Construction dynamic twin method based on internet of things and BIM

Abstract

The application provides a construction dynamic twin method based on Internet of Things and BIM, relates to the technical field of digital twin, and comprises the following steps: acquiring heterogeneous data streams collected by sensors and performing analysis and classification on the data streams to generate a standardized data set; performing semantic mapping on the standardized data set according to the association rules of BIM component identification and data attributes to generate a component associated data set; decrypting and decompressing encrypted data packets in the cloud, analyzing construction progress time sequence data by using a distributed computing framework, and generating a deviation detection result; dynamically adjusting the routing rules of the data link according to the deviation detection result, reassigning the processing priority to optimize the link configuration; updating the BIM component state based on the optimized link configuration, generating a real-time updated digital twin model by using cloud computing resources; and extracting three-dimensional visualization data from the digital twin model and mapping the three-dimensional visualization data to the BIM model interface by using a rendering technology, so that the monitoring precision and management efficiency of the construction process are effectively improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of digital twinning, in particular to a construction dynamic twinning method based on Internet of Things and BIM. BACKGROUND

[0002] The digital transformation of the construction field is a key direction to improve project management efficiency and quality. The construction dynamic twinning technology based on Internet of Things and Building Information Modeling (BIM) is a core means to realize the whole process of digitalization, which has important significance. This technology constructs a dynamic mapping between the physical world and the digital world by real-time collection of construction site data and fusion with BIM model, providing accurate decision support for construction management.

[0003] However, the existing methods have significant limitations in practical application and are difficult to meet the needs of complex construction scenarios. Currently, most solutions rely on a single data source or simple integration, which is difficult to cope with the complexity of multi-source heterogeneous data in the construction site. Traditional methods have deficiencies in data standardization, real-time synchronization and dynamic adjustment, which leads to the disconnection between the digital twinning model and the actual construction state. In addition, existing systems are limited in computing and storage capacity when dealing with massive data, making it difficult to achieve efficient data analysis and visualization. These defects limit the deep application of dynamic twinning technology in construction management.

[0004] The core challenges of the integration of Internet of Things and BIM are concentrated in the following technical factors: First, there are many types of sensor devices in the construction site, and the data formats are different. How to realize the standardized processing of heterogeneous data and accurate mapping with BIM model components is a basic problem in building a dynamic twinning model. Second, the network environment in the construction site is complex, and the real-time, reliability and security of data transmission are difficult to guarantee, which directly affects the dynamic update of the twinning model. Third, the dynamic changes of construction progress and technology require the data link to have adaptive adjustment capability, and the existing technology is difficult to achieve flexible optimization.

[0005] Therefore, how to establish a unified data exchange protocol to realize the standardization and mapping of heterogeneous data, how to optimize the data transmission mechanism to ensure real-time and reliability, and how to design an adaptive data link to cope with dynamic changes in construction, have become key problems that need to be solved in the construction dynamic twinning method based on Internet of Things and BIM. SUMMARY

[0006] The present application provides a construction dynamic twinning method based on Internet of Things and BIM, mainly comprising:

[0007] Obtain heterogeneous data streams collected by sensors, parse and classify the heterogeneous data streams using a preset format template, and generate a standardized data set;

[0008] According to the association rule of the BIM component identification and data attribute, semantic mapping is performed on the standardized data set to generate a component association data set;

[0009] If the matching accuracy of the component association data set is lower than a preset threshold, a historical deviation data is used to train a machine learning model, optimize the association rule parameters and generate an optimized component association data set;

[0010] Real-time data streams are extracted from the optimized component association data set, and an edge computing node is used to compress and encrypt the real-time data streams to generate encrypted data packets;

[0011] The encrypted data packets are sent to a cloud server through a main transmission channel, and if network jitter is detected to exceed a preset threshold, a backup transmission channel is switched to;

[0012] The encrypted data packets are decrypted and decompressed in the cloud, and a distributed computing framework is used to analyze construction progress time series data to generate deviation detection results;

[0013] According to the deviation detection results, the routing rules of the data link are dynamically adjusted, and the processing priority is re-allocated to optimize the link configuration;

[0014] Based on the optimized link configuration, the BIM component state is updated, and cloud computing resources are used to generate a real-time updated digital twin model;

[0015] Three-dimensional visualization data is extracted from the digital twin model, and rendering technology is used to map sensor data to the BIM model interface.

