as the analytic hierarchy process (AHP) or TOPSIS. The basis for the ranking includes factors such as resource effectiveness, availability, and cost. Priority ranking can be expressed by the following formula:

[0097]

[0098] Among them, P i is the priority score of resource i, w j is the weight of the jth criterion, s ij is the score of resource i on criterion j, and n is the number of evaluation criteria. In this way, an optimized resource allocation strategy is obtained. The resource allocation strategy is task-decomposed. This step uses the work breakdown structure (WBS) technique to break down the overall emergency response into a series of specific action steps. These preliminary action step sequences are optimized in time logic, using the critical path method (CPM) or program review and evaluation technique (PERT) to determine the dependencies between steps and the optimal execution order. The optimized action process can be represented by a network diagram, where nodes represent action steps and edges represent dependencies between steps. Time estimation can use the following PERT formula:

[0099]

[0100] Among them, T e is the expected time, T o is the most optimistic time estimate, T m is the most likely time estimate, T pIt is the most pessimistic time estimate. The role assignment for the optimized action process is performed. This step uses the responsibility assignment matrix (RACI matrix) technology. RACI stands for the four roles of Responsible, Accountable, Consulted, and Informed. By filling in the RACI matrix, the executors and stakeholders of each action step are clarified to obtain a detailed responsibility matrix. Based on this responsibility matrix, the communication path is designed, and graph theory algorithms such as the Dijkstra shortest path algorithm are used to optimize the flow of information, and finally form an efficient information transmission network.

[0101] Finally, the decision points of the information transmission network are identified, and decision tree analysis or influence diagram technology is used to find out the nodes where key judgments need to be made. These nodes constitute the key decision node set. For each key decision node, scenario planning technology is used to generate multiple alternatives. The generation of these alternatives takes into account different risk development situations and resource availability, and finally forms a comprehensive and executable multi-step emergency plan.

[0102] For example, in a large steel structure building project, the risk assessment report identified the key risk descriptor "excessive deformation of the main beam" through semantic analysis. Scenario element extraction converts it into a specific risk scenario model, including deformation location, degree, potential impact range, etc. Resource constraint analysis shows that the currently available emergency resources include 5 hydraulic jacks, 3 sets of temporary support frames, and 2 emergency maintenance teams. The AHP method is used to prioritize these resources, and the resource allocation strategy obtained is to first deploy temporary support frames, then use hydraulic jacks for correction, and finally reinforce them by the maintenance team. Task decomposition divides the entire emergency response into 10 specific steps, including site investigation, safety area demarcation, temporary support frame installation, etc. The PERT method is used to optimize the execution order of these steps, and the expected completion time of the entire emergency response is calculated to be 6 hours, of which the critical path includes 5 key steps. Responsibility allocation uses the RACI matrix to clarify the person in charge of each step, such as the project manager is responsible for overall coordination, the safety supervisor is responsible for on-site safety control, and the engineer is responsible for technical guidance.

[0103] The design of the information transmission network ensures that key information can be transmitted to relevant personnel within an average of 1 minute. The decision point analysis identified three key decision nodes, including whether evacuation is needed, what reinforcement plan to adopt, and whether work needs to be stopped. For these decision points, a total of 12 alternative plans were generated to form the final multi-step emergency plan.

[0104] The above describes the BIM-assisted steel structure construction safety management early warning method in the embodiment of the present application. The following describes the BIM-assisted steel structure construction safety management early warning device in the embodiment of the present application. Figure 2 In the embodiment of the present application, an embodiment of the BIM-assisted steel structure construction safety management and early warning device includes:

[0105] Processing module 201 is used to collect multi-source heterogeneous data and perform spatiotemporal alignment processing on the steel structure construction site to obtain a high-dimensional synchronized engineering data stream;

[0106] An extraction module 202, configured to perform dynamic noise reduction and feature extraction on the high-dimensional synchronized engineering data stream to obtain a standardized multi-modal feature vector;

[0107] A mining module 203 is used to perform time series correlation analysis and pattern mining on the standardized multimodal feature vector to obtain a potential risk indicator set;

[0108] An evaluation module 204 is used to perform multi-level threshold calculation and context-sensitive evaluation on the potential risk indicator set to obtain an adaptive warning trigger matrix;

[0109] An analysis module 205 is used to perform multi-dimensional cross-validation and causal chain analysis on the adaptive early warning trigger matrix to obtain a hierarchical and classified risk assessment report;

[0110] The generation module 206 is used to perform scenario analysis and response plan generation on the hierarchical and classified risk assessment report to obtain an executable multi-step emergency plan.

