Cross-industry big data sharing method and system

Abstract

The invention discloses a cross-industry big data sharing method and system, and mainly relates to the technical field of data sharing. According to the cross-industry big data sharing method disclosed by the invention, cross-industry data sharing is realized through a dynamic collaboration mechanism of four logic levels: (1) in a data acquisition stage, differential dynamic data is captured through a streaming fragmentation technology driven by industry characteristics; (2) constructing an intelligent sensing four-state nested structure in a code generation stage; (3) constructing an industry knowledge graph dynamic inference engine; and (4) executing environment-adaptive multi-element response in a distribution stage. According to the invention, by constructing a cooperation mechanism of a dynamic four-state coding system and an industry knowledge graph, the real-time limitation of static coding is broken through, and dynamic sharing, multi-level security verification and intelligent risk pre-judgment of multi-format data are realized.

Description

Technical Field

[0001] The present invention mainly relates to the technical field of data sharing, and specifically provides a cross-industry big data sharing method and system. Background Art

[0002] Cross-industry data sharing in the industrial Internet of Things environment has long faced multiple challenges of dynamics, security, and intelligent decision-making. Existing technologies mostly rely on static coding systems (such as the GS1 standard), whose pre-set fields are fixed after generation and are difficult to adapt to the real-time data update requirements of highly dynamic scenarios such as mechanical vibration and food cold chain. Taking the quality traceability scenario of a certain automotive part as an example, traditional two-dimensional codes only record the inspection results at the time of factory shipment and cannot reflect the temperature and humidity changes during the logistics process and the stress data at the assembly site, resulting in a delay of more than 3 days in tracing defective products. In the machinery industry with high-frequency data sampling (>500Hz), due to the lack of a dynamic hierarchical architecture in existing coding schemes, the sampling frequency has to be reduced to avoid communication congestion, resulting in a loss rate of more than 37% of sudden failure characteristics.

[0003] In addition, the differentiated security requirements across industries further intensify the technical contradictions: the clothing industry requires physical feature binding with millimeter-level precision to combat fabric batch forgery, while the food industry requires continuous signature verification of cold chain logistics trajectories. A unified security strategy is difficult to balance efficiency and protection intensity. A certain fresh food platform once had a positioning error of more than 8 hours in tampering incidents due to the adoption of single-level signatures.

[0004] The rigid architecture of the current industry knowledge reasoning system also restricts the release of data value. Traditional solutions use static rule engines to process data in various industries and fail to dynamically adapt the industry characteristics contained in the coding to the reasoning model. When the abnormal vibration spectrum of mechanical equipment triggers the microbial reproduction rule in the food industry, the false alarm rate is as high as 64%, seriously hindering cross-domain collaboration.

[0005] More severely, the dual constraints formed by the computing power limitation of edge devices and network fluctuations force the system to make a difficult balance among data update frequency, signature verification overhead, and risk response speed: low-latency dynamic coding requires sacrificing encryption intensity, while strong security verification leads to a sharp increase in the processing delay of high-frequency data. Existing solutions are difficult to break through this "efficiency-security-decision" ternary paradox, which has become a key bottleneck restricting cross-enterprise collaboration in Industry 4.0. Summary of the Invention

[0006] In view of the above problems, the present invention proposes a dynamic adaptive cross-industry big data sharing method and system, which break through the real-time limitation of static coding by constructing a collaborative mechanism of a dynamic four-state coding system and an industry knowledge graph, and realize dynamic sharing of multi-format data, multi-level security verification, and intelligent risk prediction.

[0007] To solve the above problems, the present invention proposes an innovative cross-industry big data sharing method and system, which realizes the nested high-frequency data stream while maintaining the coding stability by constructing a two-way driving mechanism of dynamic four-state coding and industry knowledge graph, and establishes an extensible multi-level security verification system.

