5G-based workshop digital visualization method and system

Abstract

The invention discloses a 5G-based workshop digital visualization method and system, and the method comprises the steps: obtaining collection data from a workshop equipment sensor, and obtaining a compressed data flow; transmitting the compressed data stream to a cloud processing platform, judging the integrity of the data fragments, and if the data fragments are complete, storing the data fragments to a distributed database to obtain a structured data set; for the structured data set, obtaining an equipment state feature set; performing dynamic mapping on the virtual model and the physical equipment through a space-time calibration mechanism according to the equipment state feature set, judging whether a mapping error is lower than a preset threshold value or not, and if the mapping error is lower than the threshold value, updating the virtual model to obtain a real-time synchronous digital twin model; and extracting equipment collaboration information from the real-time synchronous digital twinborn model to obtain visual presentation data of the production process. According to the invention, real-time monitoring and accurate fault prediction of the equipment cooperation state are realized, and the production efficiency and the intelligent level of equipment management are improved.

Description

Technical Field

[0001] This invention relates to the field of workshop monitoring technology, and in particular discloses a 5G-based workshop digital visualization method and system. Background Technology

[0002] With the rapid development of intelligent manufacturing, real-time monitoring and visualization of workshop production data have become crucial for improving production efficiency and management. Manufacturing enterprises urgently need to achieve transparency and intelligence in their production processes through digitalization to cope with complex and ever-changing market demands. However, existing solutions have significant shortcomings in achieving real-time visualization of all elements.

[0003] Many systems rely on traditional network transmission, which limits data acquisition and processing speed, making it difficult to meet the high-frequency, real-time data interaction needs of large-scale equipment and complex scenarios. Furthermore, data silos are prevalent, hindering efficient information integration between devices, leading to delayed production decisions and impacting overall efficiency. Against this backdrop, digital visualization of the workshop faces two core technological challenges.

[0004] First, the high-speed collection and transmission of equipment operating parameters, energy consumption data, and material flow information in the production environment presents a challenge. Due to the wide variety of equipment and the massive amount of data in the workshop, the bandwidth and latency limitations of traditional networks make real-time data transmission difficult to achieve. For example, in high-precision processing scenarios, microsecond-level changes in equipment status cannot be synchronized to the management platform in a timely manner, leading to production scheduling delays.

[0005] Secondly, there are technical bottlenecks in the dynamic mapping between physical equipment and virtual models. Workshop production scenarios are complex, with equipment states and environmental factors changing rapidly. Without an efficient spatiotemporal calibration mechanism, virtual models struggle to accurately reflect the real-time state of physical entities. For example, in collaborative operations involving multiple devices, a malfunction in one device may go undetected due to delayed model updates, potentially leading to production interruptions.

[0006] Therefore, how to achieve real-time collection, processing, and visualization of all elements in the workshop through high-speed, low-latency data transmission technology and precise digital twin mapping mechanism has become a key issue in the upgrading of intelligent manufacturing. Summary of the Invention

[0007] This invention provides a 5G-based workshop digital visualization method and system, aiming to solve at least one of the defects existing in the prior art.

[0008] One aspect of the present invention relates to a 5G-based digital visualization method for workshops, comprising the following steps: S100: Acquire data from workshop equipment sensors, perform dimensionality reduction processing on high-frequency data in the acquired data to obtain a compressed data stream. The acquired data includes production data, energy consumption data and material flow information. S200 transmits the compressed data stream to the cloud processing platform via a high-speed, low-latency network. It uses data sharding technology to partition and manage the data stream, determines the integrity of the data shards, and if the data shards are complete, stores them in a distributed database to obtain a structured data set. S300: For structured data sets, a streaming processing framework is used to perform real-time analysis of equipment status, and a clustering algorithm is used to classify equipment operating parameters to obtain a set of equipment status features. S400. Based on the device status feature set, construct a digital twin model, dynamically map the virtual model and the physical device through a spatiotemporal calibration mechanism, determine whether the mapping error is lower than a preset threshold, and update the virtual model if it is lower than the threshold to obtain a real-time synchronized digital twin model. S500 extracts equipment collaboration information from the real-time synchronized digital twin model, uses anomaly detection algorithms to identify potential faults, determines whether the equipment collaboration status is normal, and generates a fault alarm if an anomaly is found, thus obtaining visualized data of the production process.

[0009] Further, step S100 includes: S110. Obtain sensor data, energy consumption data, and material flow information from workshop equipment sensors, and perform standardized format conversion on the sensor data, energy consumption data, and material flow information to obtain a dataset in a unified format. The standardized dataset includes standardized sensor data, standardized energy consumption data, and standardized material flow information. The standardized sensor data is described as follows:

[0010] in, This represents the standardized sensor data. This represents the raw data acquired from the sensors of the workshop equipment. This represents the mean of the original data. This represents the standard deviation of the original data; Energy consumption data in a standardized format is expressed as follows:

[0011] in, This represents the normalized energy consumption data. This indicates the currently collected energy consumption value. This represents the minimum value of energy consumption data. This represents the maximum value of the energy consumption data; The standardized format for material flow information is as follows:

[0012] in, This represents material flow information in a standardized format. Represents material flow data. This represents the material flow time data. Indicates material location data. , , These represent the weighting coefficients for each dimension of the data; S120. Perform preprocessing operations on the unified format dataset through edge computing nodes to remove noise and redundant data in the unified format dataset and obtain the cleaned dataset. The cleaned dataset is represented as follows:

[0013] in, This represents the cleaned dataset. Represents the original, uniformly formatted dataset. Represents the first in the dataset Data points, Represents the mean vector of the dataset. Indicates standard deviation, This represents the noise detection threshold coefficient; S130. For the cleaned dataset, perform dimensionality reduction on the high-frequency data. If the frequency of the cleaned dataset exceeds a preset threshold, perform data aggregation to obtain a compressed data stream. The compressed data stream is represented as follows:

[0014] in, This represents the compressed data stream. This indicates the number of data blocks being aggregated. Indicates the first The weighting coefficients of each data block, Indicates the first One raw data block, Indicates the first The target dimensionality for dimensionality reduction of each data block; S140. Send the compressed data stream to the cloud storage system, determine whether there is data loss during the transmission of the compressed data stream, and if there is loss, trigger the retransmission mechanism to obtain the complete stored dataset. The following formula can be used to detect whether data loss occurs during the transmission of compressed data streams:

[0015] in, This represents the probability of data loss during data transmission. Indicates the total number of data packets transmitted. Indicates the first The integrity verification function for each data packet, when receiving data. With sending data Returns 1 if they match, otherwise returns 0.

[0016] Further, step S200 includes: S210. The compressed data stream is transmitted through a high-speed, low-latency network, and the data stream is divided into segments using data fragmentation technology to obtain fragmented data streams. The uniform division of the data stream can be achieved using the following formula:

[0017] in, Indicates the first The size of each data partition Indicates the total size of the data stream to be segmented. This indicates the preset total number of fragments. Indicates the fragment sequence number; S220. For the fragmented data stream, perform parallel processing operations on the cloud processing platform, and integrate the fragments using multi-node parallel computing to obtain the integrated data set. The integrated dataset is described as follows:

[0018] in, This represents the integrated data set. This indicates the total number of data fragments. Indicates the first Data shards, This represents a data conversion function. Indicates the first The conversion parameters for each segment; S230. If the integrated data set passes the data integrity check, the integrated data set is stored in the distributed database to verify the consistency of each shard and obtain a structured data set. Data integrity verification is described as follows:

[0019] in, Indicates data integrity verification metrics. Indicates the total number of data fields. Indicates the total number of data records. Indicates the first The field number The weighting coefficient of each record. This function indicates the consistency between the field value and the reference value, returning 1 when the data is complete and 0 when it is missing or incorrect. The consistency of each fragment is expressed as follows:

[0020] in, This indicates the consistency verification results for each fragment. This represents the total number of shards in the distributed database. Indicates the first Each segment, Indicates the first The number of data items per shard Indicates the first The first segment One data item, This indicates that the consistency verification of the data item has passed the function. This represents the overall validation function for the data item.