[0111] Through the collaborative cooperation of the above components, multi-source heterogeneous data collection and spatiotemporal alignment processing are carried out on the steel structure construction site to obtain a high-dimensional synchronized engineering data stream, which realizes all-round and multi-angle real-time monitoring of the construction site, providing a comprehensive and accurate data basis for subsequent analysis. Secondly, dynamic noise reduction and feature extraction are carried out on the high-dimensional synchronized engineering data stream to obtain a standardized multimodal feature vector, which effectively eliminates the noise interference in the data, extracts the most representative features, and improves the accuracy and efficiency of subsequent analysis. Thirdly, by performing time series correlation analysis and pattern mining on the standardized multimodal feature vector, a potential risk indicator set is obtained, and the implicit time dependency and risk pattern in the data are deeply mined, providing a scientific basis for risk warning. In addition, multi-level threshold calculation and context-sensitive evaluation are performed on the potential risk indicator set to obtain an adaptive early warning trigger matrix, which realizes the dynamic adjustment of the early warning mechanism and significantly improves the accuracy and adaptability of the early warning. At the same time, by performing multi-dimensional cross-validation and causal chain analysis on the adaptive early warning trigger matrix, a hierarchical and classified risk assessment report is obtained, which not only quantifies the risk level, but also reveals the causal relationship between risk factors, providing comprehensive support for risk management decisions. Finally, the hierarchical and classified risk assessment reports were analyzed and response plans were generated to obtain executable multi-step emergency plans, realizing intelligent management of the entire process from risk identification to emergency response. This method greatly improves the accuracy, real-time and operability of steel structure construction safety management, effectively reduces the probability of safety accidents, and improves the efficiency and effectiveness of emergency response.

[0112] Based on the same technical concept, the embodiment of the present application also provides an electronic device. Figure 3 As shown, it is a schematic diagram of the structure of the electronic device 300 provided in the embodiment of the present application, including a processor 301, a memory 302, and a bus 303. Among them, the memory 302 is used to store execution instructions, including a memory 3021 and an external memory 3022; the memory 3021 here is also called an internal memory, which is used to temporarily store the operation data in the processor 301, and the data exchanged with the external memory 3022 such as a hard disk. The processor 301 exchanges data with the external memory 3022 through the memory 3021. When the electronic device 300 is running, the processor 301 communicates with the memory 302 through the bus 303.

[0113] The present application also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. Instructions are stored in the computer-readable storage medium. When the instructions are executed on a computer, the computer executes the steps of the BIM-assisted steel structure construction safety management and early warning method.

[0114] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0115] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program codes.

[0116] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A BIM-assisted steel structure construction safety management early warning method, characterized in that: The BIM-assisted steel structure construction safety management early warning method includes: Perform multi-source heterogeneous data collection and spatiotemporal alignment processing on steel structure construction sites to obtain high-dimensional synchronized engineering data streams; Performing dynamic noise reduction and feature extraction on the high-dimensional synchronized engineering data stream to obtain a standardized multimodal feature vector; Performing time series correlation analysis and pattern mining on the standardized multimodal feature vector to obtain a set of potential risk indicators; Performing multi-level threshold calculation and context-sensitive evaluation on the potential risk indicator set to obtain an adaptive early warning trigger matrix; Perform multi-dimensional cross-validation and causal chain analysis on the adaptive early warning trigger matrix to obtain a graded and classified risk assessment report; The hierarchical and classified risk assessment reports are analyzed in a scenario-based manner and response plans are generated to obtain an executable multi-step emergency plan.

