An industrial internet-based production data edge collaborative processing method

Abstract

This invention discloses a method for edge collaborative processing of production data based on the Industrial Internet, relating to the field of edge computing technology. The method includes: loading a stable boundary index mirror and a confirmed state mirror; preprocessing production data to generate process anchor points, differential slices, and unconfirmed anchor point interval indices; generating anchor point slices based on the unconfirmed anchor point interval indices, verifying continuity, extracting external references, constructing a causal hypergraph, and dividing it into closed and open anchor point slices; registering closed anchor point slices to the stable boundary index mirror, determining the nearest common anchor point, performing reverse rollback and normalized replay on open anchor point slices, and generating repair anchor point slices and repair ledgers; generating a compensation ledger based on the repair anchor point slices and repair ledgers and submitting it hierarchically; updating the stable boundary index mirror and the confirmed state mirror after successful verification, and partially reopening after verification failure. This invention improves the controllability of edge collaborative processing of production data in weak network and network outage scenarios.

Description

Technical Field

[0001] This invention relates to the field of edge computing technology, and in particular to a method for edge collaborative processing of production data based on the Industrial Internet. Background Technology

[0002] In the context of the Industrial Internet, production equipment, control terminals, and manufacturing execution systems typically access the network via edge nodes to collect, normalize, cache, and collaboratively report changes in workstation status, batch information, and attributes. Conventional processing methods often employ an architecture that combines edge-side preprocessing with central-side centralized confirmation. When the link fluctuates, data is temporarily stored locally first, and then retransmitted by interval and uniformly verified after communication is restored, in order to achieve continuous management and consistency maintenance of production data during the collection, transmission, and aggregation process.

[0003] In complex process-related scenarios, recovery uploads are often processed according to time segments or cache order, making it difficult to uniformly constrain cross-interval reference relationships and process sequence relationships, which limits the accuracy of determining the repair starting point and compensation boundary. In addition, re-upload verification usually processes boundary confirmation and differential submission together, lacking a layered acceptance mechanism, which makes it difficult to effectively converge the reprocessing range when local anomalies occur. Summary of the Invention

[0004] In view of the problems existing in the prior art, the present invention is proposed.

[0005] Therefore, this invention provides a collaborative edge processing method for production data based on the Industrial Internet to solve the problems of limited accuracy in determining the repair starting point and compensation boundary in weak network and network outage scenarios, and the difficulty in effectively converging the reprocessing range when there are local anomalies in existing technologies.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: This invention provides a method for edge collaborative processing of production data based on the Industrial Internet, comprising: loading a stable boundary index mirror and a confirmed state mirror; preprocessing the production data accessed by the Industrial Internet; generating process anchors and dividing anchor intervals when a process state change of the same production object is detected; generating differential slices for each state change within each anchor interval; continuously adding differential slices during weak network and network outage periods to form an unconfirmed anchor interval index; generating anchor slices based on the unconfirmed anchor interval index; verifying interval continuity through single-chain predecessor relationships within the anchor interval; extracting external reference sets through cross-slice predecessor reference relationships; constructing a causal hypergraph by combining external reference relationships, process sequence, and anchor inheritance relationships; and dividing the anchor slices into closed anchor slices. The system generates repaired anchor points and open anchor point shards. Closed anchor point shards are registered to the stable boundary index mirror. For open anchor point shards, all upstream paths are traced back along the causal hypergraph. The nearest common anchor point is determined from the common upstream anchor points already registered in the stable boundary index mirror and used as the unique repair starting point. Based on the confirmed state mirror corresponding to the nearest common anchor point, the open anchor point shards are reversed and canonicalized replay is performed to generate repaired anchor point shards and repair ledgers. The repaired anchor point shards and repair ledgers are combined to generate a compensation ledger including proof layers and difference layers. Layered compensation submissions are performed in the order of proof layers first and difference layers last. After successful verification, the stable boundary index mirror and confirmed state mirror are updated. After failed verification, only the open anchor point shard interval after the nearest common anchor point is partially reopened.

[0007] As a preferred embodiment of the industrial internet-based edge collaborative processing method for production data described in this invention, the preprocessing of production data accessed by the industrial internet includes: mapping a unified production object identifier based on data collected from different sources; extracting batch information from current sampled data, work order data, and local cache and binding it to the corresponding production object identifier; mapping a unified process status code; mapping the original field name to a unified attribute identifier; uniformly converting the original sampling time; performing status coding processing on the attribute value; and forming a standardized status record.

[0008] As a preferred embodiment of the industrial internet-based edge collaborative processing method for production data described in this invention, the step of generating process anchor points and dividing anchor point intervals when a process state switch of the same production object is detected includes: the edge node maintains the currently confirmed process state for each production object; when the process state code corresponding to the current sample is inconsistent with the currently confirmed process state, the edge node enters a continuous confirmation process, determines that a process state switch has occurred, and generates a new process anchor point; the area between the previous process anchor point and the current process anchor point is determined as a complete anchor point interval, and the current process anchor point is used as the starting point of the new interval to establish the currently opened anchor point interval; within the currently opened anchor point interval, when the previous state value and the subsequent state value corresponding to the current attribute are different, a corresponding differential piece is generated.

[0009] As a preferred embodiment of the industrial internet-based edge collaborative processing method for production data described in this invention, the step of continuously adding differential shards during weak network and network outage periods to form an unconfirmed anchor point interval index includes: continuously monitoring the link status with the central side, distinguishing between normal, weak network, and network outage states based on the link status; in the weak network state, allowing differential shards to be written to disk locally first and delaying confirmation and synchronization with the central side; in the network outage state, stopping the sending of real-time confirmation requests to the central side and continuously performing differential shard addition processing locally, establishing an unconfirmed anchor point interval index for edge nodes, and writing the index information corresponding to each anchor point interval that has not yet been confirmed by the central side into the unconfirmed anchor point interval index.

[0010] As a preferred embodiment of the industrial internet-based edge collaborative processing method for production data described in this invention, the step of generating anchor point fragments based on unconfirmed anchor point interval indexes includes: locating the corresponding process anchor point chain segment and anchor point interval differential fragment stack segment based on the unconfirmed anchor point interval index, and generating an anchor point fragment for each unconfirmed anchor point interval; when the current anchor point interval termination process anchor point identifier is empty and the current open state mark is valid, the current anchor point interval is determined as the currently open anchor point interval, and an anchor point fragment corresponding to the current open state mark is generated when the fragment generation trigger condition is met; when the current anchor point interval termination process anchor point identifier is valid, the current anchor point interval is determined as a closed complete interval, and the corresponding anchor point fragment is generated preferentially.

[0011] As a preferred embodiment of the industrial internet-based edge collaborative processing method for production data described in this invention, the step of constructing a causal hypergraph by combining external reference relationships, process sequence, and anchor point inheritance relationships, and dividing anchor point segments into closed anchor point segments and open anchor point segments, includes: using each anchor point segment as a node in the causal hypergraph, establishing data dependency edges based on external reference relationships, establishing sequence edges based on process sequence, establishing anchor point inheritance edges based on anchor point inheritance relationships, and constructing a causal hypergraph; determining the anchor point inheritance consistency value and calculating the segment closure judgment value by calculating the interval continuity coefficient and the proportion of external references, and dividing the anchor point segments into closed anchor point segments and open anchor point segments according to the closure judgment condition.

[0012] As a preferred embodiment of the industrial internet-based edge collaborative processing method for production data, the step of determining the nearest common anchor point from the common upstream anchor points registered in the stable boundary index mirror and using it as the sole repair starting point includes: taking the current open anchor point to be repaired as the starting node, tracing back all upstream paths along the data dependency edges, sequence edges, and anchor point inheritance edges in the causal hypergraph; matching each upstream anchor point in the upstream path set with the registered anchor points in the stable boundary index mirror, extracting the common upstream anchor points that appear simultaneously in all upstream paths and are registered in the stable boundary index mirror, and determining the nearest common anchor point, using the nearest common anchor point as the sole repair starting point.