[0021] Further, step S300 includes: S310. Receive real-time data streams from workshop equipment sensors, parse the data streams using a streaming processing framework, extract operating parameters, and obtain a set of equipment operating parameters. The extracted runtime parameters are described as follows:

[0022] in, Indicates the first The extracted runtime parameter values, This indicates the total number of data stream elements participating in the parsing. Indicates the first The weight coefficient of each data element, Indicates the first The data element is the first... The contribution value of each parameter; The set of equipment operating parameters is described as follows:

[0023] in, This represents the final set of equipment operating parameters. This indicates the total number of extracted runtime parameters. Indicates the first One running parameter, and These represent the minimum and maximum threshold ranges of the parameters, respectively. Set operations ensure that all parameters are within the valid range. S320. For the set of equipment operating parameters, the K-means clustering algorithm is used to classify the operating parameters. If the parameter value meets the preset threshold range, the classification result is determined and the equipment status classification set is obtained. The following formula is used to determine whether the parameter values ​​after clustering meet the preset threshold range. If the condition is met, the classification result is considered valid:

[0024] in, This represents the minimum value of the preset threshold. This represents the maximum value of the preset threshold. Indicates the first The values ​​of each running parameter; S330. Based on the equipment status classification set, the classification results are partitioned and stored through a distributed storage system, and integrity verification is performed to obtain structured status data. The following formula is used to calculate the hash when partitioning classification results in a distributed storage system:

[0025] in, Indicates partition The storage hash value, Indicates partition The Middle The content of each data block This indicates the total number of data blocks in the partition. Indicates the first Redundancy of each data block and Indicates the weighting coefficient; Perform integrity verification on the stored structured state data:

[0026] in, Indicates time Integrity verification metrics This indicates the total number of data blocks participating in the verification. Indicates the first One data block, Indicates the first Cyclic redundancy check value of each data block Indicates the first data blocks Check value, and Represents the weighting coefficients for different verification methods; S340. For structured state data, a streaming processing framework is used to extract features from the structured state data to generate a first device state feature set. The first device state feature set is described as follows: in, Represents the first device state feature set. Represents the candidate feature vector. This indicates the number of dimensions in the structured state data. Indicates the first Weighting coefficients for dimensional data Indicates the first Feature mapping function for dimensional data, Indicates the first Dimensional input structured state data, Indicates the feature extraction parameters. Represents the parameter space.

[0027] Further, step S400 includes: S410: Obtain real-time operating data streams from the workshop equipment sensors of the physical equipment, parse the real-time operating data streams using streaming processing tools, extract equipment status parameters, and generate a second equipment status feature set; S420. Based on the second device state feature set, a spatiotemporal calibration mechanism is used to align the virtual model and the physical device in terms of time and space dimensions to obtain the aligned mapping relationship. S430. For the aligned mapping relationship, calculate the mapping error E between the virtual model and the physical device state, where the mapping error E represents the deviation between the virtual model output and the physical device state. If the mapping error E is lower than the preset threshold, it is determined that the model needs to be updated and an update signal is generated. S440. Based on the update signal, adjust the virtual model parameters using database tools to obtain a real-time synchronized digital twin model.

[0028] Furthermore, in step S410, the device status parameters are described as follows:

[0029] in, Indicates the first Each device status characteristic Indicates the size of the streaming processing window. Indicates the first A feature extraction function, Indicates time Structured state data, Indicates the time decay factor. Indicates the current moment. e Represents the natural constant.

[0030] Furthermore, in step S420, the aligned mapping relationship is expressed as follows:

[0031] in, The function representing the aligned mapping relationship. This represents the weighting coefficient for the time dimension. This represents the spatial dimension weight coefficient. Represents the weight matrix along the time dimension. Represents the spatial dimension weight matrix. Represents the device state vector. This represents the state vector of the virtual model.

[0032] Furthermore, in step S430, the update signal is expressed as follows:

[0033] in, Indicates an update signal. This represents the calculated mapping error value. This represents the preset error threshold. When the mapping error is lower than the threshold, an update signal of 1 is output; otherwise, 0 is output. This is used to determine whether the model needs to be updated. In step S440, the real-time synchronized digital twin model is described as follows:

[0034] in, This indicates the state of the digital twin model after real-time synchronization. This indicates the current state of the virtual model. Indicates the synchronous attenuation coefficient. Represents real system data, Represents virtual system data. This indicates an element-wise multiplication operation.

[0035] Further, step S500 includes: S510: Extract device collaboration information from the real-time synchronized digital twin model, parse the device collaboration information through streaming processing tools, and generate a first feature set containing the interaction states of multiple devices. S520. The isolated forest algorithm is used to process the first feature set and calculate the abnormal score S of the interaction state of each device. The abnormal score S represents the degree to which the collaborative state of the devices deviates from the normal value. If the abnormal score S is higher than the preset threshold, a fault alarm signal is generated. S530. Based on the fault alarm signal and combined with the production process data, the fault alarm signal and production operation parameters are integrated through database tools to generate a second dataset containing alarm status and production data. S540. Process the second dataset using visualization tools, and use dashboard generation technology to draw a dynamic view of the production process, obtaining real-time updated production status data.

[0036] Another aspect of the present invention relates to a 5G-based workshop digital visualization system, comprising: a compressed data stream acquisition module, used to acquire collected data from workshop equipment sensors, perform dimensionality reduction processing on high-frequency data in the collected data to obtain a compressed data stream, wherein the collected data includes production data, energy consumption data and material flow information; The structured data set acquisition module is used to transmit compressed data streams to the cloud processing platform via a high-speed, low-latency network. It uses data sharding technology to partition and manage the data stream, and judges the integrity of the data shards. If the data shards are complete, they are stored in a distributed database to obtain a structured data set. The device status feature set acquisition module is used to perform real-time analysis of device status using a streaming processing framework for structured data sets, and to classify device operating parameters through clustering algorithms to obtain device status feature sets. The digital twin model acquisition module is used to construct a digital twin model based on the device status feature set, dynamically map the virtual model and the physical device through a spatiotemporal calibration mechanism, determine whether the mapping error is lower than a preset threshold, and update the virtual model if it is lower than the threshold to obtain a real-time synchronized digital twin model. The visualization data acquisition module is used to extract equipment collaboration information from the real-time synchronized digital twin model, use anomaly detection algorithms to identify potential faults, determine whether the equipment collaboration status is normal, and generate fault alarms if an anomaly is found, thus obtaining visualization data of the production process.