[0008] In the first aspect, the present invention provides a cross-industry big data sharing method, which is realized through four major links: dynamic processing of industry feature perception, multi-level coding generation, intelligent reasoning driven by knowledge graph, and response distribution with environmental adaptability. In specific implementation, first, the acquisition strategy is dynamically set according to the data characteristics of different industries: for the vibration signal capture requirement of mechanical equipment, the system performs a fast Fourier transform on the original signal through the FPGA chip configured on the industrial control terminal, extracts the vibration fundamental frequency and harmonic component features with a 50ms time window. Compared with the traditional second-level window scheme, the spectral resolution is increased by 20 times; for the generation of unique fabric identification in the clothing industry, a laser scanning device with a wavelength of 850nm is used to obtain the three-dimensional topological information of the fabric surface, and the first 32 low-frequency coefficients are extracted through block discrete cosine transform to generate a feature vector, and a 256-bit unique identification code is generated through the SHA3-512 hash algorithm. Experimental tests show that the discrimination degree of this coding for the same type of fabric batches is as high as 99.99%.

[0009] A dual-threshold trigger mechanism is deployed during the acquisition process. For example, in the food cold chain scenario, the basic threshold Δ1 is set to ±0.5°C / minute, and data synchronization is only triggered when the temperature change rate exceeds Δ2 = 2.5°C / minute. Combining with the local cache mechanism of the edge node, the communication energy consumption in the normal state is reduced by 42%.

[0010] In the coding generation stage, a four-state nested structure is constructed to adapt to diverse scenarios. The static identification layer integrates industry-specific physical features (such as clothing laser hash, mechanical component wear index), and uses variable splicing rules to generate an irreversible base code; the dynamic update layer monitors data fluctuations in real time through a hardware-level dual-threshold comparator. When the mechanical vibration energy change rate exceeds the set threshold, a high-speed transmission channel is triggered and a TSN timestamp mark is inserted; the encryption authentication layer automatically selects the signature mode according to the data criticality - when it is detected that the number of dynamic data updates per second exceeds 50 times, it automatically switches from the traditional SM2 dual-signature to the BLS-12-381 aggregate signature mode. By constructing a Merkle tree from the packet hashes within 10 consecutive seconds and generating a single aggregate signature, the signature verification energy consumption is reduced by 72%.

[0011] The service extension layer pre-sets a snapshot of the industry knowledge graph sub-structure. For example, for mechanical equipment, the fault feature association rules of key components are pre-stored. When the network delay exceeds 200ms, the edge node directly analyzes the abnormal feature path based on the snapshot to achieve local fault diagnosis at the 150ms level.

[0012] Deploy multiple adaptive optimization strategies in the data distribution link: In the network bandwidth detection module, predict the communication quality in the next 5 seconds based on the ARIMA time series model. When the predicted value is lower than 100 Kbps, automatically switch the response data from the high-precision image mode to a 1KB compressed text summary, and activate the geofence blurring processing logic. For example, convert "XX Manufacturing Base in Zhejiang" to the industry code "CN-ZJ-03"; if the user permission verification is a regulatory terminal, open the complete feature hash query interface through the service extension layer and combine the presigned mechanism to achieve data traceability, shortening the traceability time of tampering events from 8 hours in the traditional solution to within 10 minutes.

[0013] Corresponding to the method flow, the system of the present invention consists of four major hardware clusters: an industry protocol adaptation module, a dynamic coding controller, an edge inference node, and an adaptive rendering gateway. The industry protocol adaptation module adopts a pluggable design. The mechanical industry subunit integrates a signal conditioning circuit and a CAN bus transparent transmission module, and is built-in with a second-order anti-aliasing filter to eliminate 50Hz power frequency interference, ensuring that the vibration spectrum sampling signal-to-noise ratio ≥ 65dB; the clothing industry subunit is equipped with a high-precision CCD sensor and a laser drive circuit, supporting surface profile scanning at the 0.1μm level, and controlling the feature extraction delay within 5ms.

[0014] The core of the dynamic coding controller is a dual-threshold comparison chipset, which includes two channels for analog circuit and digital processing - for the analog signal processing channel of the food industry, a differential amplifier and an integration circuit are set up to achieve millisecond-level calculation of the temperature change rate; the digital processing channel of the mechanical industry is configured with an FFT hardware accelerator, which can parallel process the frequency domain analysis of 8 vibration signals.