[0013] As a preferred embodiment of the industrial internet-based edge collaborative processing method for production data described in this invention, the step of performing reverse rollback and standardized replay on open anchor point fragments based on the confirmed state image corresponding to the nearest common anchor point includes: establishing a repair work status table based on the confirmed state image corresponding to the nearest common anchor point; constructing the current repair interval with the open anchor point fragments from the nearest common anchor point to the current open anchor point fragment to be repaired and the corresponding differential fragments; performing reverse rollback on the current repair interval in a backward order; and performing standardized replay on the current repair interval starting from the confirmed state corresponding to the nearest common anchor point, in the order of first the anchor point interval and then the differential fragments, to generate repaired anchor point fragments and repair accounts.

[0014] As a preferred embodiment of the industrial internet-based edge collaborative processing method for production data described in this invention, the step of submitting layered compensation data in the order of submitting the proof layer first and the difference layer last includes: performing ordered organization on the repair anchor point sealing and repair accounts to generate the proof layer and difference layer; performing aggregated digest processing on the proof layer and difference layer respectively to obtain the proof layer digest value and difference layer digest value corresponding to the current compensation batch; combining the verification value of the previously completed and verified compensation batch, the current compensation batch identifier, and the encoded content corresponding to the nearest common anchor point identifier to perform concatenation and fixed-length one-way digest operation to obtain the current compensation batch verification value; submitting the proof layer and the current compensation batch identifier to the central side; after the proof layer structure consistency verification is passed, submitting the difference layer and the current compensation batch verification value to the central side; the central side regenerates the corresponding compensation batch verification value based on the received proof layer and difference layer, and performs acceptance judgment on the current compensation batch.

[0015] As a preferred embodiment of the industrial internet-based edge collaborative processing method for production data described in this invention, the step of updating the stable boundary index image and the confirmed state image after successful verification, and performing partial reopening only on the open anchor point slab interval after the nearest common anchor point after verification failure, includes: when the acceptance judgment corresponding to the current compensation batch is successful, updating the stable boundary index image and the confirmed state image by performing a repair confirmation write-back update on each repair anchor point slab; when the acceptance judgment corresponding to the current compensation batch is failed, the edge node uses the nearest common anchor point as the starting point of the partial reopening boundary, performs partial reopening only on the open anchor point slab interval after the nearest common anchor point, and rewrites the anchor point interval differential slab corresponding to the end of the current compensation batch after the nearest common anchor point into the repair queue.

[0016] The beneficial effects of this invention are as follows: by constructing a causal hypergraph and dividing anchor point patches into closed anchor point patches and open anchor point patches, a unified organization of cross-interval correlation, patch status and processing boundaries of production data is achieved; by determining the nearest common anchor point as the unique repair starting point, the orderly convergence of the repair path and the stable closure of the compensation link are achieved, improving the continuity, accuracy and controllability of edge collaborative processing of production data in weak network and network outage scenarios. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1This is a flowchart of a method for edge collaborative processing of production data based on the Industrial Internet.

[0019] Figure 2 A flowchart for generating anchor point covers and constructing a causal hypergraph.

[0020] Figure 3 The flowchart for determining the repair starting point and performing reverse rollback and specification replay.

[0021] Figure 4 A flowchart for generating a compensation ledger, hierarchical submission, and result processing.

[0022] Figure 5 Generate frequency maps for differential slices under different network conditions.

[0023] Figure 6 The graph shows the change in the success rate of the compensated link over time under different network conditions. Detailed Implementation

[0024] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0025] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0026] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0027] Reference Figures 1-6 This is one embodiment of the present invention, which provides a method for edge collaborative processing of production data based on the Industrial Internet, including the following steps: S1. Load the stable boundary index image and the confirmed state image, preprocess the production data accessed by the industrial Internet, generate process anchor points and divide anchor point intervals when the process state of the same production object is detected to change, generate differential slices for each state change in each anchor point interval, and continuously add differential slices during weak network and network outage periods to form an unconfirmed anchor point interval index.

[0028] Load the stable boundary index image through the edge node and read the corresponding confirmed state image.

[0029] Furthermore, after the edge node starts up, it loads the stable boundary index image from the local persistent storage area and loads the confirmed state image according to the state image address in the stable boundary index image; the edge node loads the stable boundary index image into the boundary index cache area and loads the confirmed state image into the state cache area.

[0030] Among them, the stable boundary index mirror is used to record the most recently confirmed data boundary information of the current production object; the confirmed state mirror is used to record the valid state of the current production object at the time of the most recent confirmation.

[0031] It should be noted that the stable boundary index mirror includes, but is not limited to, the production object identifier, the most recently confirmed process anchor point identifier, the most recently confirmed process anchor point sequence number, the most recently confirmed process anchor point timestamp, the most recently confirmed process anchor point summary value, the confirmed status mirror address, the most recently confirmed differential film tail identifier, and the corresponding confirmation batch number.

[0032] It should be noted that the production object identifier is obtained by the edge node after performing unified processing on the production data accessed by the Industrial Internet; the most recently confirmed process anchor point identifier, the most recently confirmed process anchor point sequence number, the most recently confirmed process anchor point timestamp, and the most recently confirmed process anchor point summary value are obtained from the confirmed process anchor point record written back to the local persistent storage area after the current production object completes the previous confirmation batch; the confirmed status image address is obtained from the local storage address allocated by the edge node for the corresponding status image when saving the confirmed status image; the most recently confirmed differential fragment tail identifier is obtained from the differential fragment identifier of the last confirmed differential fragment in the previous confirmation batch; the corresponding confirmation batch number is obtained from the confirmation batch information returned by the central side when the current production object is most recently confirmed, and is synchronously written to the local persistent storage area by the edge node.

[0033] It should be noted that the confirmed status image includes, but is not limited to, the current valid batch identifier, the current confirmed process status code, the current confirmed value of each target attribute, the anchor point identifier of the most recently confirmed process, and the tail identifier of the most recently confirmed differential image.

[0034] It should be noted that the current valid batch identifier is obtained from the valid batch information corresponding to the most recent confirmation completion of the current production object, or by inheriting the current valid batch identifier from the confirmed status image when the current sampling frame does not carry valid batch information; the current confirmed process status code is obtained from the process status corresponding to the most recent confirmation completion of the current production object; the current confirmed value of each target attribute is obtained from the valid status value corresponding to each target attribute when the most recent confirmation completion of the current production object; the most recently confirmed process anchor point identifier is obtained from the process anchor point record that was most recently confirmed by the central side; the most recently confirmed differential chip tail identifier is obtained from the differential chip identifier of the last differential chip in the corresponding anchor point interval when the most recent confirmation completion of the current production object.

[0035] The production data accessed through the Industrial Internet is processed by unifying production object identification, batch association, process association, attribute mapping, time normalization, and status coding.

[0036] Furthermore, the device address, workstation number, carrier number, and product code in data collected from different sources are mapped to a unified production object identifier; batch information is extracted from the current sampled data, work order data, and local cache, and the batch information is bound to the corresponding production object identifier; the workstation status, process step number, control step number, or process segment identifier in the original data is mapped to a unified process status code; the original field name is mapped to a unified attribute identifier; numerical attributes retain their original attribute dimensions, enumerated attributes are converted to integer codes according to the coding table, and Boolean attributes are uniformly converted to binary codes; the original sampling time is uniformly converted to a millisecond-level timestamp; and standardized status records are formed through status coding processing.

[0037] It should be noted that when the same sampling frame contains several identification fields, a unique production object identifier is generated by combining them according to the priority of the production object identifier.

[0038] Among them, the priority of the production object identifier is as follows: product code, carrier number, workstation number, and equipment address. For example, when the same sampling frame contains the product code, carrier number, workstation number, and equipment address at the same time, the product code is used as the primary identification field and the carrier number is used as the secondary identification field. The product code and the carrier number are combined to generate a unique production object identifier.

[0039] It should be noted that when the current sampling frame does not carry valid batch information, the current valid batch identifier in the confirmed state image is inherited.