[0016] The technical scheme provided by the embodiment of the application can include the following beneficial effects:

[0017] The application discloses an intelligent construction management method based on an Internet of Things and BIM, which realizes semantic association mapping with BIM components by collecting heterogeneous sensor data streams and performing standardized processing, uses an edge computing node to compress and encrypt real-time data for transmission, uses a distributed computing framework in the cloud to analyze construction progress time series data, dynamically adjusts data link configuration, updates BIM component state and generates a digital twin model, and finally visualizes sensor data on the BIM interface. The method combines technologies such as the Internet of Things, BIM, edge computing, cloud computing, etc., realizes real-time collection, intelligent analysis and dynamic management of construction site data, effectively improves the monitoring accuracy and management efficiency of the construction process, and provides technical support for smart construction site construction. BRIEF DESCRIPTION OF DRAWINGS

[0018] In order to better understand and implement, the technical scheme of the application is described in detail below with reference to the drawings.

[0019] Fig. 1A flow chart of a construction dynamic twin method based on the Internet of Things and BIM according to the present application;

[0020] Fig. 2 A schematic diagram of a construction dynamic twin method based on the Internet of Things and BIM according to the present application;

[0021] Fig. 3 Another schematic diagram of a construction dynamic twin method based on the Internet of Things and BIM according to the present application. DETAILED DESCRIPTION

[0022] To further clarify the technical means and effects taken by the present application to achieve the predetermined application purposes, the exemplary embodiments will be described in detail here, and the examples are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments do not represent all the embodiments consistent with the present application. Instead, they are merely examples of methods and systems consistent with some aspects of the present application as detailed in the appended claims.

[0023] The terms used in the present application are merely for the purpose of describing specific embodiments and are not intended to limit the present application. The singular forms "a", "an" and "the" used in the present application and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used herein means any or all possible combinations of one or more associated listed items.

[0024] The specific embodiments, features and effects according to the present application are described in detail below in conjunction with the accompanying drawings and preferred embodiments.

[0025] As Figs. 1-3 The construction dynamic twin method based on the Internet of Things and BIM according to the present embodiment can specifically include:

[0026] In step S101, the heterogeneous data stream collected by the sensor is obtained, the heterogeneous data stream is parsed and classified using a preset format template, and a standardized data set is generated.

[0027] The heterogeneous data stream is acquired from the sensor, the data stream integrity is verified by using a stream processing technology, and an initial data stream is obtained. The initial data stream is matched by using a preset template, if the template is consistent with the characteristics of the data stream, the key field is extracted, and structured data is obtained. The structured data is subjected to feature decomposition by using a decision tree algorithm, if the feature meets the classification condition, a category label is assigned, and classified data is obtained. According to the classified data, the data is mapped to a unified format by using a data standardization technology, and a standardized data set is obtained. Time sequence features are extracted from the standardized data set, if the feature sequence meets the continuity condition, a dynamic data stream is generated, and a dynamic data set is obtained. The dynamic data set is grouped by using a K-means clustering algorithm, if the grouping result meets the similarity threshold, a grouped data set is generated, and a final data set is obtained. The final data set is encoded by using a data formatting technology, and a formatted data set is obtained.

[0028] In step S102, semantic mapping is performed on the standardized data set according to the association rule of the BIM component identification and data attribute, and a component association data set is generated.