[0037] The beneficial effects achieved by this invention are as follows: This invention provides a 5G-based workshop digital visualization method and system. Addressing the business scenario of efficient data collection, real-time analysis, and fault early warning for workshop equipment, it preprocesses high-frequency production data, energy consumption data, and material data through edge computing and time-series compression algorithms to reduce dimensionality. This data is then transmitted to the cloud via a high-speed, low-latency network. Data fragmentation and integrity verification ensure reliable storage of the data as a structured dataset. Furthermore, a streaming processing framework and clustering algorithms are used to analyze equipment status characteristics in real time, constructing and dynamically updating a digital twin model. A spatiotemporal calibration mechanism reduces mapping errors, and finally, an anomaly detection algorithm identifies potential faults and generates visual early warnings. This invention achieves real-time monitoring of equipment collaborative status and accurate fault prediction, improving production efficiency and the level of intelligent equipment management. Attached Figure Description

[0038] Figure 1 This is a flowchart illustrating an embodiment of a 5G-based workshop digital visualization method according to the present invention. Figure 2 This is a functional block diagram of an embodiment of a 5G-based workshop digital visualization system according to the present invention.

[0039] Explanation of icon numbers: 10. Compressed data stream acquisition module; 20. Structured data set acquisition module; 30. Equipment status feature set acquisition module; 40. Digital twin model acquisition module; 50. Visualized data acquisition module. Detailed Implementation

[0040] To better understand the above technical solutions, the following will provide a detailed explanation of the technical solutions in conjunction with the accompanying drawings and specific implementation methods.

[0041] like Figure 1 As shown, the first embodiment of the present invention proposes a 5G-based workshop digital visualization method, including the following steps: Step S100: Acquire data from the sensors of the workshop equipment, perform dimensionality reduction processing on the high-frequency data in the acquired data to obtain a compressed data stream. The acquired data includes production data, energy consumption data and material flow information.

[0042] Compressed data streams refer to simplified data sequences formed from raw data (including production data, energy consumption data, and material flow information) collected from workshop equipment sensors. For high-frequency fluctuating data (such as high-frequency sampled equipment vibration and real-time energy consumption pulse signals), dimensionality reduction techniques (such as principal component analysis, wavelet transform, and feature selection) are used to remove redundant information and noise, retaining only core features. Compressed data streams reduce data storage and transmission costs while preserving key data value (such as equipment status trends, abnormal energy consumption characteristics, and material flow nodes), providing efficient input for subsequent data analysis (such as equipment fault early warning and energy efficiency optimization).

[0043] Step S200: Transmit the compressed data stream to the cloud processing platform via a high-speed, low-latency network. Use data sharding technology to partition and manage the data stream. Determine the integrity of the data shards. If the data shards are complete, store them in a distributed database to obtain a structured data set.

[0044] Structured datasets refer to standardized, traceable collections of data formed by transmitting compressed data streams from the workshop to a cloud processing platform via high-speed, low-latency networks, partitioning them by time, device, or data type using data sharding technology, verifying the integrity of each shard (such as checking data checksums and matching shard boundaries), and storing them in a distributed database. Through end-to-end management of "transmission-sharding-verification-storage," structured datasets transform fragmented compressed data into structured data resources with a unified format, complete relationships, and support for efficient querying and analysis, providing reliable data support for workshop production monitoring, energy efficiency analysis, and process optimization.

[0045] Step S300: For the structured data set, a streaming processing framework is used to analyze the equipment status in real time, and the equipment operating parameters are classified by a clustering algorithm to obtain the equipment status feature set.

[0046] Equipment status feature sets refer to a set of core features extracted from structured cloud-based datasets. These features are derived by using a streaming processing framework to analyze high-frequency operating parameters (such as production data and energy consumption data) in real time, and then classifying these parameters using clustering algorithms (such as K-means and DBSCAN). The extracted features characterize the equipment's normal operation, abnormal fluctuations, and potential faults. By condensing discriminative status indicators from real-time data streams, equipment status feature sets enable quantitative descriptions and pattern recognition of equipment operating states, providing feature-level data support for equipment monitoring, fault warning, and performance optimization.

[0047] Step S400: Based on the device status feature set, construct a digital twin model, dynamically map the virtual model and the physical device through a spatiotemporal calibration mechanism, determine whether the mapping error is lower than a preset threshold, and update the virtual model if it is lower than the threshold to obtain a real-time synchronized digital twin model.

[0048] A real-time synchronized digital twin model refers to a virtual model that maps to all elements of a physical device, driven by a set of device state features. Through a spatiotemporal calibration mechanism (time axis alignment, spatial parameter matching), the virtual model dynamically correlates with the physical device's operating state. The mapping error between the virtual model and the physical device (such as state feature deviation and timing synchronization difference) is continuously calculated. When the mapping error is below a preset threshold (e.g., feature deviation ≤ 5%, timing difference ≤ 100ms), the virtual model is updated based on real-time data from the physical device, ultimately forming a digital mirror that is highly consistent with and continuously synchronized with the physical device in terms of state, behavior, and performance. The real-time synchronized digital twin model, through a closed loop of "feature-driven - dynamic mapping - error calibration - model update," achieves real-time and accurate replication of the physical device, providing a digital carrier for equipment monitoring and optimized control.

[0049] Step S500: Extract equipment collaboration information from the real-time synchronized digital twin model, use an anomaly detection algorithm to identify potential faults, determine whether the equipment collaboration status is normal, generate a fault alarm if an anomaly is found, and obtain visualized data of the production process.

[0050] The visualization of the production process refers to extracting collaborative operation information between equipment (such as interaction timing, parameter matching degree, and material flow efficiency) from a real-time synchronized digital twin model. This information is then used to identify potential faults where collaborative parameters deviate from normal ranges through anomaly detection algorithms (such as Isolation Forest and LSTM anomaly prediction). When anomalies are detected in equipment collaboration, structured fault alarms are generated. Finally, this data is integrated into a visualization dataset containing real-time equipment status, collaborative relationship topology, anomaly alarm identifiers, and trend curves. The visualization of the production process transforms abstract equipment collaboration data and fault information into intuitive and perceptible visual forms (such as dynamic charts, topology diagrams, and alarm pop-ups), providing visual decision support for production monitoring, fault response, and process optimization.

[0051] Furthermore, the 5G-based workshop digital visualization method provided in this embodiment includes step S100: step S110, acquiring sensor data, energy consumption data and material flow information from workshop equipment sensors, and performing standardized format conversion on the sensor data, energy consumption data and material flow information to obtain a unified format dataset.

[0052] The standardized dataset includes standardized sensor data, standardized energy consumption data, and standardized material flow information. The standardized sensor data is described as follows:

[0053] In formula (1), This represents the standardized sensor data. This represents the raw data acquired from the sensors of the workshop equipment. This represents the mean of the original data. This represents the standard deviation of the original data.

[0054] Energy consumption data in a standardized format is expressed as follows:

[0055] In formula (2), This represents the normalized energy consumption data. This indicates the currently collected energy consumption value. This represents the minimum value of energy consumption data. This represents the maximum value of the energy consumption data.

[0056] The standardized format for material flow information is as follows:

[0057] In formula (3), This represents material flow information in a standardized format. Represents material flow data. This represents the material flow time data. Indicates material location data. , , These represent the weighting coefficients for each dimension of the data.

[0058] In a car manufacturing workshop setting, acquiring data from sensors on equipment is crucial for achieving smart manufacturing. Sensor data includes information such as temperature, pressure, and vibration; energy consumption data relates to the electricity consumed during equipment operation; and material flow information records the movement of raw materials and semi-finished products on the production line. Assume a welding robot's sensors collect temperature data 10 times per second in JSON format, containing fields such as temperature value and timestamp; energy consumption data is recorded in kilowatt-hours and generated once per minute; and material flow information is recorded in XML (Extensible Markup Language) format, recording pallet numbers and locations. Standardized format conversion requires unifying this data into JSON format and standardizing the timestamp to UTC (Universal Time Coordinated) format to ensure data consistency across devices. For example, temperature data {"temp":25.5", "timestamp":"2025-08-08T09:15:00Z"}, energy consumption data {"power":1.2", "timestamp":"2025-08-08T09:15:00Z"}, and material data {"pallet_id":"P123", "location":"station_1", "timestamp":"2025-08-08T09:15:00Z"}. In this embodiment, a unified format facilitates subsequent processing and reduces system complexity.