[0040] It should be noted that each process status code corresponds to a unique process position in the process sequence table.

[0041] It should be noted that the encoding table used to convert enumerated attributes to integer codes is obtained by extracting the attribute encoding table from the local configuration library of the edge node. The attribute encoding table is issued by the central side and cached in the local configuration library of the edge node during edge node initialization.

[0042] It should be noted that when the original data lacks time information, the receiving time of the edge node is used as the current sampling time and marked as a time-addition record in the local record.

[0043] It should be noted that standardized status records include, but are not limited to, production object identifiers, batch identifiers, process status codes, attribute identifiers, current attribute values, millisecond-level timestamps, source link markers, and network status markers.

[0044] It should be noted that the production object identifier is obtained by mapping after unified processing of the production object identifier; the batch identifier is obtained from the batch information in the current sampled data, work order data, or local cache, and is obtained by inheriting the currently valid batch identifier in the confirmed status image when it is missing; the process status code is obtained by mapping after performing process association processing on the workstation status, process step number, control step number, or process segment identifier; the attribute identifier is obtained by mapping after the original field name is processed by attribute mapping; the current attribute value is obtained by processing the corresponding original attribute value in the current sample frame after status encoding; the millisecond-level timestamp is obtained by time normalization processing of the original sampling time, or by obtaining the time received by the edge node when the original data is missing time information; the source link mark is obtained by the access link type identifier of the current production data; and the network status mark is obtained by the edge node's real-time monitoring of the current link status.

[0045] Furthermore, edge nodes write standardized state records into a sampling cache queue and group them according to the production object identifier.

[0046] When a process state change is detected for the same production object, a process anchor point is generated, and the anchor point interval is divided by the adjacent process anchor points. Within each anchor point interval, a differential piece is generated for each state change.

[0047] Furthermore, the edge node maintains the current confirmed process status for each production object, using the current confirmed process status as the benchmark for judging process status switching. The edge node reads the standardized status record of the current production object from the sampling cache queue and obtains the process status code corresponding to the current sample. When the process status code corresponding to the current sample is consistent with the current confirmed process status, it is determined that no process status switching has occurred, and the current sample record continues to be included in the currently opened anchor point range. When the process status code corresponding to the current sample is inconsistent with the current confirmed process status, the edge node does not immediately generate a new process anchor point and enters the continuous confirmation process.

[0048] It should be noted that the current confirmed process status takes priority from its own confirmed status mirror; if the current production object has generated a new process anchor point in this round of operation, then the process status corresponding to the latest generated process anchor point is taken as the latest current confirmed process status.

[0049] It should be noted that the continuous confirmation process specifically involves the edge node continuously monitoring the process status code within a number of subsequent sampling periods; the process status switch of the current production object is determined to have occurred if and only if the new process status code reaches the number of consecutive sampling confirmations in the number of consecutive samplings.

[0050] It should be noted that the example of the number of consecutive sampling confirmations is set to 3 consecutive sampling periods.

[0051] Furthermore, when a process state switch is determined to be valid, the edge node immediately generates a new process anchor point. The edge node appends the newly generated process anchor point to the tail of the process anchor point chain of the corresponding production object. After the current process anchor point is generated, a complete anchor point interval is defined between the previous process anchor point and the current process anchor point, and the currently opened anchor point interval is established with the current process anchor point as the starting point of the new interval. For the currently opened anchor point interval, the edge node generates differential slices according to the attribute changes in the standardized state record. The edge node determines the previous state value for the current attribute, and the edge node uses the current value of the attribute in the current standardized state record as the subsequent state value. If the previous state value and subsequent state value corresponding to the current attribute are the same, no new differential slice is generated. If the previous state value and subsequent state value corresponding to the current attribute are different, a new differential slice is generated. After the edge node generates a differential slice, it immediately writes the current differential slice to the tail of the differential slice stack corresponding to the current anchor point interval, and synchronously updates the differential slice tail pointer and local sequence number of the current anchor point interval.

[0052] It should be noted that the new process anchor points include, but are not limited to, the current process anchor point identifier, the production object identifier, the batch identifier, the current process status code, the current process anchor point sequence number, the previous process anchor point identifier, the previous process anchor point summary value, the current process anchor point timestamp, and the current process anchor point summary value.

[0053] It should be noted that the current process anchor point identifier is generated and obtained by the edge node according to the anchor point generation order corresponding to the current production object after confirming the process state switch of the current production object; the production object identifier is obtained from the production object identifier in the standardized state record that triggered the process state switch; the batch identifier is obtained from the batch identifier in the standardized state record that triggered the process state switch, or obtained by inheriting the current valid batch identifier in the confirmed state image when the current sampling frame does not carry valid batch information; the current process state code is obtained from the process state code in the standardized state record that triggered the process state switch; the current process anchor point sequence number is generated and obtained by the edge node within the same production object according to the order of process anchor point generation; the previous process anchor point... The identifier is obtained from the most recently confirmed process anchor record of the current production object, or from the previous process anchor at the end of the current process anchor chain if a new process anchor has been generated in this round of operation; the summary value of the previous process anchor is obtained from the summary value in the process anchor record corresponding to the identifier of the previous process anchor; the timestamp of the current process anchor is obtained from the millisecond-level timestamp corresponding to the standardized status record that triggered the current process state switch; the summary value of the current process anchor is obtained by reading the production object identifier, batch identifier, current process status code, current process anchor sequence number, previous process anchor identifier, previous process anchor summary value and current process anchor timestamp in the current process anchor in the order of fields by the edge node, performing summary generation processing on the aforementioned fields, and writing it into the current process anchor record.

[0054] It should be noted that the generation granularity of differential slices is the state change of the same production object, the same anchor point range, and the same attribute identifier.

[0055] It should be noted that the order of reading the previous state value is as follows: first, read the next state value corresponding to the most recent difference piece in the current anchor point interval; if no difference piece corresponding to the current attribute has been generated in the current anchor point interval, then read the current confirmed value of the current attribute in the confirmed state image.

[0056] It should be noted that each differential slice includes, but is not limited to, differential slice identifier, process anchor point identifier, production object identifier, attribute identifier, previous state value, subsequent state value, positive change amount, negative change amount, local sequence number, single-chain predecessor pointer within the interval, cross-slice predecessor reference set, differential slice generation timestamp, and network status flag.

[0057] Within each interval, the single-chain predecessor pointer only points to the previous generated differential fragment within the same anchor point interval, maintaining the single-chain order relationship within the current anchor point interval. The cross-fragment predecessor reference set is filled only when the current differential fragment is a derived differential fragment, recording the related differential fragments referenced by the current differential fragment. Each referenced differential fragment in the cross-fragment predecessor reference set is a differential fragment that was already written to disk before the current differential fragment was generated, ensuring that the predecessor relationship does not form a cycle.

[0058] It should be noted that the differential piece identifier is generated and obtained by edge nodes within the current anchor point interval corresponding to the current production object, according to the order in which differential pieces are generated; the process anchor point identifier is obtained through the starting process anchor point corresponding to the currently opened anchor point interval to which the current differential piece belongs; the production object identifier is obtained through the production object identifier in the current standardized state record; the attribute identifier is obtained through the attribute identifier in the current standardized state record; the previous state value is obtained first through the subsequent state value corresponding to the most recent differential piece within the current anchor point interval; if the differential piece corresponding to the current attribute has not yet been generated within the current anchor point interval, it is obtained through the current confirmed value of the current attribute in the confirmed state mirror; the subsequent state value is obtained through the current value of the attribute in the current standardized state record; positive change The transformation quantity is calculated by the edge node based on the direction and magnitude of the change of the subsequent state value relative to the previous state value. The reverse transformation quantity is calculated by the edge node based on the reverse direction and magnitude of the change of the previous state value relative to the subsequent state value. The local sequence number is generated and obtained by the edge node in increments according to the order of differential fragment generation within the current anchor point interval. The single-chain predecessor pointer within the interval is obtained by the differential fragment identifier of the previous generated differential fragment within the current anchor point interval. The cross-fragment predecessor reference set is obtained by the set of differential fragment identifiers actually referenced by the current differential fragment during the generation process. The differential fragment generation timestamp is obtained by the millisecond-level timestamp corresponding to the current normalized state record. The network state marker is obtained by the edge node's real-time monitoring of the link state when the current differential fragment is generated.