[0015] The edge inference node includes three major functional components: (1) The atlas snapshot cache pool adopts a hierarchical storage architecture and maintains a cache hit rate of more than 85% through the LRU-K algorithm; (2) The rule engine loading unit is built-in with an industry rule mapping table. For example, when scanning the mechanical industry code, it automatically loads the Drools rule library containing the bearing wear failure model; (3) The local signature verification module integrates a hardware accelerator for BLS signatures, which can directly call GPU resources through the PCIe bus to complete the aggregated signature verification, shortening the batch processing time of 500 data packets from 5.2 seconds to 0.3 seconds.

[0016] The adaptive rendering gateway deploys a dual-core heterogeneous processor. The NPU core runs the bandwidth prediction model and calculates the optimal coding strategy in real time; the CPU core executes JWT token parsing and geofence matching. When the geographical location triggers the blurring rule, it calls the preset obfuscation algorithm to convert the precise address into a generalization identifier such as "Southwest Region - Supplier 03", reducing the risk of commercial sensitive information leakage by 89%.

[0017] Compared with the prior art, the beneficial effects of the present invention are: (1) Dynamic acquisition bound to the industry: By accelerating the vibration fundamental frequency calculation through FPGA, the data processing time in the 500Hz sampling scenario is reduced from 22ms to 0.8ms. Combining with a dual-threshold trigger to build a multi-level response mechanism, the transmission rate of mechanical fault features without loss reaches 98.7%.

[0018] (2) Four-state nested coding structure: The combination of laser hash embedding in the static layer and TSN timestamp injection in the dynamic layer realizes a time positioning error of tampering events ≤ 10ms; in the scenario of high-frequency updates at 1000 times per second, the overall signature energy consumption of the system is reduced to 28% of the traditional scheme.

[0019] (3) Knowledge graph dynamic adaptation: The cooperation of the subgraph snapshot pre-storage mechanism and the edge rule engine keeps the abnormal diagnosis accuracy above 91% in the offline state, which is 4 times higher than the traditional cloud scheme.

[0020] (4) Environment adaptive rendering: The combined strategy of the ARIMA bandwidth prediction model and geographical fence fuzzification shortens the first-screen loading time of mobile users by 82%, while meeting the EU GDPR privacy protection standard.

[0021] Through the deep cooperation of the method process and the hardware system, this solution achieves three breakthrough indicators in the industrial Internet of Things scenario: end-to-end processing delay ≤ 20ms (traditional scheme ≥ 220ms), cross-industry rule misjudgment rate ≤ 1.5% (original benchmark ≥ 12%), and high-frequency data integrity rate ≥ 99.9%. Taking the implementation data of a certain automotive parts manufacturer as an example, the product recall events caused by data lag are reduced by 93% after the system is launched, verifying the industrial practical value of the technical solution.

[0022] The present invention will be explained and described in detail below in conjunction with the drawings and specific embodiments. Description of the Drawings

[0023] Figure 1 Shows the schematic flow chart of the cross-industry big data sharing method of the present invention; Figure 2 Shows the system architecture diagram of the cross-industry big data sharing of the present invention. Specific Embodiments

[0024] To facilitate the understanding of the present invention, the present invention will be described more comprehensively below with reference to the relevant drawings. Several embodiments of the present invention are given in the drawings, but the present invention can be implemented in different forms and is not limited to the embodiments described in the text. On the contrary, providing these embodiments is to make the disclosure of the present invention more thorough and comprehensive.

[0025] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the technical field to which this invention belongs. The terms used in the description of the present invention herein are for the purpose of describing specific embodiments only and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more of the related listed items.

[0026] In view of the problems in the background art, the present invention proposes a dynamic adaptive cross-industry big data sharing method and system, which realizes real-time nesting of high-frequency data streams, adjustable security policies, and accurate risk prediction through constructing a collaborative mechanism of industry-aware coding system and edge intelligent reasoning. The core of the cross-industry big data sharing method of the present invention lies in: dynamically adjusting the acquisition and processing process according to the data characteristics of the target industry, generating a four-state code integrating static identifiers, real-time updated fields, multi-level security verification, and service extension interfaces, and embedding industry rule identifiers in the coding layer to drive the self-adaptation of the knowledge graph reasoning model, and finally outputting differentiated multi-modal responses in combination with the network environment and user permissions.