[0059] Step S120: Perform preprocessing operations on the unified format dataset through edge computing nodes to remove noise and redundant data in the unified format dataset, and obtain the cleaned dataset.

[0060] The cleaned dataset is represented as follows:

[0061] In formula (4), This represents the cleaned dataset. Represents the original, uniformly formatted dataset. Represents the first in the dataset Data points, Represents the mean vector of the dataset. Indicates standard deviation, This represents the noise detection threshold coefficient.

[0062] Edge computing nodes preprocess datasets in a standardized format. Noisy data, such as outliers like 50℃ (normal range 20-30℃) from temperature sensors due to interference, is filtered using a sliding window averaging method. Redundant data, such as duplicate timestamps, is removed. The cleaned dataset retains only valid data, reducing storage and computational burden. For example, 100 temperature data points within 10 seconds are compressed to 50 after denoising and deduplication, reducing the data volume by 50% and improving processing efficiency.

[0063] Step S130: For the cleaned dataset, perform dimensionality reduction on the high-frequency data. If the frequency of the cleaned dataset exceeds a preset threshold, perform data aggregation to obtain a compressed data stream.

[0064] The compressed data stream is represented as follows:

[0065] In formula (5), This represents the compressed data stream. This indicates the number of data blocks being aggregated. Indicates the first The weighting coefficients of each data block, Indicates the first One raw data block, Indicates the first The target dimensionality for dimensionality reduction of a data block.

[0066] High-frequency data dimensionality reduction processing is applied to temperature data collected 10 times per second. If the preset frequency threshold is 5 times per second, an aggregation operation is used to average every two data points, generating a data stream of 5 times per second. For example, 10 temperature data points [25.5, 25.7, 25.6, 25.8...] are aggregated into [25.6, 25.7...], halving the data volume, preserving trend information, and reducing transmission and storage costs.

[0067] Step S140: Send the compressed data stream to the cloud storage system, determine whether there is data loss during the transmission of the compressed data stream, and if there is loss, trigger the retransmission mechanism to obtain the complete stored dataset.

[0068] The following formula can be used to detect whether data loss occurs during the transmission of compressed data streams:

[0069] In formula (6), This represents the probability of data loss during data transmission. Indicates the total number of data packets transmitted. Indicates the first The integrity verification function for each data packet, when receiving data. With sending data Returns 1 if they match, otherwise returns 0.

[0070] The compressed data stream is sent to the cloud storage system via the MQTT (Message Queuing Telemetry Transport) protocol. During transmission, edge nodes record the checksum of each batch of data, which is then verified by the cloud. If data packet loss is detected (e.g., only 9 out of 10 packets are received), a retransmission mechanism is triggered, and the edge nodes resend the lost packets until the cloud confirms complete reception. This retransmission mechanism ensures data integrity, for example, preventing the loss of critical temperature data when the network is unstable, thus guaranteeing the accuracy of subsequent analysis.

[0071] Preferably, the 5G-based workshop digital visualization method provided in this embodiment includes step S200: step S210, transmitting the compressed data stream through a high-speed, low-latency network, and using data fragmentation technology to divide the data stream to obtain the fragmented data stream.

[0072] The uniform division of the data stream can be achieved using the following formula:

[0073] In formula (7), Indicates the first The size of each data partition Indicates the total size of the data stream to be segmented. This indicates the preset total number of fragments. Indicates the segment number.

[0074] The compressed data stream is transmitted via a high-speed, low-latency network to ensure real-time performance and reliability. High-speed, low-latency networks, such as 5G (5th Generation Mobile Communication Technology) networks or dedicated industrial Ethernet, can control data transmission latency to the millisecond level, meeting the production line's demands for real-time data processing. The welding robot generates a compressed temperature data stream per second, containing approximately five JSON-formatted data entries, each about 100 bytes, totaling 500 bytes per second. Through the 5G network, data is transmitted from edge nodes to the cloud processing platform with latency as low as 5 milliseconds, ensuring smooth and real-time data transmission. Compared to traditional 4G networks, 5G's high bandwidth supports larger-scale data transmission and reduces the risk of network congestion.

[0075] Data sharding technology divides compressed data streams into smaller fragments to optimize transmission and processing efficiency. For example, a 500-byte temperature data stream within one second is divided into five 100-byte fragments, each containing one temperature record and a timestamp. After sharding, each fragment is appended with metadata, such as a fragment number and a checksum, facilitating subsequent integration and verification. In this embodiment, sharding reduces the amount of data transmitted in a single transaction, minimizing the impact of network interruptions on overall data transmission, while also facilitating parallel processing in the cloud. During sharding, edge nodes record the hash value of each data fragment to ensure fragment integrity.

[0076] Step S220: For the fragmented data stream, perform parallel processing operations on the cloud processing platform, and integrate the fragments using multi-node parallel computing to obtain the integrated data set.

[0077] The integrated dataset is described as follows:

[0078] In formula (8), This represents the integrated data set. This indicates the total number of data fragments. Indicates the first Data shards, This represents a data conversion function. Indicates the first The transformation parameters for each segment.

[0079] After receiving the fragmented data stream, the cloud processing platform integrates it using multi-node parallel computing. Multiple computing nodes are deployed in the cloud, such as four nodes processing five fragments in parallel, with each node handling one or two fragments, integrating them into a complete dataset. Nodes read the timestamps and data content from the fragments and reconstruct the data stream in chronological order. For example, if the timestamps of the five fragments are from 2025-08-08T09:15:00.000Z to 2025-08-08T09:15:00.400Z, after integration, they form a continuous temperature data sequence. This parallel processing method reduces the processing time from 500 milliseconds in serial processing to approximately 150 milliseconds, significantly improving efficiency.

[0080] Step S230: If the integrated data set passes the data integrity check, the integrated data set is stored in the distributed database to verify the consistency of each shard and obtain a structured data set.

[0081] Data integrity verification is described as follows:

[0082] In formula (9), Indicates data integrity verification metrics. Indicates the total number of data fields. Indicates the total number of data records. Indicates the first The field number The weighting coefficient of each record. This function indicates the consistency between the field value and the reference value. It returns 1 when the data is complete and 0 when it is missing or incorrect.

[0083] The consistency of each fragment is expressed as follows:

[0084] In formula (10), This indicates the consistency verification results for each fragment. This represents the total number of shards in the distributed database. Indicates the first Each segment, Indicates the first The number of data items per shard Indicates the first The first segment One data item, This indicates that the consistency verification of the data item has passed the function. This represents the overall validation function for the data item.

[0085] The integrated dataset undergoes data integrity verification to ensure reliability. For example, the cloud platform calculates the overall checksum of the integrated data and compares it with the checksum sent by the edge nodes. If the checksums match, the dataset is considered complete; if inconsistencies are found, such as corrupted data shards, the platform requests the edge nodes to retransmit those shards. The verification process ensures no data loss or tampering. For example, if a shard's temperature value deviates abnormally from the normal range by 20-30°C, a retransmission mechanism is triggered to ensure data accuracy. The verified integrated dataset is stored in a distributed database, such as MongoDB or Cassandra, to support high concurrency access and scalability. In this embodiment, data is partitioned by timestamp, with temperature data stored in hourly partitions for faster querying. For example, temperature data from 09:00:00Z to 10:00:00Z on August 8, 2025, is stored in the same partition, improving query efficiency by approximately 30%.