[0059] In weak network and network outage states, differential pieces are continuously added to the corresponding differential piece stack according to the anchor point interval, forming a process anchor point chain, an anchor point interval differential piece stack, and an unconfirmed anchor point interval index.

[0060] Furthermore, edge nodes continuously monitor the link status with the central side and classify the network status into normal, weak, and offline states based on the link status. When the network status changes from normal to weak or offline, the edge node accumulates data for the currently opened anchor intervals primarily using local records. In the weak network state, the edge node allows differential shards to be written to disk locally first, delaying confirmation and synchronization with the central side. In the offline state, the edge node stops sending real-time confirmation requests to the central side and only continues to perform differential shard appending processing locally. The edge node establishes an unconfirmed anchor interval index record for each anchor interval that has not yet been confirmed by the central side.

[0061] It should be noted that the network status is categorized into normal, weak, and offline states based on link status. Specifically, edge nodes send link probe requests to the central side according to the probe cycle, recording the probe response latency, acknowledgment response timeout, and number of consecutive no-response times within the detection window. When all probe requests within the detection window receive a response within the acknowledgment time limit, it is determined to be in a normal state. When probe requests within the detection window still receive responses, but the average response latency exceeds the upper limit of the weak network acknowledgment latency, it is determined to be in a weak network state. When the acknowledgment success rate of link probe requests within the detection window is lower than the lower limit of the weak network success rate, it is determined to be in a weak network state. When no probe response is received for several consecutive times, such as three consecutive times, it is determined to be in an offline state. When a connection interruption with the central side is detected, it is determined to be in an offline state.

[0062] It should be noted that the upper limit for weak network confirmation latency is set to 800 milliseconds in the example; the lower limit for weak network success rate is set to 80% in the example.

[0063] It should be noted that the weak network state indicates that the corresponding edge node and the central side still maintain communication capability but the communication quality is reduced; the disconnected network state indicates that the corresponding edge node and the central side cannot conduct real-time confirmation interaction.

[0064] It should be noted that an unconfirmed anchor range index is composed of several unconfirmed anchor range index records.

[0065] It should be noted that when the network status flag changes from normal to weak network or offline, the process anchor point generation rules and differential piece generation rules are not changed.

[0066] It should be noted that the differential slice appending process specifically involves: receiving a new standardized state record; determining if the current attribute state has changed; generating a new differential slice for the changed attribute; appending the new differential slice to the end of the differential slice stack corresponding to the current anchor point interval; and updating the differential slice tail pointer, local sequence number, and network state flag of the current anchor point interval.

[0067] It should be noted that during weak network and network outage states, edge nodes are not allowed to overwrite already written differential fragments, nor are they allowed to merge differential fragments across anchor point intervals. They are only allowed to continuously append to the tail of the differential fragment stack corresponding to the current anchor point interval.

[0068] It should be noted that unconfirmed anchor interval index records include, but are not limited to, production object identifier, current anchor interval start process anchor identifier, current anchor interval end process anchor identifier or current open status marker, current anchor interval first difference fragment identifier, current anchor interval last difference fragment identifier, current anchor interval difference fragment quantity, current anchor interval network status marker, and index update timestamp.

[0069] It should be noted that the production object identifier is obtained through the production object identifier corresponding to the currently unconfirmed anchor point interval; the starting process anchor point identifier of the current anchor point interval is obtained through the starting process anchor point record corresponding to the currently unconfirmed anchor point interval; the ending process anchor point identifier and the current open status flag of the current anchor point interval are set and obtained by the edge node according to the closed status of the currently unconfirmed anchor point interval; the first differential fragment identifier of the current anchor point interval is obtained through the differential fragment identifier of the first differential fragment written in the currently unconfirmed anchor point interval; the last differential fragment identifier of the current anchor point interval is obtained through the differential fragment identifier of the most recently appended write in the currently unconfirmed anchor point interval; the number of differential fragments in the current anchor point interval is obtained by the edge node in real time by accumulating the number of differential fragments written in the currently unconfirmed anchor point interval; the network status flag of the current anchor point interval is obtained by the edge node in real time by monitoring the link status when the currently unconfirmed anchor point interval is formed or updated; and the index update timestamp is obtained through the local time corresponding to the most recently updated index record of the currently unconfirmed anchor point interval by the edge node.

[0070] It should be noted that when the current anchor interval has closed with the next process anchor, the termination process anchor identifier of the current anchor interval is filled with the corresponding next process anchor identifier; when the current anchor interval has not yet had a next process anchor, the termination process anchor identifier of the current anchor interval is empty, and the current open state flag is set to valid, indicating that the current anchor interval is in a continuous appending state; after each new differential fragment is appended, the edge node synchronously updates the tail differential fragment identifier, differential fragment quantity, and index update timestamp in the corresponding unconfirmed anchor interval index record.

[0071] It should be noted that when the link is restored, the edge nodes use the unconfirmed anchor interval index as the direct input source for generating anchor fragments.

[0072] It should be noted that, in order to verify that the differential sharding generation of production data can maintain continuity and reliability under different network conditions, three network environments were set up: normal, weak network, and network outage. Figure 5As shown, the results indicate that under normal network conditions, the generation frequency of differential fragments remains relatively stable, enabling rapid and continuous generation. Under weak network conditions, due to network bandwidth and latency issues, the generation frequency of differential fragments decreases, and the number of generated differential fragments fluctuates. Under network outage conditions, differential fragments continue to be generated locally, but due to the inability to synchronize with the central terminal, the generation frequency drops significantly, even approaching zero. Figure 5 The response capability under different network conditions can be clearly seen, verifying that the present invention can still guarantee the generation of production data in unstable or disconnected network environments, providing a foundation for subsequent data synchronization and repair.

[0073] S2. Generate anchor point covers based on the index of unconfirmed anchor point intervals, verify the continuity of intervals through the single-chain predecessor relationship within the anchor point interval, extract the external reference set through cross-piece predecessor reference relationship, construct a causal hypergraph by combining external reference relationship, process sequence and anchor point inheritance relationship, and divide the anchor point covers into closed anchor point covers and open anchor point covers.

[0074] Based on the index of the unconfirmed anchor point interval, locate the corresponding process anchor point chain segment and anchor point interval differential slice stack segment, and generate an anchor point seal that corresponds one-to-one with a single unconfirmed anchor point interval. For the currently open anchor point interval, generate an anchor point seal corresponding to the current open state mark when the seal generation trigger condition is met.

[0075] Furthermore, the edge node reads the unconfirmed anchor point interval index record and, based on the production object identifier, the starting process anchor point identifier, the ending process anchor point identifier or current open status flag, the first difference fragment identifier, the last difference fragment identifier, and the number of difference fragments in the current anchor point interval, locates the process anchor point chain segment and the anchor point interval difference fragment stack segment corresponding to the current unconfirmed anchor point interval; the edge node generates an anchor point encapsulation for each unconfirmed anchor point interval. It should be noted that when the anchor point identifier of the current anchor point interval termination process is empty and the current open status mark is valid, it means that the current anchor point interval is the currently open anchor point interval; when the anchor point identifier of the current anchor point interval termination process is valid, it means that the current anchor point interval is a complete interval that has been closed by the next process anchor point.

[0076] It should be noted that each anchor point corresponds one-to-one with a single unconfirmed anchor point interval and does not cross different unconfirmed anchor point intervals.

[0077] It should be noted that each anchor point includes an anchor point identifier, a production object identifier, the starting process anchor point identifier of the current anchor point interval, the ending process anchor point identifier or current open status marker of the current anchor point interval, the first differential fragment identifier of the current anchor point interval, the last differential fragment identifier of the current anchor point interval, the number of differential fragments in the current anchor point interval, the network status marker of the current anchor point interval, the differential fragment identifier sequence within the anchor point, the external reference set, the anchor point inheritance marker, and the anchor point status marker.