2. The BIM-assisted steel structure construction safety management early warning method according to claim 1 is characterized in that: The multi-source heterogeneous data collection and spatiotemporal alignment processing are performed on the steel structure construction site to obtain a high-dimensional synchronized engineering data stream, including: Performing three-dimensional laser scanning on the steel structure components to obtain high-precision point cloud data, and performing noise filtering and sparse processing on the high-precision point cloud data to obtain structural geometric contour data; Performing feature extraction and parametric modeling on the structural geometric profile data to obtain a steel structure BIM model, and supplementing the steel structure BIM model with attribute information to obtain a semantic BIM model; Arrange strain sensors at key nodes of the steel structure to obtain real-time strain data, and perform signal denoising and outlier detection on the real-time strain data to obtain a valid strain data set; Deploy environmental monitoring equipment at the construction site to obtain multi-dimensional environmental parameter data, and perform data standardization and interpolation processing on the multi-dimensional environmental parameter data to obtain a continuous environmental state sequence; Perform real-time positioning system configuration on construction personnel and equipment to obtain dynamic location information, and perform trajectory smoothing and prediction analysis on the dynamic location information to obtain spatiotemporal behavior pattern data; The semantic BIM model, effective strain data set, continuous environmental state sequence and spatiotemporal behavior pattern data are timestamp aligned to obtain a multi-source data synchronization matrix, and the multi-source data synchronization matrix is ​​spatially aligned and scaled to obtain a high-dimensional synchronized engineering data stream.

3. The BIM-assisted steel structure construction safety management early warning method according to claim 1 is characterized in that: The step of dynamically reducing noise and extracting features from the high-dimensional synchronized engineering data stream to obtain a standardized multimodal feature vector includes: Performing adaptive wavelet transform on the high-dimensional synchronized engineering data stream to obtain a multi-scale decomposition result, and performing threshold denoising on the multi-scale decomposition result to obtain a smoothed data stream; Performing principal component analysis on the smoothed data stream to obtain a reduced-dimensionality feature set, and performing nonlinear mapping on the reduced-dimensionality feature set to obtain a high-order feature representation; Performing a time-frequency domain transformation on the high-order feature representation to obtain a time-frequency feature matrix, and performing peak detection and spectrum analysis on the time-frequency feature matrix to obtain key frequency features; Performing statistical moment extraction on the key frequency features to obtain statistical feature vectors, and normalizing the statistical feature vectors to obtain a standardized feature set; The standardized feature set is subjected to multimodal fusion to obtain a fused feature tensor, and the fused feature tensor is subjected to tensor decomposition to obtain a standardized multimodal feature vector.

4. The BIM-assisted steel structure construction safety management early warning method according to claim 1 is characterized in that: The time series correlation analysis and pattern mining are performed on the standardized multimodal feature vector to obtain a potential risk indicator set, including: Performing time series decomposition on the standardized multimodal feature vector to obtain trend, cycle and random components, and performing autocorrelation analysis on the trend, cycle and random components to obtain time-dependent characteristics; Performing dynamic time warping on the time-dependent characteristics to obtain an aligned feature sequence, and performing sliding window segmentation on the aligned feature sequence to obtain local feature fragments; Performing frequent pattern mining on the local feature fragments to obtain repeated feature combinations, and performing association rule learning on the repeated feature combinations to obtain association rules between features; Performing graph structure modeling on the association rules between the features to obtain a feature association network, and performing community detection on the feature association network to obtain highly correlated feature clusters; Importance scores are performed on the highly correlated feature clusters to obtain key feature subsets, and semantic interpretation is performed on the key feature subsets to obtain a potential risk indicator set.