[0086] Furthermore, in the 5G-based workshop digital visualization method provided in this embodiment, step S300 includes: Step S310: Receive real-time data streams from workshop equipment sensors, parse the data streams using a streaming processing framework, extract operating parameters, and obtain a set of equipment operating parameters.

[0087] The extracted runtime parameters are described as follows:

[0088] In formula (11), Indicates the first The extracted runtime parameter values, This indicates the total number of data stream elements participating in the parsing. Indicates the first The weight coefficient of each data element, Indicates the first The data element is the first... The contribution value of each parameter.

[0089] The set of equipment operating parameters is described as follows:

[0090] In formula (12), This represents the final set of equipment operating parameters. This indicates the total number of extracted runtime parameters. Indicates the first One running parameter, and These represent the minimum and maximum threshold ranges of the parameters, respectively. Set operations ensure that all parameters are within the valid range.

[0091] The welding robot's sensors collect data such as vibration, current, and rotation speed in real time, forming a data stream. This data stream is analyzed using a streaming processing framework to extract key operating parameters, thereby forming a set of equipment operating parameters.

[0092] The sensor generates approximately 1000 bytes of data per second, containing vibration frequency, current intensity, and rotational speed values. For example, the sensor might generate a data stream with a vibration frequency of 50-60Hz, a current intensity of 100-120A, and a rotational speed of 1500-1600rpm. These data streams are parsed in real time using streaming frameworks such as Apache Flink or Kafka Streams, transforming the raw byte streams into a structured set of runtime parameters for subsequent analysis.

[0093] Step S320: For the set of equipment operating parameters, the K-means clustering algorithm is used to classify the operating parameters. If the parameter values ​​meet the preset threshold range, the classification result is determined, and the equipment status classification set is obtained.

[0094] The following formula is used to determine whether the parameter values ​​after clustering meet the preset threshold range. If the condition is met, the classification result is considered valid:

[0095] In formula (13), This represents the minimum value of the preset threshold. This represents the maximum value of the preset threshold. Indicates the first The values ​​of each running parameter.

[0096] For the set of equipment operating parameters, the K-means clustering algorithm is used to classify the operating parameters to determine the equipment status. In this embodiment, K=3 is set, and the operating parameters are divided into three categories: normal, warning, and fault. The preset threshold ranges are vibration frequency 50-55Hz, current 100-110A, and speed 1500-1550rpm. If the operating parameter values ​​fall within these ranges, they are classified as normal; if the vibration frequency exceeds 60Hz or the current is below 90A, it may be classified as faulty. For example, if the data collected by a welding robot shows a vibration frequency of 62Hz and a current of 85A, the K-means clustering algorithm will classify it as a faulty state. This classification method quickly identifies equipment anomalies through cluster analysis and generates a set of equipment status classifications.

[0097] Step S330: Based on the device status classification set, the classification results are partitioned and stored through a distributed storage system, and integrity verification is performed to obtain structured status data.

[0098] The following formula is used to calculate the hash when partitioning classification results in a distributed storage system:

[0099] In formula (14), Indicates partition The storage hash value, Indicates partition The Middle The content of each data block This indicates the total number of data blocks in the partition. Indicates the first Redundancy of each data block and This represents the weighting coefficient.

[0100] Perform integrity verification on the stored structured state data:

[0101] In formula (15), Indicates time Integrity verification metrics This indicates the total number of data blocks participating in the verification. Indicates the first One data block, Indicates the first Cyclic redundancy check value of each data block Indicates the first data blocks Check value, and This represents the weighting coefficients for different verification methods.

[0102] Device status classification sets are partitioned and stored using a distributed storage system for efficient management and retrieval. Distributed storage systems, such as HBase or Cassandra, partition the classification results by timestamp. For example, classification data from 10:00:00Z to 11:00:00Z on August 8, 2025, is stored in the same partition, with each partition containing records of normal, warning, and fault states. Before storage, the distributed storage system performs integrity checks, calculating the hash value of the classification data and comparing it with the checksum sent by the sensors. If the check passes, the data is considered complete; if data is missing from a partition, the system requests the sensors to retransmit that portion of the data, thus ensuring data reliability and traceability.

[0103] Step S340: For the structured state data, a streaming processing framework is used to extract features from the structured state data to generate a first device state feature set.

[0104] The first device state feature set is described as follows: In formula (16), Represents the first device state feature set. Represents the candidate feature vector. This indicates the number of dimensions in the structured state data. Indicates the first Weighting coefficients for dimensional data Indicates the first Feature mapping function for dimensional data, Indicates the first Dimensional input structured state data, Indicates the feature extraction parameters. Represents the parameter space.

[0105] For structured state data, the streaming processing framework further extracts features to generate a first equipment state feature set. For example, the mean and variance of vibration frequency are extracted from normal structured state data as feature values; abnormal current fluctuation frequencies are extracted from faulty structured state data. Assume that a device's mean vibration frequency is 52Hz and variance is 1.5Hz² under normal conditions, while its current fluctuation frequency is 3 times per second under fault conditions. These feature values ​​form the equipment state feature set, supporting subsequent equipment health prediction and maintenance scheduling. The first equipment state feature set is used for machine learning model training to identify potential fault modes and optimize maintenance plans.

[0106] Preferably, in the 5G-based workshop digital visualization method provided in this embodiment, step S400 includes: Step S410: Obtain real-time operating data streams from the workshop equipment sensors of the physical equipment, parse the real-time operating data streams using streaming processing tools, extract equipment status parameters, and generate a second equipment status feature set.

[0107] The equipment status parameters are described as follows:

[0108] In formula (17), Indicates the first Each device status characteristic Indicates the size of the streaming processing window. Indicates the first A feature extraction function, Indicates time Structured state data, Indicates the time decay factor. Indicates the current moment. e Represents the natural constant.

[0109] The press's sensors collect pressure, temperature, and displacement data in real time, forming a data stream of approximately 800 bytes per second. This stream contains pressure values ​​of 200-250 kPa, temperature of 50-60°C, and displacement of 0.5-1.0 mm. Streaming tools such as Apache Spark Streaming parse these data streams, transforming the raw data into a structured set of equipment status parameters, such as average pressure, temperature fluctuation range, and displacement frequency. Specifically, a press's data stream might show a pressure of 230 kPa, a temperature of 55°C, and a displacement of 0.7 mm. After parsing, this data generates a set containing these parameters, facilitating subsequent spatiotemporal alignment.

[0110] Step S420: Based on the second device state feature set, a spatiotemporal calibration mechanism is used to align the virtual model and the physical device in terms of time and space dimensions to obtain the aligned mapping relationship.

[0111] The aligned mapping relationship is expressed as follows:

[0112] In formula (18), The function representing the aligned mapping relationship. This represents the weighting coefficient for the time dimension. This represents the spatial dimension weight coefficient. Represents the weight matrix along the time dimension. Represents the spatial dimension weight matrix. Represents the device state vector. This represents the state vector of the virtual model.