[0078] It should be noted that the sealing mark is generated by combining the production object mark, the starting process anchor mark of the current anchor point interval, and the ending process anchor mark of the current anchor point interval; when the current anchor point interval is not yet closed, the current open status mark replaces the ending process anchor mark of the current anchor point interval in the generation of the sealing mark.

[0079] It should be noted that edge nodes prioritize generating anchor point flaps for already closed complete intervals; for currently open anchor point intervals, anchor point flaps corresponding to the currently open state are generated when the flap generation trigger condition is met.

[0080] It should be noted that the generation trigger conditions include the first generation trigger condition, the second generation trigger condition, and the third generation trigger condition.

[0081] It should be noted that the first trigger condition is that the edge node detects that the link status between it and the central side has recovered from a weak network state or a network outage state to a communicable state; the second trigger condition is that the number of differential fragments written in the currently open anchor point interval has reached the fragment generation confirmation number; the third trigger condition is that the duration for which the currently open state marker remains valid has reached the fragment generation confirmation duration.

[0082] The example of the number of films to be confirmed is 5 differential films written consecutively; the example of the duration of the film to be confirmed is 30 seconds.

[0083] It should be noted that the starting point for the duration of the current open state flag's continued validity is the moment when the edge node sets the current open state flag to valid. Specifically, when the edge node detects that the current anchor point interval has not yet produced a subsequent process anchor point, it sets the current anchor point interval's termination process anchor point identifier to null and sets the current open state flag to valid, indicating that the current anchor point interval is in a continuous appending state. At this point, the timer for the continued validity of the current open state flag begins. During the period when the current open state flag remains valid, when the continuous timer reaches the sealing generation confirmation duration, it is determined that the third condition for generation triggering is met. The duration of continued validity is not recalculated using the index update timestamp after each appending differential fragment as the restart time.

[0084] It should be noted that the sealing generation trigger condition is met when any one of the first, second, or third trigger conditions is satisfied.

[0085] The continuity of the current anchor point interval is verified by comparing the predecessor pointing relationship of the single chain within the interval with the actual arrangement order of the differential pieces in the differential piece stack of the anchor point interval. The continuity verification result of the interval corresponding to the anchor point is formed by combining the number of consecutive matches and the number of consecutive mismatches.

[0086] Furthermore, the edge node, based on the first and last difference piece identifiers of the current anchor interval, sequentially traverses all difference pieces within the current anchor interval along the difference piece stack, and reads the predecessor pointer of the single chain within each difference piece. The edge node compares the predecessor pointer relationship of the single chain between two adjacent difference pieces within the current anchor interval with the actual arrangement order of the difference pieces in the anchor interval difference piece stack. When the predecessor pointer of the single chain within the interval of the later difference piece accurately points to the predecessor pointer of the earlier difference piece, the edge node will proceed with the next step. During fragmentation, a single consecutive match is recorded as one instance of continuous matching; a single chain predecessor pointer within the interval of the next difference fragment does not point to the previous difference fragment is recorded as one instance of continuous mismatch; a single pointer being null is recorded as one instance of continuous mismatch; a single pointer pointing to a difference fragment outside the anchor point interval is recorded as one instance of continuous mismatch; after completing the traversal, the edge node forms the interval continuity verification result corresponding to the current anchor point fragment based on the number of consecutive matches and the number of consecutive mismatches, and writes the interval continuity verification result into the fragmentation status flag of the current anchor point fragment.

[0087] It should be noted that when the current anchor interval contains only one differential piece, there are no adjacent differential piece pairs available for comparison. The edge node no longer performs the single-chain predecessor pointing relationship verification in the interval according to the adjacent differential pieces. At this time, the number of adjacent differential piece pairs that should be formed by the differential piece stack order of the anchor interval is 1, the actual number of consecutive matches is 1, the number of consecutive mismatches is 0, and the continuity of the current anchor interval is directly recorded as complete continuity.

[0088] Identify the external reference relationships between the current anchor interval and the differential pieces outside the anchor interval, extract the referenced differential pieces located outside the current anchor interval to form an external reference set, perform deduplication on the extracted external reference set, and obtain the final external reference set.

[0089] Furthermore, when traversing all differential fragments within the current anchor point interval, the edge node reads the cross-fragment predecessor reference set of each differential fragment one by one to identify the external reference relationship between the current anchor point interval and the differential fragments outside the anchor point interval. If the cross-fragment predecessor reference set of the current differential fragment is empty, the current differential fragment does not generate an external reference. If the cross-fragment predecessor reference set of the current differential fragment is not empty, the edge node checks each referenced differential fragment in the cross-fragment predecessor reference set to see if it is located within the current anchor point interval. When the referenced differential fragment is outside the current anchor point interval, the edge node writes the differential fragment identifier corresponding to the referenced differential fragment into the external reference set of the current anchor point fragment. When the referenced differential fragment is inside the current anchor point interval, the edge node does not write the referenced differential fragment into the external reference set. After all differential fragments have been traversed, the edge node performs deduplication processing on the duplicate differential fragment identifiers in the external reference set of the current anchor point fragment to form the final external reference set corresponding to the current anchor point fragment.

[0090] Each anchor point wafer is used as a node in the causal hypergraph, and data dependency edges, sequence edges, and anchor point inheritance edges are established between wafers based on external reference relationships, process sequence, and anchor point inheritance relationships, and a causal hypergraph is constructed.

[0091] Furthermore, edge nodes use each anchor point slicing as a node in the causal hypergraph. After the nodes are established, data dependency edges between slicings are established through external reference relationships, sequential edges between slicings are established through process sequence, and anchor point inheritance edges between slicings are established through anchor point inheritance relationships. For the relationship between the same target anchor point slicing and multiple current anchor point slicings, it is organized into a hyperedge, forming the hyperedge structure of the causal hypergraph.

[0092] It should be noted that the method of establishing edge relationships through external references is as follows: when the external reference set of the current anchor fragment contains an external differential fragment identifier, the edge node retrieves the target anchor fragment to which the corresponding external differential fragment identifier belongs, and establishes a data dependency edge between the target anchor fragment and the current anchor fragment. The edge direction of the data dependency edge is from the target anchor fragment to the current anchor fragment. When there are several current anchor fragments that reference differential fragments in the same target anchor fragment, the edge node organizes the relationship between the same target anchor fragment and several current anchor fragments into a hyperedge, forming a one-to-many causal propagation relationship.

[0093] It should be noted that the method of establishing edge relationships through process sequence is as follows: when two anchor point seals correspond to the same production object and the same batch, and the status codes of their respective processes satisfy the adjacent relationship in the process sequence table, the edge node establishes a sequential edge between the preceding anchor point seal and the following anchor point seal, and the direction of the sequential edge is from the preceding anchor point seal to the following anchor point seal; when the same production object has several branch process segments, the corresponding sequential edge is established according to the branch relationship defined in the process sequence table.

[0094] It should be noted that the method of establishing edge relationships through anchor point inheritance is as follows: when the previous process anchor point identifier and the previous process anchor point summary value in the current anchor point interval starting process anchor point record of the current anchor point segment match the corresponding anchor point record of the other anchor point segment, the edge node establishes an anchor point inheritance edge between the other anchor point segment and the current anchor point segment, with the edge direction pointing from the previous anchor point segment to the current anchor point segment.

[0095] It should be noted that edge construction based on external references takes precedence over edge construction based on process sequence. When both external references and process sequence relationships exist, edge nodes prioritize retaining edges corresponding to external references and record the process sequence relationship as an auxiliary sequence attribute. When both anchor inheritance relationships and external reference relationships exist, edge nodes retain both anchor inheritance relationships and external reference relationships.

[0096] It should be noted that the auxiliary order attribute is recorded as an additional attribute field of the corresponding edge in the causal hypergraph.