[0029] The association rule of the component identification and data attribute is obtained from the standardized data set, the component identification is matched by using a semantic mapping technology, if the rule is consistent with the attribute characteristics, the associated field is extracted, and a component association data set is obtained. According to the component association data set, the component identification is grouped by using a K-means clustering algorithm, if the grouping result meets a preset similarity threshold, the component category is determined, and a classified data set is obtained. The associated field of the component identification is extracted from the classified data set, the field is structured by using a preset template, if the field conforms to the template format, a structured data set is generated. According to the structured data set, the consistency of the component identification and data attribute is verified by using a rule verification technology, if the consistency meets a preset condition, a verification data set is obtained. The semantic features associated with the component are extracted from the verification data set, the semantic features are analyzed by using a feature decomposition technology, if the features meet the decomposition condition, a semantic data set is obtained. According to the semantic data set, the data set is mapped to a unified format by using a format conversion technology, if the format conversion result meets the standard specification, a formatted data set is obtained. The time sequence features associated with the component are extracted from the formatted data set, if the feature sequence meets the continuity condition, a dynamic association data set is generated.

[0030] In step S103, if the matching accuracy of the component association data set is lower than a preset threshold, a machine learning model is trained by using historical deviation data, the association rule parameters are optimized, and an optimized component association data set is generated.

[0031] The component association dataset is obtained from the repository, and it is determined whether the matching accuracy of the component association dataset is lower than a preset threshold. If the matching accuracy is lower than the preset threshold, historical deviation data is obtained from the repository to obtain a deviation dataset. A random forest model is trained according to the deviation dataset, the hyperparameters of the model are set to default values, the model weight is adjusted, and an optimized training model is obtained. The optimized training model is used to analyze the component association rule, the rule parameters are extracted, and a parameter set is obtained. It is determined whether the deviation value of the parameter set is outside a preset range. If the deviation value is outside the preset range, the parameter set is adjusted by a gradient descent algorithm, the learning rate is set to a preset value, the number of iterations is set to a preset number, and an adjusted parameter set is obtained. The association rule is updated according to the adjusted parameter set, and a new association rule set is generated. The matching accuracy is recalculated using the new association rule set, and an updated matching accuracy is obtained. It is determined whether the updated matching accuracy is lower than the preset threshold. If the updated matching accuracy is lower than the preset threshold, secondary deviation features are extracted from the deviation data, and the random forest model is repeatedly trained to obtain a further optimized training model.

[0032] In step S104, real-time data streams are extracted from the optimized component association dataset, and the data streams are compressed and encrypted using an edge computing node to generate encrypted data packets.

[0033] From the optimized component association dataset, a structured real-time data stream is obtained, and the structured real-time data stream generates a continuous data stream according to a pre-established field filtering rule. An Apache Flink framework on an edge computing node receives the continuous data stream, sets an LZ4 compression level for stream compression, and obtains a compressed data stream. If the volume of the compressed data stream exceeds a preset threshold, the compressed data stream is cut into pieces of a fixed size to obtain an ordered data piece set. Each data piece in the ordered data piece set is encrypted using an AES-256 algorithm in cooperation with a pre-distributed key to obtain an encrypted data piece set. The encrypted data piece set is packaged by the edge computing node in sequence, a metadata header containing data length and piece number is added, and a complete encrypted data packet is obtained. The SHA-256 algorithm is used to calculate the hash value of the metadata header of the complete encrypted data packet. If the hash value matches a pre-stored check code, the complete encrypted data packet is sent to an authenticated target node address through the MQTT protocol.

[0034] It should be noted that the Apache Flink framework is an open source stream processing framework designed for distributed data stream and batch data processing. It supports high throughput and low latency real-time stream processing, and can also process batch data (i.e., “batch processing”). Flink provides powerful stream computing capabilities and is suitable for large-scale data analysis, machine learning, real-time monitoring, etc.

[0035] LZ4 is a fast compression algorithm, especially suitable for streaming data compression, which allows users to set the compression level, and can trade off between compression speed and compression ratio.

[0036] AES-256 algorithm and SHA-256 algorithm are commonly used encryption and hash algorithms, among which AES-256 algorithm: used for encrypting data, protecting the confidentiality of data, through symmetric key encryption and decryption. SHA-256 algorithm: used to generate the hash value of data, to ensure the integrity of data and verify its tamper resistance.

[0037] Step S105, send the encrypted data packet to the cloud server through the main transmission channel, if the network jitter is detected to exceed the preset threshold, switch to the standby transmission channel.