5. The BIM-assisted steel structure construction safety management early warning method according to claim 1 is characterized in that: The step of performing multi-level threshold calculation and context-sensitive evaluation on the potential risk indicator set to obtain an adaptive early warning trigger matrix includes: Performing historical data statistical analysis on the potential risk indicator set to obtain indicator distribution characteristics, and performing multi-level quantile calculation on the indicator distribution characteristics to obtain an initial threshold set; Performing expert knowledge fusion on the initial threshold set to obtain an adjusted threshold range, and performing fuzzy set division on the adjusted threshold range to obtain a multi-level fuzzy threshold; Applying a dynamic time window to the multi-level fuzzy threshold to obtain a time-varying threshold sequence, and associating the time-varying threshold sequence with context information to obtain a context-sensitive threshold; Performing multi-dimensional cross-combination on the context-sensitive threshold to obtain a threshold decision tree, and performing pruning optimization on the threshold decision tree to obtain a simplified decision rule; The simplified decision rules are subjected to matrix mapping to obtain a warning trigger condition matrix, and an adaptive adjustment mechanism is designed for the warning trigger condition matrix to obtain an adaptive warning trigger matrix.

6. The BIM-assisted steel structure construction safety management early warning method according to claim 1 is characterized in that: The adaptive early warning trigger matrix is ​​subjected to multi-dimensional cross-validation and causal chain analysis to obtain a hierarchical and classified risk assessment report, including: Performing historical case matching on the adaptive warning trigger matrix to obtain a set of similar warning scenarios, and performing result feedback analysis on the set of similar warning scenarios to obtain a warning accuracy assessment; Performing multi-fold cross validation on the early warning accuracy assessment to obtain a stability index, and performing sensitivity analysis on the stability index to obtain key influencing factors; Constructing a causal inference model for the key influencing factors to obtain a potential causal relationship network, and performing path analysis on the potential causal relationship network to obtain a risk transmission chain; Conduct scenario simulation on the risk propagation chain to obtain multiple possible risk evolution paths, and conduct probability assessment on the multiple possible risk evolution paths to obtain risk level classification; The risk level classification is scored comprehensively in multiple dimensions to obtain a risk severity matrix, and the risk severity matrix is ​​structured to obtain a graded and classified risk assessment report.

7. The BIM-assisted steel structure construction safety management and early warning method according to claim 1 is characterized in that: The scenario analysis and response plan generation of the hierarchical and classified risk assessment report to obtain an executable multi-step emergency plan includes: Performing semantic analysis on the hierarchically classified risk assessment report to obtain key risk descriptors, and extracting scenario elements from the key risk descriptors to obtain a risk scenario model; Performing resource constraint analysis on the risk scenario model to obtain a list of available emergency resources, and prioritizing the list of available emergency resources to obtain a resource allocation strategy; Decomposing the resource allocation strategy into tasks to obtain a preliminary action step sequence, and optimizing the preliminary action step sequence in time sequence logic to obtain an optimized action process; Performing role assignment on the optimized action process to obtain a responsibility matrix, and performing communication path design on the responsibility matrix to obtain an information transmission network; Decision points are identified on the information transmission network to obtain a set of key decision nodes, and alternative plans are generated for the set of key decision nodes to obtain an executable multi-step emergency plan.

8. A BIM-assisted steel structure construction safety management and early warning device, used to implement the BIM-assisted steel structure construction safety management and early warning method as described in any one of claims 1 to 7, characterized in that: The BIM-assisted steel structure construction safety management early warning device includes: The processing module is used to collect multi-source heterogeneous data and perform spatiotemporal alignment processing on the steel structure construction site to obtain a high-dimensional synchronized engineering data stream; An extraction module, used for dynamically reducing noise and extracting features from the high-dimensional synchronized engineering data stream to obtain a standardized multimodal feature vector; A mining module, used for performing time series correlation analysis and pattern mining on the standardized multimodal feature vector to obtain a set of potential risk indicators; An evaluation module, used for performing multi-level threshold calculation and context-sensitive evaluation on the potential risk indicator set to obtain an adaptive early warning trigger matrix; An analysis module is used to perform multi-dimensional cross-validation and causal chain analysis on the adaptive early warning trigger matrix to obtain a hierarchical and classified risk assessment report; A generation module is used to perform scenario-based analysis and response plan generation on the hierarchical and classified risk assessment report to obtain an executable multi-step emergency plan.

9. A computer device, characterized in that: include: A processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the computer device is running, the processor and the memory communicate via the bus, and when the machine-readable instructions are executed by the processor, the steps of the BIM-assisted steel structure construction safety management and early warning method as described in any one of claims 1 to 7 are performed.

Images