[0027] In specific implementation, first, a programmable hardware cluster is used to dynamically configure the data acquisition window. The food industry uses a sliding window to extract the temperature change rate feature, the machinery industry captures the vibration fundamental frequency by micro-batch processing, and the clothing industry introduces a laser scanning device to generate the topological fingerprint hash of the fabric surface, solving the problem of insufficient adaptability of the traditional solution for feature extraction.

[0028] In the coding generation stage, a dual-threshold trigger mechanism is innovatively designed. When the sensor data fluctuation is lower than the set range, the network transmission is not triggered, and the data is only cached in the local edge node. When it exceeds the emergency threshold, a jump synchronization channel is started, and at the same time, the encryption policy is dynamically switched - in the high-frequency scenario, BLS aggregate signature is used to bind and verify multiple data, and the communication load is reduced by more than 65%.

[0029] To overcome the inference delay problem caused by network jitter, the system pre-stores the binary snapshot of the key entity subgraph in the coding service extension layer, enabling the edge node to perform local risk path derivation in the offline state. Combining with the Drools rule set dynamically loaded by the industry identifier, the accuracy of mechanical fault identification is ensured to be increased to 94.2%. In the distribution stage, a network quality perception model is constructed. In a weak network environment, the response data is automatically compressed to within 1KB and the key index text is preferentially returned, establishing a secure transmission baseline while ensuring the user experience.

[0030] In a first aspect, the present invention provides a cross - industry big data sharing method, which realizes cross - industry data sharing through a dynamic collaboration mechanism at four logical levels: (1) In the data collection stage, differential dynamic data is captured through a streaming sharding technology driven by industry characteristics; (2) In the encoding generation stage, an intelligent perception four - state nested structure is constructed; (3) An industry knowledge graph dynamic inference engine is constructed; (4) In the distribution stage, a multi - response adaptive to the environment is executed. Specifically: S1. Industry - aware data collection: Select a feature extraction algorithm and a collection window based on the target industry type. The food industry uses a 2 - second sliding window to match the temperature change rate calculation period, the machinery industry uses a 50 - ms micro - batch processing window to capture the fundamental frequency of the vibration spectrum, and the clothing industry introduces a fabric surface light - induced fingerprint collection device to generate laser topology hash features. Among them, the generation of fabric light - induced fingerprint hash includes: Use an 850 - nm laser beam to scan the fabric surface to generate a 1024×1024 - pixel reflected grayscale matrix; Perform a block discrete cosine transform on the matrix, and take the first 32 coefficients to splice into a feature vector; Input the feature vector into the SHA3 - 512 algorithm to generate a 256 - bit laser hash identifier, which is used as a mandatory field in the static identification layer.

[0031] Specifically, for the vibration data of the machinery industry, the fundamental frequency of vibration is calculated in real - time through an FPGA - accelerated fast Fourier transform chip, and the spectrum energy data is packed into micro - batches every 50 ms; for the surface features of clothing fabrics, a laser scanning device generates a surface topology matrix at a wavelength of 850 nm, extracts a compressed feature vector through a block DCT transform and generates a SHA3 - 512 hash value, so that the hash uniqueness of the same batch of fabrics reaches 99.99%. During the collection process, the data change rate is dynamically monitored, and whether to trigger subsequent processing is determined by industry - specific thresholds.

[0032] S2. Dynamic four - state encoding generation: Generate a four - state encoding including a static identification layer, a dynamic update layer, an encryption authentication layer, and a service extension layer. Specifically, in the present invention, the four - state encoding adopts a distributed storage structure, where the static identification layer is stored in the central database and mapped to the blockchain ledger; the dynamic update layer is implemented through a non - volatile cache module of the edge node; the encryption authentication layer and the service extension layer are hosted by the cloud security container and provide API access interfaces.

[0033] The static identification layer is stored in the central database and mapped to the blockchain ledger, integrating industry - customized factors and blockchain unique identifiers. The clothing industry embeds a laser hash, the machinery industry embeds a wear index based on historical load, and an irreversible identifier is generated by assembling industry codes.

[0034] The dynamic update layer is stored in the non-volatile cache module of the edge node, and a dual-threshold trigger incremental update mechanism is adopted. When the change amount of sensor data is lower than the base threshold Δ1, it is only locally cached. When the change rate exceeds the emergency threshold Δ2, jump synchronization is started and the signature strength is degraded.