[0038] Send the encrypted data packet to the cloud server through the main transmission channel, get the real-time network jitter value, judge whether the real-time network jitter value exceeds the preset threshold, get the network state of the main transmission channel. If the real-time network jitter value exceeds the preset threshold, activate the channel switching mechanism, switch from the main transmission channel to the standby transmission channel, determine the switching completion state. Send the encrypted data packet through the standby transmission channel, get the network jitter value of the standby transmission channel, judge the transmission stability, get the standby transmission channel state. Use the sliding window function of pandas library to analyze the network jitter value sequence of the standby transmission channel, judge whether the network jitter value sequence returns to below the preset threshold. If the main transmission channel is restored to be feasible, switch back from the standby transmission channel to the main transmission channel, get the transmission performance of the main transmission channel after switching, judge the transmission stability, get the main transmission channel state. Through the encrypted data packet of the main transmission channel or the standby transmission channel, get the receiving confirmation information of the cloud server, judge the data integrity, get the transmission result. According to the transmission result, use Log4j log recording tool to store the transmission state and channel switching record, determine the business continuity, get the system running state.

[0039] It should be noted that Pandas is a core library in Python for data processing, analysis and cleaning, which provides efficient and flexible data structures (such as Series and DataFrame), making data operation simple and fast. Log4j is a tool for recording logs, which provides flexible configuration options and multiple log output methods, helping developers efficiently record, manage and analyze log information, supporting the recording and management of transmission state and channel switching events.

[0040] Step S106, decrypt and decompress the encrypted data packet in the cloud, analyze the construction progress time series data using a distributed computing framework, and generate a deviation detection result.

[0041] The encrypted data packet is obtained from the cloud and is encrypted using the AES algorithm. The encrypted data packet is decrypted using a preset key to obtain a decrypted data packet. The decrypted data packet is processed by a ZIP decompression algorithm to generate construction progress time series data. The construction progress time series data is analyzed using an Apache Spark distributed computing framework to extract construction progress features and obtain a progress feature set. If the deviation of the progress feature set from a preset progress plan exceeds a preset threshold, the progress is marked as abnormal to obtain a deviation mark. According to the deviation mark, the K-Means clustering algorithm is used to group abnormal data to generate an abnormal grouping result. The adjustment direction is determined by analyzing the abnormal grouping result, and an adjustment scheme is generated. For the adjustment scheme, Matplotlib is used to generate a visual time series chart to obtain progress monitoring output.

[0042] It should be noted that the above Apache Spark is an open source big data processing framework for fast and general data processing tasks. It supports distributed computing, can process large-scale data sets, and has high efficiency. Matplotlib is a widely used Python plotting library that provides an easy-to-use way to generate various static, dynamic, and interactive charts.

[0043] In step S107, the routing rules of the data link are dynamically adjusted according to the deviation detection result, and the processing priority is re-allocated to optimize the link configuration.

[0044] Real-time network probe data is acquired, the real-time network probe data including a link traffic matrix, and three characteristic values of byte number per second, packet loss rate, and delay are extracted. The characteristic values are subjected to Z-score standardization processing to obtain a standardized characteristic set. The standardized characteristic set is input into a K-Means clustering device with a preset K value of 2 to obtain an abnormal traffic marking result. If the proportion of abnormal traffic in the abnormal traffic marking result exceeds a preset threshold of 5%, a routing rule draft is generated according to a historical routing table. The routing rule draft is input into a routing strategy module of FRRouting to obtain a routing table update scheme containing a next hop address and a weight value. According to the weight value in the routing table update scheme, a server weight parameter is set on a HAProxy load balancer, and a traffic control queue of a corresponding priority is created through a Linux TC module. Real-time monitoring data of the HAProxy is acquired, and a link load variance value is calculated. If the load variance value exceeds a preset threshold of 0.2, a minimum connection number algorithm is used to recalculate the server weight parameter. The recalculated weight parameter is written into a HAProxy configuration file, a hot update operation is performed, and an updated link configuration snapshot is obtained. Link state data of the updated link configuration snapshot is continuously collected, the traffic characteristic values are recalculated and input into the K-Means clustering device, and a new abnormal traffic marking result is obtained.