[0097] Calculate the closure judgment value for each anchor point, and classify closed anchor point blocks and open anchor point blocks based on the closure judgment value. The anchor point block state is classified by uniformly representing the interval continuity coefficient, the proportion of external references, and the consistency value of anchor point inheritance.

[0098] Furthermore, for each anchor point patch, the number of adjacent differential patch pairs that should be formed by the anchor point interval differential patch stack order and the actual number of consecutive matches are counted. By calculating the ratio of the actual number of consecutive matches to the number of adjacent differential patch pairs that should be formed by the anchor point interval differential patch stack order, the interval continuity coefficient of the current anchor point patch is obtained.

[0099] Among them, the closer the interval continuity coefficient of the current anchor point is to 1, the more complete the interval continuity within the current anchor point is.

[0100] It should be noted that when the number of differential pieces within the current anchor point is greater than 1, the number of adjacent differential piece pairs that should be formed by the differential piece stack order of the anchor point interval is the number of differential pieces within the current anchor point minus 1; when the number of differential pieces within the current anchor point is equal to 1, the number of adjacent differential piece pairs that should be formed by the differential piece stack order of the anchor point interval is 1.

[0101] Furthermore, the total number of all cross-section predecessor references within the anchor section is counted, along with the number of deduplicated references in the external reference set of the current anchor section. When the total number of all cross-section predecessor references within the anchor section is greater than 0, the external reference ratio of the current anchor section is obtained by calculating the ratio of the number of deduplicated references in the external reference set of the current anchor section to the total number of cross-section predecessor references within the anchor section.

[0102] The closer the external reference ratio of the current anchor point wrapper is to 0, the lower the dependence of the current anchor point wrapper on the external differential wrapper.

[0103] Furthermore, when the anchor identifier and summary value of the previous process in the anchor record of the current anchor interval of the current anchor point segment match simultaneously in the corresponding preceding anchor point segment, the anchor point inheritance consistency value of the anchor point segment is set to 1; when either the anchor identifier or the summary value of the previous process in the anchor record of the current anchor interval of the current anchor point segment cannot match in the corresponding preceding anchor point segment, the anchor point inheritance consistency value of the anchor point segment is set to 0; the segment closure judgment value is calculated by combining the interval continuity coefficient of the current anchor point segment, the external reference ratio of the current anchor point segment, and the anchor point inheritance consistency value of the anchor point segment.

[0104] Among them, the sealing closure judgment value is used to comprehensively characterize the degree of interval continuity and integrity of the current anchor point sealing, the degree of dependence on the external differential sheet of the sealing, and the degree of consistency of anchor point inheritance.

[0105] It should be noted that the calculation of the seal closure judgment value is expressed as: ; in, This indicates the seal closure determination value. This represents the interval continuity coefficient of the current anchor point sealing area. This indicates the percentage of external references to the current anchor point encapsulation. This indicates that the anchor point of the anchor point seal inherits the same value.

[0106] It should be noted that, The range of values ​​is .

[0107] Furthermore, based on the sealing closure determination value, the anchor point sealing is marked as a closed anchor point sealing and an open anchor point sealing; when the sealing closure determination value is equal to 1, the current anchor point sealing is marked as a closed anchor point sealing; when the sealing closure determination value is not equal to 1, the current anchor point sealing is marked as an open anchor point sealing.

[0108] S3. Register the closed anchor point fragment to the stable boundary index mirror. For the open anchor point fragment, backtrack all upstream paths along the causal hypergraph. Determine the nearest common anchor point from the common upstream anchor points already registered in the stable boundary index mirror and use it as the unique repair starting point. Perform reverse rollback and normalized replay on the open anchor point fragment based on the confirmed state mirror corresponding to the nearest common anchor point to generate the repaired anchor point fragment and repair account.

[0109] Register the closed anchor point seal to the stable boundary index mirror, and summarize them to form the stable boundary set corresponding to the current production object.

[0110] Furthermore, the edge nodes read the set of closed anchor point slices and register each closed anchor point slice to the stable boundary index mirror. During the registration process, the edge nodes write the production object identifier, batch identifier, starting process anchor point identifier of the current anchor point interval, ending process anchor point identifier of the current anchor point interval, ending process anchor point summary value, slice status flag, slice confirmation sequence number, and corresponding confirmed status mirror address corresponding to each closed anchor point slice into the boundary registration area of ​​the stable boundary index mirror. The edge nodes summarize all the closed anchor point slice boundary information registered by the current production object in the stable boundary index mirror to form the stable boundary set corresponding to the current production object.

[0111] Among them, the stable boundary set is used to limit the boundary range that subsequent open anchor point sealing can be traced back and used as a candidate repair starting point.

[0112] It should be noted that the anchor point identifier of the current anchor point interval termination process and the anchor point summary value of the termination process jointly represent the boundary endpoint of the current closed anchor point seal; the seal confirmation sequence number is generated and obtained by the edge nodes in ascending order according to the order in which the closed anchor point seals are confirmed; the corresponding confirmed state image address is obtained by the local storage address allocated by the edge nodes when saving the state image corresponding to the current closed anchor point seal.

[0113] It should be noted that the stable boundary set includes the registered anchor point identifiers of each termination process, the summary values ​​of each termination process anchor point, the confirmation sequence number of each sealing plate, and the corresponding confirmed status mirror address.

[0114] Based on the open anchor point sealing, all upstream paths are traced back along the causal hypergraph to extract the common upstream anchor points that have been registered in the stable boundary index mirror. When the candidate common upstream anchor points are not unique, the unique nearest common anchor point is determined as the repair starting point based on the nearest common anchor point preference value.

[0115] Furthermore, the edge node selects the current open anchor point fragment to be repaired from the set of open anchor point fragments. Starting from the current open anchor point fragment to be repaired, it backtracks all upstream paths along the data dependency edges, sequence edges, and anchor point inheritance edges in the causal hypergraph. During the backtracking process, the edge node records the anchor point fragment nodes passed through in each upstream path and their corresponding starting process anchor point identifier, ending process anchor point identifier, and ending process anchor point summary value, forming the upstream path set of the current open anchor point fragment to be repaired.

[0116] It should be noted that when there is no upstream node for the current open anchor point patch to be repaired, the edge node marks the current open anchor point patch to be repaired as having no common upstream path input, and directly determines the nearest common anchor point as the registered closed anchor point patch boundary with the largest patch confirmation sequence number in the stable boundary index mirror of the current production object.

[0117] Furthermore, the edge node matches each upstream anchor point in the upstream path set with the registered anchor points in the stable boundary index mirror, extracts common upstream anchor points that appear simultaneously in all upstream paths and are already registered in the stable boundary index mirror, forming a common upstream anchor point set; when there is only one common upstream anchor point in the common upstream anchor point set, the edge node directly determines the current common upstream anchor point as the nearest common anchor point; when the number of common upstream anchor points in the common upstream anchor point set is not less than 2, the edge node calculates the nearest common anchor point preference value for each candidate common upstream anchor point, and determines the candidate common upstream anchor point with the largest preference value as the nearest common anchor point.

[0118] It should be noted that the nearest common anchor point is used to comprehensively compare the validity of boundary registration and the anchor point distance with the current open anchor point to be repaired among several common upstream anchor points to determine the unique nearest common anchor point.

[0119] Furthermore, the preferred value for the nearest common anchor point is calculated as follows: ; in, Indicates the first The nearest common anchor point preference value among candidate common upstream anchor points Indicates the first Valid boundary registration values ​​for each candidate common upstream anchor point This indicates that the currently unrepaired open anchor point seal is related to the first... Anchor distance values ​​between candidate common upstream anchor points.

[0120] It should be noted that, The range of values ​​is .

[0121] It should be noted that when the first When all candidate common upstream anchor points have been effectively registered in the stable boundary index mirror, and the corresponding termination process anchor point identifier and termination process anchor point summary value can both be matched in the stable boundary index mirror, the valid boundary registration value is 1; when the th... When the first candidate common upstream anchor point is not registered in the stable boundary index mirror, the valid boundary registration value is 0. If neither the termination process anchor identifier nor the termination process anchor summary value corresponding to a candidate common upstream anchor point can be matched in the stable boundary index mirror, the valid value of the boundary registration is 0.