[0035] Preferably, before the dual-threshold comparison of the present invention, sensor baseline calibration needs to be performed. The original data stream is collected for 5 seconds for sliding window filtering; the standard deviation σ of each window is calculated. When σ > the set threshold, the automatic gain control (AGC) circuit is triggered, and the calibrated data can be used for the change rate calculation.

[0036] The degradation of the signature strength includes the following situations: when the number of updates per second of the dynamic layer > 50, the BLS-12-381 aggregated signature is used to replace the SM2+RSA dual-level signature. The generation logic of the BLS-12-381 aggregated signature is as follows: (a)The edge node generates a Merkle tree for the dynamic data sequence within the Δt period and inputs the root hash value into the signature algorithm; (b)After the signature verification end verifies the root hash consistency through the preset public key, the corresponding dynamic data subset is extracted for integrity verification.

[0037] Specifically, a dual-threshold comparator is introduced into the dynamic update layer, and the implementation method of dual-threshold trigger is as follows: The base threshold Δ1 is set according to industry sensitivity. For example, in the food industry, Δ1 = ±0.5℃ / minute; in the machinery industry, Δ1 = 5% vibration energy change rate; in the clothing industry, Δ1 = ±2% fabric stretch rate).

[0038] The emergency threshold Δ2 = 5×Δ1, and when jump synchronization is triggered, the dynamic layer data packet is attached with a red priority mark, and the encryption authentication layer inserts a time-sensitive network timestamp in the signature field.

[0039] When the data fluctuation exceeds Δ1 but does not reach the emergency threshold Δ2 = 5×Δ1, only local logging is performed; when it exceeds Δ2, a time-sensitive network (TSN) chip is used to inject a high-priority mark, and a packet jump direct transmission channel is started.

[0040] The encryption authentication layer deploys a policy selector to switch between dual-level signatures and aggregated signatures according to the data update frequency. At the central node verification end, when the buffer queue depth > 100 packets or the update frequency > 50Hz, the BLS-12-381 aggregated signature is activated, and the Merkle tree root hash of the dynamic data packets within 10 seconds is singly signed, reducing the signature verification overhead by 72%.

[0041] Preferably, add obfuscated metadata tags to the encryption authentication layer. When a geofence is triggered, the service extension layer generates an additional signature field σ' = Sig(Ψ || original hash) containing the obfuscation factor Ψ; the verification end needs to verify both the legality of the obfuscation factor and the integrity of the original data.

[0042] S3. Collaborative knowledge graph construction: Dynamically load the corresponding inference model based on the industry rule identifiers embedded in the static layer. When the network latency exceeds the standard, dynamically load the micrograph snapshot of the service extension layer at the edge node, and perform offline data verification and local inference.

[0043] The method for generating the above micrograph snapshot is as follows: (1) The central graph engine regularly extracts key entity subgraphs according to industry rules, including the fabric batch - supplier association chain in the clothing industry and the equipment serial number - replacement record chain in the machinery industry; (2) Compress the subgraph structure into a binary format and pre - embed it into the four - state encoding of the corresponding product through the service extension layer. Specifically, in the present invention, the compression process uses a combined algorithm of prefix encoding + Huffman encoding: First, extract the attribute prefixes of the subgraph entity relationships to form a dictionary; then perform Huffman compression on the repeated paths, and control the compression ratio within 30% of the original data.

[0044] (3) When the edge device detects that the network latency > 200ms, call the local parsing engine to execute a SPARQL query based on the snapshot.

[0045] Specifically, the central engine regularly extracts the graph sub - structure (such as the component replacement record chain of mechanical equipment) and compresses it into a binary snapshot for pre - storage in the encoding service extension layer. At the edge node, the parsing module loads the corresponding Drools rule library through the industry identifier (such as "M - 05" to identify the mechanical bearing detection rule). When it detects that the network latency > 200ms, directly perform inference based on the local snapshot to ensure that the response time ≤ 150ms.

[0046] S4. Adaptive data service distribution: Dynamically adjust the response data granularity according to user permissions and network environment, and return a combination of static layer thumbnail and dynamic index summary text in a weak network environment. Specifically, use an adaptive rendering gateway with an ARIMA prediction model built - in. When the predicted bandwidth < 100Kbps, preferentially return a combination of static - encoded thumbnail and summary text, and trigger the geofence obfuscation module to convert sensitive addresses into geographical region codes. If the user holds a regulatory permission token, the service extension layer opens the original laser hash and full vibration spectrum query interfaces.