[0045] It should be noted that FRRouting (FRR) is a powerful routing protocol implementation suite that allows users to manage and configure complex routing policies in Linux systems. The routing policy module provides network administrators with flexible ways to control routing decisions, filtering, and path selection, suitable for various routing protocol environments.

[0046] HAProxy is a high-efficiency load balancer used to distribute traffic to multiple backend servers, and a weight can be set for each server to control the proportion of traffic allocation. Linux TC is a powerful traffic control tool that can be used for traffic classification, bandwidth allocation, and priority management to ensure reasonable utilization and efficient distribution of traffic.

[0047] In step S108, the BIM component state is updated based on the optimized link configuration, and a real-time updated digital twin model is generated using cloud computing resources.

[0048] A parameter set is obtained from the link configuration, the parameter set including link delay and computing task allocation information. The parameter set is parsed to determine a state update rule of a BIM component, the state update rule including a computing priority adjustment condition. If the link delay in the parameter set exceeds a preset threshold, the computing task is redistributed through a cloud computing resource to obtain an adjusted resource scheduling scheme. According to the resource scheduling scheme, the computing priority of the BIM component is dynamically adjusted to generate updated component state data. Through the cloud computing resource, a synchronization operation is performed on the updated component state data to obtain a real-time updated data synchronization result. The data synchronization result is mapped to a virtual model to generate a preliminary digital twin model. If the update frequency of the preliminary digital twin model is lower than a preset threshold, a K-means algorithm is used for cluster analysis of the model update frequency to obtain optimized model generation parameters. According to the optimized model generation parameters, the rendering frequency of the digital twin model is adjusted to generate a final real-time updated digital twin model.

[0049] In step S109, three-dimensional visualization data is extracted from the digital twin model, and a rendering technique is used to map sensor data to a BIM model interface.

[0050] Specifically, three-dimensional visualization data is obtained from the digital twin model, a JSON parsing tool is used to parse the data structure, the correspondence between data points and the BIM model is determined, and mapping parameters of the three-dimensional visualization data are obtained. According to the mapping parameters, sensor data and geometric nodes of the BIM model are matched through WebGL to generate preliminary visualization interface data. If the refresh frequency of the preliminary visualization interface data is lower than a preset threshold, an ARIMA model is used to optimize the refresh frequency to obtain adjusted refresh frequency parameters. According to the adjusted refresh frequency parameters, WebSocket is used to dynamically update the visualization interface data to generate real-time synchronized interface display data. The real-time synchronized interface display data is distributed stored through HDFS to obtain distributed stored interface data. Key frame data is extracted from the distributed stored interface data, a Diff algorithm is used to update the BIM model interface to generate a final dynamically updated visualization interface. If the data synchronization delay of the final dynamically updated visualization interface exceeds a preset threshold, a Ketama algorithm is used to redistribute cloud computing resources to obtain an optimized resource scheduling scheme, and the final dynamically updated visualization interface is updated.

[0051] It is worth noting that WebGL (Web Graphics Library) is a powerful web-based graphics rendering technology that allows developers to perform efficient 3D graphics rendering in the browser. Through WebGL, geometric data in BIM models and real-time sensor data can be combined to achieve dynamic and interactive visualization. This method not only improves the real-time monitoring capability of BIM models, but also provides users with a more intuitive data display experience. WebSocket is a communication protocol that establishes a persistent, full-duplex (bidirectional) communication channel between the client (such as a browser) and the server. It allows both parties to exchange data in real time on an open connection, overcoming the limitations of the request-response mode in traditional HTTP protocols, and is particularly suitable for applications with high real-time requirements.

[0052] HDFS (Hadoop Distributed File System) is one of the core components of the Hadoop ecosystem, used for large-scale data storage. HDFS is designed to run on commodity hardware and can efficiently and fault-tolerantly store and manage large amounts of data, making it particularly suitable for big data processing; Diff algorithm is an algorithm used to compare two files or data sets and find their differences, its core purpose is to identify the differences between two versions of text or data.