[0113] For equipment status parameters, a spatiotemporal calibration mechanism ensures that the virtual model and the physical equipment are aligned in both time and space. This mechanism uses timestamps as a reference and combines them with the equipment's position coordinates for alignment. For example, if the press sensor data includes a timestamp of 2025-08-08 14:00:00Z and position coordinates X:10, Y:20, the virtual model adjusts its state based on the same timestamp and coordinates, generating an aligned mapping relationship. During calibration, the system eliminates time discrepancies by comparing the time-series data of the physical equipment and the virtual model; and ensures spatial consistency through coordinate matching.

[0114] Step S430: For the aligned mapping relationship, calculate the mapping error E between the virtual model and the physical device state, where the mapping error E represents the deviation between the virtual model output and the physical device state. If the mapping error E is lower than a preset threshold, it is determined that the model needs to be updated and an update signal is generated.

[0115] The updated signal is described as follows: (19) In formula (19), Indicates an update signal. This represents the calculated mapping error value. This represents the preset error threshold. When the mapping error is lower than the threshold, an update signal of 1 is output; otherwise, 0 is output. This is used to determine whether the model needs to be updated.

[0116] Based on the aligned mapping relationship, the mapping error E between the virtual model and the physical equipment state is calculated. For example, if the virtual model predicts a pressure of 235 kPa, while the actual pressure is 230 kPa, the mapping error E is 5 kPa. In this embodiment, a preset threshold of 10 kPa is used. If the mapping error E is less than 10 kPa, it indicates that the virtual model prediction is accurate and no update is needed; if the mapping error E exceeds the threshold, such as 15 kPa, an update signal is generated. Specifically, if the mapping error E of a certain stamping press is 12 kPa, an update signal is triggered.

[0117] Step S440: Based on the update signal, adjust the virtual model parameters using database tools to obtain a real-time synchronized digital twin model.

[0118] The real-time synchronized digital twin model is described as follows:

[0119] In formula (20), This indicates the state of the digital twin model after real-time synchronization. This indicates the current state of the virtual model. Indicates the synchronous attenuation coefficient. Represents real system data, Represents virtual system data. This indicates an element-wise multiplication operation.

[0120] The update signals adjust the virtual model parameters using database tools such as MongoDB. For example, based on historical data analysis, the system adjusts the virtual model's pressure prediction parameter from 235 kPa to 230 kPa and the temperature parameter from 56°C to 55°C. The adjusted virtual model forms a real-time synchronized digital twin model.

[0121] Digital twin models maintain dynamic consistency with physical equipment through continuous updates. For example, during the operation of a stamping press, if the sensor data stream shows that the pressure suddenly rises to 260 kPa, the digital twin model adjusts its parameters based on the updated signal, and the predicted value is quickly corrected to 258 kPa, reducing the mapping error E to 2 kPa.

[0122] Furthermore, in the 5G-based workshop digital visualization method provided in this embodiment, step S500 includes: Step S510: Extract device collaboration information from the real-time synchronized digital twin model, parse the device collaboration information using a streaming processing tool, and generate a first feature set containing the interaction states of multiple devices.

[0123] The first feature set is expressed as follows:

[0124] In formula (21), Describes the first feature set. Indicates the number of device pairs participating in the interaction. Indicates the first Interaction strength coefficient of each device pair Indicates the first Frequency of interaction between devices Indicates the first The reciprocal of the response time of each device pair.

[0125] Equipment collaboration information is extracted from the digital twin model to capture the interaction between the stamping machine, welding robot, and conveyor belt. The stamping machine generates a data stream per second with a pressure of 230 kPa and a temperature of 55°C, the welding robot records a welding time of 3 seconds and a current of 150 A, and the conveyor belt records a speed of 0.8 m / s.

[0126] The equipment collaboration information is parsed by the streaming processing tool Apache Flink to generate a first feature set containing the interaction state, such as the collaboration time difference between the press and the welding robot being 2 seconds, and the material transfer frequency between the conveyor belt and the press being 0.5 times / second.

[0127] Step S520: The first feature set is processed using the isolated forest algorithm to calculate the anomaly score S of the interaction state of each device. The anomaly score S represents the degree to which the collaborative state of the devices deviates from the normal value. If the anomaly score S is higher than the preset threshold, a fault alarm signal is generated.

[0128] The anomaly score S for each device's interaction state is calculated using the following formula:

[0129] In formula (22), Indicates the first Anomalies in the interaction state of each device. Indicates sample Average path length in an isolated forest express The average path length normalization factor for constructing a binary search tree from samples is used. The closer the anomaly score is to 1, the more abnormal it is, and the closer it is to 0, the more normal it is.

[0130] sample Average path length in isolated forests The expression is as follows:

[0131] In formula (23), Indicates sample Average path length across all isolated trees This represents the total number of isolated trees in an isolated forest. Indicates sample In the The average path length from the root node to the leaf node in an isolated tree is the length of the path. The shorter the average path length, the more likely the sample is to be isolated.

[0132] The fault alarm signal is derived using the following formula: (twenty four) In formula (24), This indicates the generation status of the fault alarm signal. This represents the calculated abnormal score. This represents the preset anomaly detection threshold. When the anomaly score exceeds the threshold, [the system will detect anomalies]. An alarm is generated if the value is 1, otherwise... An integer equal to 0 will not generate an alarm.

[0133] The Isolation Forest algorithm is used to analyze the first feature set and calculate an anomaly score S to identify anomalies in equipment coordination. For example, under normal circumstances, the coordination time difference between the stamping machine and the welding robot should be 1-3 seconds. If the time difference reaches 5 seconds at a certain moment, the Isolation Forest algorithm calculates an S value of 0.85, which is higher than the preset threshold of 0.7, indicating an anomaly in coordination and generating a fault alarm signal. Preferably, the Isolation Forest constructs random trees to segment the data, quickly identifying data points that deviate from the normal pattern, making it suitable for real-time processing of high-dimensional data.

[0134] Step S530: Based on the fault alarm signal and combined with the production process data, integrate the fault alarm signal and production operation parameters through database tools to generate a second dataset containing alarm status and production data.

[0135] The second dataset was obtained using the following formula:

[0136] In formula (25), This represents the generated second dataset. This indicates the total number of fault alarm signals. Indicates the first One fault alarm signal data, This represents the corresponding production operation parameter data. This indicates a database connection operation.

[0137] The alarm status is described as follows:

[0138] In formula (26), Indicates an alarm status. This represents the preset fault threshold parameter. This indicates the operating parameter values ​​of the current production process. Indicates a time window. This represents the alarm judgment function.

[0139] Fault alarm signals are combined with production process data and integrated using database tools such as Cassandra. For example, a fault alarm signal indicates an abnormality in the coordination between the stamping machine and the welding robot, and production data shows that the current production cycle is 10 products per minute. Cassandra stores alarm status and production parameters in partitions by device ID (Identity Document) and timestamp, forming a second dataset. For example, it might contain data records such as alarm time 2025-08-08 14:05:00, abnormal equipment as a stamping machine, and production cycle of 10 pieces / minute, thus facilitating the tracing of the root cause of the problem.

[0140] Step S540: Process the second dataset using visualization tools, and use dashboard generation technology to draw a dynamic view of the production process to obtain real-time updated production status data.

[0141] The dynamic view is described as follows:

[0142] In formula (27), This represents the dynamic view data of the dashboard at time t. This indicates the total number of parameters monitored during the production process. Indicates the first The weighting coefficients of each parameter, Indicates the first The parameters at time... Visualized numerical values, Indicates the first The parameters at time... The state function is 1 when the parameters are normal and 0 when they are abnormal.