[0047] The cross - industry big data sharing system of the present invention is composed of four major hardware clusters. Each module is used to implement the above - mentioned cross - industry big data sharing method at the physical layer. The system includes: Industry protocol adaptation module: Integrates an FPGA-accelerated food temperature change rate calculation unit, a mechanical vibration spectrum analysis DSP chip, and a clothing laser fingerprint scanning module. Among them: The mechanical subunit integrates an acceleration sensor signal conditioning circuit and a CAN bus transparent transmission module. The signal conditioning circuit is built-in with a 50Hz power frequency noise filter to ensure that the vibration spectrum sampling signal-to-noise ratio ≥ 60dB; The clothing subunit is equipped with a laser drive circuit and a high-precision CCD sensor, supporting 0.1μm-level surface topology scanning, and the feature extraction time ≤ 5ms / time; The food subunit is provided with 16 PT100 temperature sensor interfaces, and each channel is configured with an independent anti-aliasing filter, and the temperature change rate calculation error ≤ 0.1℃ / min.

[0048] Dynamic coding controller: Built-in with a dual-threshold comparator, a signature policy selection circuit, and a TSN timestamp injection module.

[0049] The dual-threshold comparator module adopts a hybrid circuit design: The analog circuit part processes the temperature change rate differential signal in the food industry, and the digital circuit part runs the FFT algorithm in the mechanical industry, and the determination delay ≤ 0.5ms.

[0050] The signature policy selector monitors the CPU load and queue depth of the edge node in real time. When the CPU utilization rate > 85%, it automatically switches to the aggregated signature mode and completes the BLS signature generation through a hardware accelerator.

[0051] Furthermore, in the present invention, a aggregated signature parsing device is deployed at the signature verification end, including: Hash tree reconstruction unit: Reconstructs a Merkle tree according to the received packet timestamp sequence; Delay tolerance verification module; Allows the block hash value verification to be completed within Δt ± 10% time.

[0052] The service extension layer stores the binary data of the industry map snapshot and provides an API interface through a cloud security container.

[0053] Edge inference module: Equipped with a map snapshot cache pool and a lightweight rule engine, supporting risk path derivation in offline mode. Among them: The cache pool stores the binary data of the industry map snapshot, and adopts the LRU-K replacement strategy to maintain a cache hit rate of more than 80%.

[0054] The rule engine loads the bearing fault mode feature library dedicated to the mechanical industry, matches the vibration spectrum features through a sliding time window, and the confidence calculation error ≤ 3%.

[0055] Among them, the operation logic of the edge inference module is: (a)After receiving the dynamic layer data, preferentially query the local cache snapshot. If not hit, initiate a synchronization request to the central graph engine; (b)The local rule engine loads the Drools rule subset corresponding to the industry identifier. The mechanical industry rule library includes a bearing fault modal feature library; (c)The inference result is attached with a credibility mark. When the credibility < 80%, request secondary verification from the central server.

[0056] Adaptive rendering gateway: Built-in network bandwidth prediction model and geofence fuzzification module, where: The network bandwidth prediction model establishes an ARIMA time series prediction based on historical transmission delays, predicts the bandwidth fluctuation trend in the next 5 seconds, and dynamically selects the JSON / ProtoBuf encoding format; The fuzzification processing module maintains a geofence - fuzzy rule mapping table. When the requested IP exceeds the preset regional range, convert the production address to the industry code + geographical region code. That is, when the IP address of the request does not belong to the preset list, trigger the "supplier code + GEO6" conversion logic. For example, map "XX Factory in Jiangsu" to "CN_EAST_F03".