[0053] In the above embodiments, the description of each embodiment has its own focus, and the parts not detailed or described in a certain embodiment can be referred to the relevant description of other embodiments.

[0054] The above is only the preferred embodiment of the present application, and does not limit the present application in any form. Although the present application has been disclosed as above, it is not intended to limit the present application. Any person skilled in the art can make some changes or modifications to the above disclosed technical content without departing from the scope of the present application, and any simple modification, equivalent change and modification of the above embodiments based on the technical essence of the present application still belong to the scope of the present application.

Claims

1. A construction dynamic twin method based on Internet of Things and BIM, characterized in that, The method comprises: acquiring a heterogeneous data stream collected by a sensor, parsing and classifying the heterogeneous data stream using a preset format template to generate a standardized data set; performing semantic mapping on the standardized data set according to an association rule of a BIM component identifier and a data attribute to generate a component association data set; if the matching accuracy of the component association data set is lower than a preset threshold, training a machine learning model using historical deviation data, optimizing the association rule parameters and generating an optimized component association data set; extracting a real-time data stream from the optimized component association data set, compressing and encrypting the real-time data stream using an edge computing node to generate an encrypted data packet; sending the encrypted data packet to a cloud server through a main transmission channel, and switching to a backup transmission channel if network jitter is detected to exceed a preset threshold; decrypting and decompressing the encrypted data packet in the cloud, analyzing construction progress time sequence data using a distributed computing framework to generate a deviation detection result; according to the deviation detection result, dynamically adjusting the routing rules of the data link, and reassigning the processing priority to optimize the link configuration, including: acquiring real-time network probe data and extracting characteristic values; the real-time network probe data includes a link flow matrix; the characteristic values include the number of bytes per second, packet loss rate and delay; performing Z-score standardization processing on the characteristic values to obtain a standardized feature set; inputting the standardized feature set into a K-Means clusterer with a preset K value of 2 to obtain an abnormal traffic marking result; if the proportion of abnormal traffic in the abnormal traffic marking result exceeds a preset threshold of 5%, generating a routing rule draft according to a historical routing table; inputting the routing rule draft into a routing strategy module to obtain a routing table update scheme containing a next hop address and a weight value; according to the weight value in the routing table update scheme, setting server weight parameters on a load balancer and creating a traffic control queue of corresponding priority; acquiring real-time monitoring data and calculating the load variance value of each link; if the load variance value exceeds a preset threshold, recalculating the server weight parameters using the minimum connection number algorithm; writing the recalculated server weight parameters into a configuration file and performing a hot update operation to obtain an updated link configuration snapshot; continuously collecting link state data of the updated link configuration snapshot, recalculating the traffic characteristic values and inputting them into the K-Means clusterer to obtain a new abnormal traffic marking result; updating the BIM component state based on the optimized link configuration, and generating a real-time updated digital twin model using cloud computing resources; extracting three-dimensional visualization data from the digital twin model and mapping sensor data to the BIM model interface using rendering technology.

2. The method of claim 1, wherein, The acquisition of a heterogeneous data stream collected by a sensor, the parsing and classification of the heterogeneous data stream using a preset format template to generate a standardized data set, comprises: acquiring a heterogeneous data stream from a sensor, verifying data stream integrity using stream processing technology to obtain an initial data stream; matching the initial data stream with a preset template, if the template and the data stream characteristics are consistent, extracting key fields to obtain structured data; The structured data is decomposed by using a decision tree algorithm, and if the feature meets the classification condition, a category label is assigned to obtain classified data; According to the classified data, a data standardization technology is used to map to a unified format to obtain a standardized data set; From the standardized data set, time series features are extracted, and if the time series features meet the continuity condition, a dynamic data stream is generated to obtain a dynamic data set; The dynamic data set is grouped by using a K-means clustering algorithm, and if the grouping result meets the similarity threshold, a grouped data set is generated to obtain a final data set; The final data set is encoded by using a data formatting technology to obtain a formatted data set.