[0143] The second dataset is processed using the visualization tool Tableau to generate a dynamic dashboard view. This dynamic view displays the press pressure curve, welding robot current fluctuation graph, and conveyor belt speed trend graph, highlighting abnormal time points, such as the collaborative anomaly at 14:05:00. The dashboard also shows real-time changes in production cycle time, such as a drop to 8 pieces / minute due to an anomaly. Preferably, the dashboard supports user interaction, allowing users to zoom in on the abnormal time period and view detailed parameters. In this embodiment, the dynamic dashboard view intuitively reflects the production status, facilitating rapid response to anomalies by workshop managers.

[0144] Please see Figure 2This embodiment provides a 5G-based workshop digital visualization system for executing the aforementioned 5G-based workshop digital visualization method. The 5G-based workshop digital visualization system includes a compressed data stream acquisition module 10, a structured data set acquisition module 20, an equipment status feature set acquisition module 30, a digital twin model acquisition module 40, and a visualization presentation data acquisition module 50. The compressed data stream acquisition module 10 acquires collected data from workshop equipment sensors, performs dimensionality reduction processing on high-frequency data in the collected data to obtain a compressed data stream. The collected data includes production data, energy consumption data, and material flow information. The structured data set acquisition module 20 transmits the compressed data stream to a cloud processing platform via a high-speed, low-latency network, uses data fragmentation technology to partition and manage the data stream, and determines the integrity of data fragments; if complete, the fragments are stored. The data is stored in a distributed database to obtain a structured data set. The equipment status feature set acquisition module 30 uses a streaming processing framework to analyze the equipment status in real time, classifies equipment operating parameters using a clustering algorithm, and obtains an equipment status feature set. The digital twin model acquisition module 40 constructs a digital twin model based on the equipment status feature set, dynamically maps the virtual model to the physical equipment using a spatiotemporal calibration mechanism, determines whether the mapping error is below a preset threshold, and updates the virtual model if it is below the threshold, resulting in a real-time synchronized digital twin model. The visualization data acquisition module 50 extracts equipment collaboration information from the real-time synchronized digital twin model, identifies potential faults using an anomaly detection algorithm, determines whether the equipment collaboration status is normal, and generates a fault alarm if an anomaly is found, thus obtaining visualized data of the production process.

[0145] The 5G-based workshop digital visualization method and system provided in this embodiment, compared with existing technologies, addresses the business scenario problems of efficient data collection, real-time analysis, and fault early warning for workshop equipment. It preprocesses high-frequency production data, energy consumption data, and material data through edge computing and time-series compression algorithms to reduce dimensionality, transmits the data to the cloud via a high-speed, low-latency network, and employs data fragmentation and integrity verification to ensure reliable storage as a structured data set. Then, it uses a streaming processing framework and clustering algorithms to analyze equipment status characteristics in real time, constructs and dynamically updates a digital twin model, reduces mapping errors using a spatiotemporal calibration mechanism, and finally identifies potential faults and generates visual early warnings through anomaly detection algorithms. This embodiment achieves real-time monitoring of equipment collaborative status and accurate fault prediction, improving production efficiency and the level of intelligent equipment management.

[0146] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention. Clearly, those skilled in the art can make various alterations and modifications to the invention without departing from its spirit and scope. Thus, if these modifications and modifications of the invention fall within the scope of the claims and their equivalents, the invention is also intended to include these modifications and modifications.

Claims

1. A 5G-based digital visualization method for workshops, characterized in that, Includes the following steps: S100. Acquire data from workshop equipment sensors, perform dimensionality reduction processing on high-frequency data in the acquired data to obtain a compressed data stream. The acquired data includes production data, energy consumption data, and material flow information. S200 transmits the compressed data stream to the cloud processing platform via a high-speed, low-latency network, manages the data stream by partition, judges the integrity of the data fragments, and if the data fragments are complete, stores them in a distributed database to obtain a structured data set. S300. For the structured data set, a streaming processing framework is used to perform real-time analysis of the device status, and a clustering algorithm is used to classify the device operating parameters to obtain a device status feature set. S400. Based on the device status feature set, construct a digital twin model, dynamically map the virtual model and the physical device through a spatiotemporal calibration mechanism, determine whether the mapping error is lower than a preset threshold, and if it is lower than the threshold, update the virtual model to obtain a real-time synchronized digital twin model. S500 extracts equipment collaboration information from the real-time synchronized digital twin model, uses anomaly detection algorithms to identify potential faults, determines whether the equipment collaboration status is normal, and generates a fault alarm if an anomaly is found, thus obtaining visualized data of the production process.

2. The 5G-based workshop digital visualization method according to claim 1, characterized in that, Step S100 includes: S110. Obtain sensor data, energy consumption data, and material flow information from workshop equipment sensors, and perform standardized format conversion on the sensor data, energy consumption data, and material flow information to obtain a dataset in a unified format; The standardized dataset includes standardized sensor data, standardized energy consumption data, and standardized material flow information. The standardized sensor data is described as follows: in, This represents the standardized sensor data. This represents the raw data acquired from the sensors of the workshop equipment. This represents the mean of the original data. This represents the standard deviation of the original data; Energy consumption data is presented in a standardized format as follows: in, This represents the normalized energy consumption data. This indicates the currently collected energy consumption value. This represents the minimum value of energy consumption data. This represents the maximum value of the energy consumption data; The standardized format for material flow information is as follows: in, This represents material flow information in a standardized format. This represents material flow data. This represents the material flow time data. Indicates material location data. , , These represent the weighting coefficients for each dimension of the data; S120. Perform preprocessing operations on the unified format dataset through edge computing nodes to remove noise and redundant data in the unified format dataset and obtain the cleaned dataset. The cleaned dataset is represented as follows: in, This represents the cleaned dataset. Represents the original, uniformly formatted dataset. Represents the first in the dataset Data points, Represents the mean vector of the dataset. Indicates standard deviation, This represents the noise detection threshold coefficient; S130. For the cleaned dataset, perform dimensionality reduction on the high-frequency data. If the frequency of the cleaned dataset exceeds a preset threshold, perform data aggregation to obtain a compressed data stream. The compressed data stream is represented as follows: in, This represents the compressed data stream. This indicates the number of data blocks being aggregated. Indicates the first The weighting coefficients of each data block, Indicates the first One raw data block, Indicates the first The target dimensionality for dimensionality reduction of each data block; S140. Send the compressed data stream to the cloud storage system, determine whether there is data loss during the transmission of the compressed data stream, and if there is loss, trigger the retransmission mechanism to obtain the complete stored dataset. The following formula can be used to detect whether data loss occurs during the transmission of compressed data streams: in, This represents the probability of data loss during data transmission. Indicates the total number of data packets transmitted. Indicates the first The integrity verification function for each data packet, when receiving data. With sending data Returns 1 if they match, otherwise returns 0.

3. The 5G-based workshop digital visualization method according to claim 1, characterized in that, Step S200 includes: S210. The compressed data stream is transmitted through a high-speed, low-latency network, and the data stream is divided using data fragmentation technology to obtain a fragmented data stream. The uniform division of the data stream fragments is achieved using the following formula: in, Indicates the first The size of each data partition Indicates the total size of the data stream to be segmented. This indicates the preset total number of fragments. Indicates the fragment sequence number; S220. For the fragmented data stream, perform parallel processing operations on the cloud processing platform, and integrate the fragments using multi-node parallel computing to obtain the integrated data set. The integrated dataset is described as follows: in, This represents the integrated data set. This indicates the total number of data fragments. Indicates the first Data shards, Represents data conversion functions, Indicates the first The conversion parameters for each segment; S230. If the integrated data set passes the data integrity check, the integrated data set is stored in the distributed database to verify the consistency of each shard and obtain a structured data set. Data integrity verification is described as follows: in, Indicates data integrity verification metrics. Indicates the total number of data fields. Indicates the total number of data records. Indicates the first The field number The weighting coefficient of each record. This function indicates the consistency between the field value and the reference value, returning 1 when the data is complete and 0 when it is missing or incorrect. The consistency of each fragment is expressed as follows: in, This indicates the consistency verification results for each fragment. This represents the total number of shards in the distributed database. Indicates the first Each segment, Indicates the first The number of data items per shard Indicates the first The first segment One data item, This indicates that the consistency verification of the data item has passed the function. This represents the overall validation function for the data item.