[0057] The industry - specific dynamic acquisition and encoding mechanism of the present invention significantly optimizes the transmission efficiency: The micro - batch processing of the mechanical vibration fundamental frequency reduces the data delay from 220 ms to 18 ms, and the window matching error is reduced by 92%. The dual - threshold trigger strategy effectively balances energy consumption and data integrity. In the food cold chain scenario, the communication energy consumption for regular updates is reduced by 38%, while the recognition accuracy for emergency events is increased to 99.7%. The dynamic switching scheme of the security policy breaks through the computing power restriction. The BLS aggregated signature saves 72% of the signature energy consumption in the scenario of thousands of updates per second in the mechanical industry, and the time traceability accuracy of tampering operations reaches ±10 ms through timestamp injection.

[0058] The synergistic effect of edge inference snapshot and rule dynamic loading increases the response speed in a weak network environment by 4 times, and the misjudgment rate of clothing shrinkage batch identification is reduced from 8.9% to 1.3%. The geofence - driven fuzzification distribution strategy takes into account privacy and regulatory requirements. The fuzzification processing of the production address reduces the commercial leakage risk by 90%, while the complete feature hash ensures the non - repudiation of regulatory traceability. Multi - dimensional actual measurements show that the proposed solution improves the end - to - end processing efficiency by more than 300% in cross - industry scenarios, and reduces the security verification resource consumption by 68%, providing a reliable technical foundation for the deep collaboration of industrial Internet of Things.

[0059] References in the specification to "one embodiment" or "an embodiment" mean that the particular features, structures, or characteristics described in connection with the embodiment are included in at least one exemplary embodiment or technique disclosed in accordance with the present application. The appearances of the phrase "in one embodiment" in various places in the specification are not necessarily all referring to the same embodiment.

[0060] The present application disclosure also relates to an apparatus for performing operations in a text. The apparatus may be specifically constructed for the required purpose or it may comprise a general-purpose computer selectively activated or reconfigured by a computer program stored in the computer. Such a computer program may be stored in a computer-readable medium, such as, but not limited to, any type of disk, including floppy disks, optical disks, CD-ROMs, magneto-optical disks, read-only memory (ROM), random access memory (RAM), EPROM, EEPROM, magnetic or optical cards, application specific integrated circuit (ASIC) or any type of medium suitable for storing electronic instructions, and each may be coupled to a computer system bus. In addition, the computers mentioned in the specification may include a single processor or may be an architecture involving multiple processors for increased computing power.

[0061] The processes and displays presented herein inherently do not involve any particular computer or other device. Various general-purpose systems may also be used with programs in accordance with the teachings herein, or it may prove convenient to construct more specialized devices to perform one or more method steps. The structures for various such systems are discussed in the following description. Additionally, any specific programming language sufficient to implement the techniques and embodiments of the present application disclosure may be used. Various programming languages may be used to implement the present disclosure, as discussed herein.

[0062] In addition, the language used in this specification has been principally selected for readability and instructional purposes and may not have been selected to depict or limit the disclosed subject matter. Accordingly, the present application disclosure is intended to illustrate rather than limit the scope of the concepts discussed herein.

Claims

1. A cross-industry big data sharing method, characterized in that, It includes the following steps: S1. Industry-aware data collection: Select feature extraction algorithms and collection windows based on the target industry type; S2. Generation of dynamic four-state encoding: Generate four-state encoding including a static identification layer, a dynamic update layer, an encryption authentication layer, and a service extension layer, where: The static identification layer is stored in the central database and mapped to the blockchain ledger, integrating industry customization factors and blockchain unique identifiers; The dynamic update layer adopts an incremental update mechanism triggered by double thresholds: When the change amount of sensor data is lower than the basic threshold Δ1, it is only locally cached. When the change rate exceeds the emergency threshold Δ2, jump synchronization is started and the signature strength is degraded; The encryption authentication layer deploys a signature policy selector to switch between two-level signatures or aggregated signatures according to the data update frequency; The service extension layer stores industry map snapshots and provides localized business interfaces; S3. Collaborative knowledge graph construction: Dynamically load the corresponding inference model based on the industry rule identifiers embedded in the static layer. When the network latency exceeds the standard, the miniature map snapshot of the service extension layer is dynamically loaded at the edge node to perform offline data verification and local inference; S4. Adaptive data service distribution: Dynamically adjust the response data granularity according to user permissions and network environment, and return a combination of static layer thumbnail and dynamic metric summary text in a weak network environment.