3. The method of claim 1, wherein, The association rules of the BIM component identification and data attributes are used to perform semantic mapping on the standardized data set to generate a component association data set, including: The association rules of the BIM component identification and data attributes are obtained from the standardized data set, and the semantic mapping technology is used to match the component identification, and if the rules and attribute features are consistent, the associated fields are extracted to obtain the component association data set; According to the component association data set, the K-means clustering algorithm is used to group the component identification, and if the grouping result meets the preset similarity threshold, the component category is determined to obtain a classified data set; From the classified data set, the associated fields of the component identification are extracted, and the associated fields are structured by using a preset template, and if the associated fields meet the template format, a structured data set is generated; According to the structured data set, the consistency of the component identification and data attributes is verified by using a rule verification technology, and if the consistency meets the preset condition, a verification data set is obtained; From the verification data set, the semantic features of the component association are extracted, and the semantic features are analyzed by using a feature decomposition technology, and if the semantic features meet the decomposition condition, a semantic data set is obtained; According to the semantic data set, the semantic data set is mapped to a unified format by using a format conversion technology, and if the format conversion result meets the standard specification, a formatted data set is obtained; From the formatted data set, the time series features of the component association are extracted, and if the time series features meet the continuity condition, a dynamic association data set is generated.

4. The method of claim 1, wherein, If the matching accuracy of the component association data set is lower than the preset threshold, a historical deviation data is used to train a machine learning model, the association rule parameters are optimized, and an optimized component association data set is generated, including: The component association data set is obtained from the storage library, and it is judged whether the matching accuracy of the component association data set is lower than the preset threshold; If the matching accuracy is lower than the preset threshold, the historical deviation data is obtained from the storage library to obtain a deviation data set; According to the deviation data set, a random forest model is trained, the hyperparameters of the model are set to default values, the model weights are adjusted, and an optimized training model is obtained; The optimized training model is used to analyze the component association rules to extract rule parameters to obtain a parameter set; It is judged whether the deviation value of the parameter set is beyond the preset range; If the deviation value is out of a preset range, the parameter set is adjusted by a gradient descent algorithm, a learning rate is set as a preset value, an iteration number is set as a preset number, and an adjusted parameter set is obtained; An associated rule is updated according to the adjusted parameter set, and a new associated rule set is generated; The matching accuracy is recalculated by using the new associated rule set, and updated matching accuracy is obtained; It is judged whether the updated matching accuracy is lower than a preset threshold value; If the updated matching accuracy is lower than the preset threshold value, secondary deviation features are extracted from the deviation data, the random forest model is repeatedly trained, and a further optimized training model is obtained.

5. The method of claim 1, wherein, The real-time data stream is extracted from the optimized component association data set, and the edge computing node is used to compress and encrypt the real-time data stream to generate an encrypted data packet, including: The structured real-time data stream is obtained from the optimized component association data set, and the real-time data stream is generated into a continuous data stream according to a pre-established field filtering rule; The continuous data stream is received by using a framework on the edge computing node, and is compressed in a streaming manner to obtain a compressed data stream; If the volume of the compressed data stream exceeds a preset threshold value, the compressed data stream is cut into pieces with a fixed size to obtain an ordered data piece set; Each data piece in the ordered data piece set is encrypted by using an AES-256 algorithm in cooperation with a pre-distributed key to obtain an encrypted data piece set; The encrypted data piece set is packaged by the edge computing node according to the serial number, a metadata header containing data length and piece number is added, and a complete encrypted data packet is obtained; The hash value of the metadata header of the complete encrypted data packet is calculated by using a SHA-256 algorithm, and if the hash value matches a pre-stored check code, the complete encrypted data packet is sent to a target node address that has been authenticated.