[0122] It should be noted that the anchor point distance value is determined by the sequence number of the starting process of the current open anchor point sealing patch to be repaired and the number of anchor points. Obtain the difference between the process anchor number of each candidate common upstream anchor point.

[0123] It should be noted that, Take an integer that is not less than 0.

[0124] It should be noted that the larger the recent common anchor point preference value, the more suitable the current candidate common upstream anchor point is as the recent common anchor point for the current open anchor point sealing patch to be repaired.

[0125] It should be noted that when there are several candidate common upstream anchors with the same nearest common anchor value, the edge node preferentially selects the candidate common upstream anchor with the larger sealing confirmation sequence number as the nearest common anchor; when the sealing confirmation sequence number is still the same, the edge node preferentially selects the candidate common upstream anchor with the later termination process anchor timestamp as the nearest common anchor.

[0126] Based on the confirmed state mirror corresponding to the most recent common anchor point, a repair work status table is established. The open anchor point sealing and corresponding differential pieces after the most recent common anchor point and before the current open anchor point sealing are used to form the current repair interval. The reverse rollback is performed along the current repair interval to restore the repair work status to the confirmed state corresponding to the most recent common anchor point.

[0127] Furthermore, the edge node reads the confirmed state image corresponding to the nearest common anchor point based on the corresponding confirmed state image address in the stable boundary index image, and establishes a repair work status table; it determines the current repair interval based on the upstream path relationship between the nearest common anchor point and the current open anchor point slice to be repaired; it performs reverse rollback in reverse order from back to front; when all differential slices in the current repair interval have completed the reverse rollback, the status in the repair work status table is rolled back to the confirmed state corresponding to the nearest common anchor point.

[0128] The repair status table is used to record the temporary status changes of the currently unrepaired open anchor point during the repair process. The initial status of the repair status table is generated by mirroring the confirmed status corresponding to the nearest common anchor point.

[0129] It should be noted that the repair work status table includes the production object identifier, the current valid batch identifier, the current confirmed process status code, the current status value of each target attribute, the current rolled-back differential film tail identifier, and the current repair start point marker.

[0130] It should be noted that the current interval to be repaired consists of all open anchor point patches along the path from the nearest common anchor point to the current open anchor point patch to be repaired, as well as the corresponding anchor point interval difference patches.

[0131] It should be noted that the reverse rollback process involves the edge node starting from the tail difference piece of the current anchor interval of the open anchor fragment to be repaired, and traversing the difference pieces in reverse along the single-chain predecessor pointer within the interval. Once the current anchor interval rollback is complete, it switches to the tail difference piece of the anchor interval corresponding to the previous open anchor fragment along the upstream path of the causal hypergraph, and continues the reverse rollback. Each time a difference piece is rolled back, the edge node adjusts the corresponding attribute in the repair status table based on the reverse change recorded in the current difference piece. The current state value is restored to the previous state value; when the current differential slice has a cross-slice predecessor reference set, and all differential slices referenced in the cross-slice predecessor reference set have completed rollback, the edge node performs rollback on the current differential slice; when the current differential slice has a cross-slice predecessor reference set, and there are referenced differential slices in the cross-slice predecessor reference set that have not yet completed rollback, the edge node first marks the current differential slice as a differential slice to be rolled back, and then performs rollback again after all referenced differential slices corresponding to the current differential slice have completed rollback.

[0132] Starting with the confirmed state corresponding to the nearest common anchor point, the current interval to be repaired is replayed according to the order of anchor point interval first and then differential piece. The repaired state sequence is reconstructed based on the positive change of differential piece and cross-piece predecessor reference relationship.

[0133] Furthermore, starting from the confirmed state corresponding to the nearest common anchor point, the edge node performs standardized replay on all differential pieces within the current repair interval, in the order of first the anchor point interval and then the differential pieces. After completing the standardized replay of all differential pieces within the anchor point interval corresponding to each open anchor point, the state snapshot in the current repair work state table is recorded as the replay result of the current open anchor point piece, and the standardized replay continues to be performed on the next open anchor point piece until all open anchor point pieces within the current repair interval have completed standardized replay, resulting in the state sequence after standardized replay.

[0134] It should be noted that the replay specification is as follows: edge nodes select open anchor point fragments to be replayed sequentially according to the path direction from the nearest common anchor point to the current open anchor point fragment to be repaired; within each open anchor point fragment, replay is performed sequentially from the fragment corresponding to the first difference fragment identifier to the fragment corresponding to the last difference fragment identifier, according to the arrangement order in the difference fragment stack of the anchor point interval; when replaying a difference fragment, the current status value of the corresponding attribute in the repair work status table is updated according to the positive change recorded in the current difference fragment; when the current difference fragment has a cross-fragment predecessor reference set, and all difference fragments referenced in the cross-fragment predecessor reference set have been replayed, the edge node performs replay on the current difference fragment; when the current difference fragment has a cross-fragment predecessor reference set, and there are referenced difference fragments in the cross-fragment predecessor reference set that have not yet been replayed, the edge node first marks the current difference fragment as a difference fragment to be replayed, and then performs replay again after all referenced difference fragments corresponding to the current difference fragment have been replayed.

[0135] It should be noted that the replayed state sequence is used to replace the original unconfirmed state sequence corresponding to the original open anchor point seal within the current repair interval, forming the repaired seal boundary and seal internal state.

[0136] Based on the replayed state sequence, repair anchor points are regenerated to correspond one-to-one with the original open anchor points, and repair ledgers are generated simultaneously, recording the mapping relationship between the original open anchor points and repair anchor points, repair boundary information, and repair status information.

[0137] Furthermore, after the entire current repair interval has been replayed according to the standardization, the edge nodes regenerate repair anchor fragments based on the state sequence after the standardization replay. While generating repair anchor fragments, a corresponding repair account is generated for each repair anchor fragment. When the current production object has several open anchor fragments, the edge nodes execute the following steps in the order from upstream to downstream of the open anchor fragments in the causal hypergraph: determine the nearest common anchor, perform reverse rollback, perform standardization replay, and generate repair anchor fragments and repair accounts, thus generating the corresponding set of repair anchor fragments and set of repair accounts.

[0138] Among them, the repaired anchor point cover and the original open anchor point cover maintain a one-to-one correspondence on the boundary of the anchor point interval. The differential cover identifier sequence, external reference set, anchor point inheritance mark and cover status mark of the repaired anchor point cover are recalculated and written according to the results of the replay according to the specification. The repair ledger is used to record the mapping relationship between the original open anchor point cover and the repaired anchor point cover, as well as the boundary information of this repair.

[0139] It should be noted that the repaired anchor point sealing includes the repaired sealing identifier, the corresponding original open anchor point sealing identifier, the production object identifier, the anchor point identifier of the starting process after repair, the anchor point identifier of the ending process after repair, the first differential piece identifier after repair, the tail differential piece identifier after repair, the number of differential pieces after repair, the external reference set after repair, and the sealing status mark after repair.

[0140] It should be noted that each repair entry includes the repair entry identifier, the corresponding original open anchor point sealing fragment identifier, the corresponding repair anchor point sealing fragment identifier, the most recent common anchor point identifier, the rollback start differential fragment identifier, the replay termination differential fragment identifier, the sealing fragment status mark before repair, the sealing fragment status mark after repair, and the repair completion timestamp.

[0141] Among them, the most recent public anchor point identifier in the repair account is used to identify the unique repair starting point of the current repair process; the pre-repair sealing status marker is used to record the open status of the original open anchor point sealing before repair; and the post-repair sealing status marker is used to record the status of the repaired anchor point sealing after the standard replay is completed.