2. The cross-industry big data sharing method according to claim 1, wherein In the dynamic update layer of step S2, the implementation method of double-threshold triggering is: The basic threshold Δ1 is set according to the industry: for the food industry, Δ1 = ±0.5°C / minute; for the machinery industry, Δ1 = 5% vibration energy change rate; for the clothing industry, Δ1 = ±2% fabric stretch rate; The emergency threshold Δ2 = 5×Δ1, and when jump synchronization is triggered, the dynamic layer data packet is attached with a red priority mark, and the encryption authentication layer inserts a time-sensitive network timestamp in the signature field; The degraded signature strength includes the following situations: When the number of updates per second of the dynamic layer > 50, the BLS-12-381 aggregated signature is used to replace the SM2+RSA two-level signature.

3. The cross-industry big data sharing method according to claim 2, wherein The generation logic of the BLS-12-381 aggregated signature is: (a)The edge node generates a Merkle tree for the dynamic data sequence within the Δt period and inputs the root hash value into the signature algorithm; (b)After the signature verification end verifies the root hash consistency through the pre-set public key, it extracts the corresponding period of dynamic data subset for integrity verification.

4. The cross-industry big data sharing method according to claim 1, characterized in that In step S1: The food industry uses a 2-second sliding window to match the temperature change rate calculation period; The machinery industry uses a 50ms micro-batch processing window to capture the fundamental frequency of the vibration spectrum; The clothing industry introduces a fabric surface light induction fingerprint collection device to generate laser topology hash features.

5. The cross-industry big data sharing method according to claim 4, wherein The fabric light induction fingerprint hash generation in step S1 includes: Using an 850nm laser beam to scan the fabric surface to generate a 1024×1024 pixel reflection grayscale matrix; Performing block discrete cosine transform on the matrix and taking the first 32 coefficients to be spliced into a feature vector; Inputting the feature vector into the SHA3-512 algorithm to generate a 256-bit laser hash identifier as an optional field in the static identification layer.

6. The method according to any one of claims 1-5, characterized in that, The method for generating the miniature map snapshot in step S3 is: (1) The central graph engine regularly extracts key entity subgraphs according to industry rules, including the fabric batch - supplier association chain in the clothing industry and the equipment serial number - replacement record chain in the machinery industry; (2) Compress the subgraph structure into a binary format and embed it into the four - state encoding of the corresponding product through the service extension layer; (3) When the edge device detects that the network latency > 200ms, call the local parsing engine to execute SPARQL queries based on the snapshot.

7. A cross-industry big data sharing system for implementing the method according to any one of claims 1-6, characterized in that, The system includes: Industry protocol adaptation module: Integrate the food temperature change rate calculation unit accelerated by FPGA, the mechanical vibration spectrum analysis DSP chip, and the clothing laser fingerprint scanning module; Dynamic encoding controller: Built - in dual - threshold comparator, signature policy selection circuit, and TSN timestamp injection module; Edge inference module: Equipped with a graph snapshot cache pool and a lightweight rule engine, supporting risk path derivation in offline mode; Adaptive rendering gateway: Built - in network bandwidth prediction model and geofence fuzzification module.

8. The system according to claim 7, wherein In the dynamic encoding controller: The dual - threshold comparator adopts an analog - digital hybrid circuit design, and the machinery industry version integrates a fast Fourier transform hardware accelerator; The signature policy selection circuit monitors the depth of the data queue in real - time and automatically activates the aggregated signature mode when the buffered data packets > 100.

9. The cross-industry big data sharing system according to claim 7, wherein, The operation logic of the edge inference module is: (a) After receiving the dynamic layer data, first query the local cache snapshot. If not hit, initiate a synchronization request to the central graph engine; (b) The local rule engine loads the Drools rule subset corresponding to the industry identifier. The machinery industry rule library contains a bearing fault modal feature library; (c) The inference result is attached with a credibility mark. When the credibility < 80%, request secondary verification from the central server.

10. The cross-industry big data sharing system according to any one of claims 7-9, characterized in that, The adaptive rendering gateway is built - in with: Network bandwidth prediction model: Establish an ARIMA time - series prediction based on historical transmission delays and dynamically select the JSON / ProtoBuf encoding format; Geofence fuzzification module: When the requested IP exceeds the preset regional range, convert the production address into an industry code + geographical region code.

Images