6. The method of claim 1, wherein, The encrypted data packet is sent to the cloud server through the main transmission channel, and if it is detected that network jitter exceeds a preset threshold value, the backup transmission channel is switched to, including: The encrypted data packet is sent to the cloud server through the main transmission channel, the real-time network jitter value is obtained, it is judged whether the real-time network jitter value exceeds the preset threshold value, and the network state of the main transmission channel is obtained; If the real-time network jitter value exceeds the preset threshold value, the channel switching mechanism is activated, the main transmission channel is switched to the backup transmission channel, and the switching completion state is determined; The encrypted data packet is sent through the backup transmission channel, the network jitter value of the backup transmission channel is obtained, the transmission stability is judged, and the backup transmission channel state is obtained; The sliding window function of the pandas library is used to analyze the network jitter value sequence of the backup transmission channel, and it is judged whether the network jitter value sequence returns to below the preset threshold value; If the main transmission channel is restored to be feasible, the main transmission channel is switched back from the backup transmission channel, the transmission performance of the main transmission channel after switching is obtained, the transmission stability is judged, and the main transmission channel state is obtained; The encrypted data packet is obtained through the main transmission channel or the backup transmission channel, the receiving confirmation information of the cloud server is obtained, the data integrity is judged, and the transmission result is obtained; According to the transmission result, a log recording tool is used to store transmission status and channel switching records, determine service continuity, and obtain system operation status.

7. The method of claim 1, wherein, The encrypted data packet is decrypted and decompressed in the cloud, and a distributed computing framework is used to analyze construction progress time series data to generate deviation detection results, including: An encrypted data packet is obtained from the cloud, which is encrypted using the AES algorithm; The encrypted data packet is decrypted using a preset key to obtain a decrypted data packet; The decrypted data packet is processed by a ZIP decompression algorithm to generate construction progress time series data; The construction progress time series data is analyzed and construction progress features are extracted to obtain a progress feature set; If the deviation of the progress feature set from the preset progress plan exceeds a preset threshold, it is marked as abnormal and a deviation mark is obtained; According to the deviation mark, the K-Means clustering algorithm is used to group abnormal data to generate an abnormal grouping result; The progress adjustment direction is determined by analyzing the abnormal grouping result to generate an adjustment scheme; A visual time series graph is generated for the adjustment scheme to obtain progress monitoring output.

8. The method of claim 1, wherein, The BIM component state is updated based on the optimized link configuration, and cloud computing resources are used to generate a real-time updated digital twin model, including: A parameter set is obtained from the link configuration, including link delay and computing task allocation information; The parameter set is parsed to determine the state update rule of the BIM component, including the computing priority adjustment condition; If the link delay in the parameter set exceeds a preset threshold, the computing task is redistributed through cloud computing resources to obtain an adjusted resource scheduling scheme; According to the adjusted resource scheduling scheme, the computing priority of the BIM component is dynamically adjusted to generate updated component state data; Through cloud computing resources, a synchronization operation is performed on the updated component state data to obtain a real-time updated data synchronization result; The real-time updated data synchronization result is mapped to the virtual model to generate a preliminary digital twin model; If the update frequency of the preliminary digital twin model is lower than a preset threshold, the K-means algorithm is used to cluster analyze the update frequency of the model to obtain optimized model generation parameters; According to the optimized model generation parameters, the rendering frequency of the digital twin model is adjusted to generate the final real-time updated digital twin model.

9. The method of claim 1, wherein, The three-dimensional visualization data is extracted from the digital twin model, and the sensor data is mapped to the BIM model interface using rendering technology, including: Three-dimensional visualization data is obtained from the digital twin model, the data structure is parsed, and the correspondence between data points and BIM models is determined to obtain mapping parameters for three-dimensional visualization data; According to the mapping parameters, the sensor data is matched with the geometric nodes of the BIM model to generate preliminary visualization interface data; If the refresh frequency of the preliminary visualization interface data is lower than a preset threshold, the ARIMA model is used to optimize the refresh frequency to obtain adjusted refresh frequency parameters; According to the adjusted refresh frequency parameter, the visual interface data is dynamically updated to generate real-time synchronized interface display data; The real-time synchronized interface display data is distributed stored to obtain distributed stored interface data; Key frame data is extracted from the distributed stored interface data, a Diff algorithm is used to update a BIM model interface to generate a final dynamic updated visual interface; If the data synchronization delay of the final dynamic updated visual interface exceeds a preset threshold, a Ketama algorithm is used to re-distribute cloud computing resources to obtain an optimized resource scheduling scheme, and the final dynamic updated visual interface is updated.

Images