4. The 5G-based workshop digital visualization method according to claim 1, characterized in that, Step S300 includes: S310. Receive real-time data streams from workshop equipment sensors, parse the data streams using a streaming processing framework, extract operating parameters, and obtain a set of equipment operating parameters. The extracted runtime parameters are described as follows: in, Indicates the first The extracted runtime parameter values, This indicates the total number of data stream elements participating in the parsing. Indicates the first The weight coefficient of each data element, Indicates the first The data element is the first... The contribution value of each parameter; The set of equipment operating parameters is described as follows: in, This represents the final set of equipment operating parameters. This indicates the total number of extracted runtime parameters. Indicates the first One running parameter, and These represent the minimum and maximum threshold ranges of the parameters, respectively. Set operations ensure that all parameters are within the valid range. S320. For the set of equipment operating parameters, the K-means clustering algorithm is used to classify the operating parameters. If the parameter value meets the preset threshold range, the classification result is determined, and the equipment status classification set is obtained. The following formula is used to determine whether the parameter values ​​after clustering meet the preset threshold range. If the condition is met, the classification result is considered valid: in, This represents the minimum value of the preset threshold. This represents the maximum value of the preset threshold. Indicates the first The values ​​of each running parameter; S330. Based on the device status classification set, the classification results are partitioned and stored through a distributed storage system, and integrity verification is performed to obtain structured status data. The following formula is used to calculate the hash when partitioning classification results in a distributed storage system: in, Indicates partition The storage hash value, Indicates partition The Middle The content of each data block This indicates the total number of data blocks in the partition. Indicates the first Redundancy of each data block and Indicates the weighting coefficient; Perform integrity verification on the stored structured state data: in, Indicates time Integrity verification metrics This indicates the total number of data blocks participating in the verification. Indicates the first One data block, Indicates the first Cyclic redundancy check value of each data block Indicates the first data blocks Check value, and Represents the weighting coefficients for different verification methods; S340. For the structured state data, a streaming processing framework is used to extract features from the structured state data to generate a first device state feature set. The first device state feature set is described as follows: in, Represents the first device state feature set. Represents the candidate feature vector. This indicates the number of dimensions in the structured state data. Indicates the first Weighting coefficients for dimensional data Indicates the first Feature mapping function for dimensional data, Indicates the first Dimensional input structured state data, Indicates the feature extraction parameters. Represents the parameter space.

5. The 5G-based workshop digital visualization method according to claim 1, characterized in that, Step S400 includes: S410. Obtain real-time operating data streams from the workshop equipment sensors of the physical equipment, parse the real-time operating data streams using a streaming processing tool, extract equipment status parameters, and generate a second equipment status feature set. S420. Based on the second device state feature set, a spatiotemporal calibration mechanism is used to align the virtual model and the physical device in terms of time and space dimensions to obtain the aligned mapping relationship. S430. For the aligned mapping relationship, calculate the mapping error E between the virtual model and the physical device state, where the mapping error E represents the deviation between the virtual model output and the physical device state. If the mapping error E is lower than the preset threshold, it is determined that the model needs to be updated and an update signal is generated. S440. Based on the update signal, adjust the virtual model parameters using database tools to obtain a real-time synchronized digital twin model.

6. The 5G-based workshop digital visualization method according to claim 5, characterized in that, In step S410, the device status parameters are described as follows: in, Indicates the first Each device status characteristic Indicates the size of the streaming processing window. Indicates the first Feature extraction functions, Indicates time Structured state data, Indicates the time decay factor. Indicates the current moment. e Represents the natural constant.

7. The 5G-based workshop digital visualization method according to claim 6, characterized in that, In step S420, the aligned mapping relationship is expressed as follows: in, The function representing the aligned mapping relationship. This represents the weighting coefficient for the time dimension. This represents the spatial dimension weight coefficient. Represents the weight matrix along the time dimension. Represents the spatial dimension weight matrix. Represents the device state vector. This represents the state vector of the virtual model.

8. The 5G-based workshop digital visualization method according to claim 7, characterized in that, In step S430, the update signal is described as follows: in, Indicates an update signal. This represents the calculated mapping error value. This represents the preset error threshold. When the mapping error is lower than the threshold, an update signal of 1 is output; otherwise, 0 is output. This is used to determine whether the model needs to be updated. In step S440, the real-time synchronized digital twin model is described as follows: in, This indicates the state of the digital twin model after real-time synchronization. This indicates the current state of the virtual model. Indicates the synchronous attenuation coefficient. Represents real system data, Represents virtual system data. This indicates an element-wise multiplication operation.

9. The 5G-based workshop digital visualization method according to claim 1, characterized in that, Step S500 includes: S510. Extract device collaboration information from the real-time synchronized digital twin model, parse the device collaboration information using a streaming processing tool, and generate a first feature set containing the interaction states of multiple devices. S520. The first feature set is processed using the isolated forest algorithm to calculate the abnormal score S of the interaction state of each device. The abnormal score S represents the degree to which the collaborative state of the devices deviates from the normal value. If the abnormal score S is higher than the preset threshold, a fault alarm signal is generated. S530. Based on the fault alarm signal and combined with the production process data, the fault alarm signal and production operation parameters are integrated through a database tool to generate a second dataset containing alarm status and production data. S540. The second dataset is processed using visualization tools, and a dynamic view of the production process is drawn using dashboard generation technology to obtain real-time updated production status data.

10. A 5G-based workshop digital visualization system, used to execute the 5G-based workshop digital visualization method as described in any one of claims 1 to 9, characterized in that, include: The compressed data stream acquisition module (10) is used to acquire collected data from workshop equipment sensors, perform dimensionality reduction processing on high-frequency data in the collected data, and obtain a compressed data stream. The collected data includes production data, energy consumption data and material flow information. The structured data set acquisition module (20) is used to transmit the compressed data stream to the cloud processing platform through a high-speed, low-latency network, and to manage the data stream by partitioning it using data sharding technology. It judges the integrity of the data shards and stores them in a distributed database if they are complete, thus obtaining a structured data set. The device status feature set acquisition module (30) is used to perform real-time analysis of the device status using a streaming processing framework for the structured data set, and to classify the device operating parameters through a clustering algorithm to obtain the device status feature set; The digital twin model acquisition module (40) is used to construct a digital twin model based on the device status feature set, dynamically map the virtual model and the physical device through a spatiotemporal calibration mechanism, determine whether the mapping error is lower than a preset threshold, and update the virtual model if it is lower than the threshold to obtain a real-time synchronized digital twin model. The visualization presentation data acquisition module (50) is used to extract equipment collaboration information from the real-time synchronized digital twin model, use anomaly detection algorithm to identify potential faults, determine whether the equipment collaboration status is normal, generate a fault alarm if abnormal, and obtain visualization presentation data of the production process.

Images