[0142] S4. Combine the repaired anchor point sharding and repaired ledger to generate a compensation ledger including the proof layer and the difference layer. Perform layer compensation submission in the order of proof layer submission first and difference layer submission last. After the verification passes, update the stable boundary index mirror and the confirmed state mirror. After the verification fails, perform local reopening only on the open anchor point sharding interval after the nearest common anchor point.

[0143] By combining the sealing of repair anchor points and the orderly organization of repair accounts, a proof layer carrying repair boundary information and a difference layer carrying repair differential data are generated respectively, forming the compensation ledger corresponding to the current compensation batch.

[0144] Furthermore, the edge nodes read the set of repair anchor point seals and the set of repair accounts, and generate a compensation ledger based on the set of repair anchor point seals and the set of repair accounts; the repair accounts are ordered according to the order of the repair completion timestamps and repair account identifiers in the set of repair accounts; the edge nodes are ordered according to the order of the post-repair start process anchor point identifiers and repair seal identifiers in the set of repair anchor point seals; and the proof layer and difference layer corresponding to the current compensation batch are generated based on the arrangement results.

[0145] It should be noted that the compensation ledger includes an proof layer and a difference layer. The proof layer is used to carry the repair boundary, repair mapping relationship and ledger-level verification information; the difference layer is used to carry the actual difference data content corresponding to the repair anchor point sealing.

[0146] It should be noted that the proof layer includes the repair account identification sequence, the corresponding original open anchor point sealing fragment identification sequence, the corresponding repair anchor point sealing fragment identification sequence, the most recent common anchor point identification sequence, the rollback start differential fragment identification sequence, the replay termination differential fragment identification sequence, the sealing fragment status mark sequence before repair, the sealing fragment status mark sequence after repair, the repair completion timestamp sequence, and the current compensation batch identifier.

[0147] It should be noted that the differential layer includes the identifier of the first differential layer after repair, the identifier of the last differential layer after repair, the number of differential layers after repair, the set of external references after repair, the status marker of the layer after repair, and the sequence of differential layer content corresponding to the layer at the repair anchor point.

[0148] It should be noted that the current compensation batch identifier is generated and obtained by the edge nodes by incrementing the current compensation ledger to be submitted in the order of submission.

[0149] Aggregate digest processing is performed on the proof layer and the difference layer respectively, and the compensation batch verification value is calculated by combining the previous compensation batch verification value, the current most recent common anchor point identifier and the current compensation batch identifier, so as to form a unified verification basis for the layered compensation submission.

[0150] Furthermore, after generating the proof layer and the difference layer, the edge nodes perform aggregate digest processing on the proof layer and the difference layer respectively to obtain the proof layer digest value and the difference layer digest value; and calculate the compensation batch verification value for the current compensation batch.

[0151] Among them, the compensation batch verification value is used to uniformly represent the proof layer content, difference layer content and submission continuity relationship with the previous compensation batch in the current compensation batch, and serves as the direct verification basis for the central side to perform acceptance of the current compensation batch.

[0152] It should be noted that the proof layer summary value is obtained by encoding the repair account fields in the proof layer in sequence and then performing summary generation processing; the difference layer summary value is obtained by encoding the repair anchor point sealing field and the corresponding difference fragment content in the difference layer in sequence and then performing summary generation processing.

[0153] It should be noted that the calculated verification value for the compensation batch is expressed as follows: ; in, This indicates the verification value of the current compensation batch. This indicates the verification value of the previous compensation batch that has been submitted and passed verification. This represents the current compensation batch proof layer summary value. This represents the current compensated batch difference stratified summary value. This indicates the nearest common anchor point identifier corresponding to the current compensation batch. Indicates the current compensation batch identifier. This indicates that encoding processing will be performed on the corresponding field. This indicates that a fixed-length one-way digest operation will be performed on the encoded byte string formed by concatenating the above fields in order. Indicates according to , , , as well as The corresponding encoded results are concatenated sequentially according to their order.

[0154] It should be noted that the fixed-length one-way digest operation example uses... Digest algorithm or Digest algorithm.

[0155] It should be noted that, Used to uniquely represent the compensation ledger status of the current compensation batch at the current edge node.

[0156] It should be noted that when there is no verification value from a previous completed and verified compensation batch in the current compensation batch, Obtained by initial compensation verification base value of edge nodes.

[0157] It should be noted that, Obtained by the nearest common anchor point identifier corresponding to each repaired account in the current compensation batch; when the number of repaired accounts corresponding to the current compensation batch is not less than 1. Obtained by the nearest common anchor point identifier with the earliest sealing confirmation sequence number in the current compensation batch.

[0158] It should be noted that, To convert the corresponding fields into concatenable encoded byte strings, numeric fields are encoded with a fixed byte length, while identifier fields are encoded by combining the field length with the field content.

[0159] The layered compensation submission is performed to the central side in the order of first submitting the proof layer and then submitting the difference layer. The central side then performs acceptance judgment on the current compensation batch based on the consistency of the proof layer structure and the verification value of the compensation batch.

[0160] Furthermore, after generating the proof layer, difference layer, and compensation batch verification value corresponding to the current compensation batch, the edge node first submits the proof layer to the central side, and then submits the difference layer to the central side. Specifically, the edge node first sends the proof layer and the current compensation batch identifier to the central side. After receiving the proof layer, the central side performs a structural consistency check on the proof layer. When the central side determines that the structural consistency check of the proof layer passes, the central side returns a proof layer reception pass flag to the edge node. When the central side determines that the structural consistency check of the proof layer fails, the central side returns a proof layer reception failure flag to the edge node and terminates the submission of the difference layer for the current compensation batch. When the edge node receives the proof layer reception pass flag, the edge node sends the difference layer and the current compensation batch verification value to the central side.

[0161] It should be noted that the structural consistency verification includes: verification of the integrity of the repaired account field, verification of the mapping integrity between the original open anchor point shard identifier and the corresponding repaired anchor point shard identifier, verification of the consistency between the most recent common anchor point identifier and the registered boundary in the current production object stable boundary index mirror, and verification of the rationality of the state transition between the shard status mark before repair and the shard status mark after repair.

[0162] Furthermore, after receiving the differential layer, the central side writes the repaired differential fragment content sequence in the differential layer into the verification buffer, and recalculates the corresponding compensation batch verification value based on the received proof layer and differential layer. When the compensation batch verification value recalculated by the central side is consistent with the compensation batch verification value submitted by the edge node, the central side determines that the current compensation batch verification has passed; when the compensation batch verification value recalculated by the central side is inconsistent with the compensation batch verification value submitted by the edge node, the central side determines that the current compensation batch verification has failed.

[0163] After the current compensation batch passes verification, the repair anchor point is written to the local confirmation area, and the stable boundary index image and the confirmed status image are updated simultaneously to complete the confirmation of the repair result corresponding to the current compensation batch.

[0164] Furthermore, once the central side determines that the current compensation batch has passed verification, the edge node writes the set of repair anchor points in the current compensation batch into the local confirmation area and simultaneously updates the stable boundary index image and the confirmed status image. For each repair anchor point in the current compensation batch, a repair confirmation write-back update is performed. After the stable boundary index image and the confirmed status image are updated, the current compensation batch is marked as completed and the verification value of the current compensation batch is written into the local completed compensation batch record, which serves as the source for calculating the verification value of the previous completed and verified compensation batch in the next compensation batch calculation.

[0165] It should be noted that the repair confirmation write-back update specifically involves: marking the corresponding original open anchor point cover as replaced; registering the repaired anchor point cover as a new confirmed anchor point cover; writing the anchor point identifier of the terminated process after repair, the summary value of the anchor point of the terminated process after repair, the cover status mark after repair, and the corresponding compensation batch identifier into the stable boundary index image; and writing the snapshot of the repair work status after the specification replay is completed into the local storage address corresponding to the confirmed status image.

[0166] Among them, the confirmed anchor point seal refers to the repair anchor point seal that has been verified and written into the local confirmation area after being approved by the current compensation batch.

[0167] It should be noted that when updating the stable boundary index mirror, the edge node prioritizes replacing the unconfirmed boundary corresponding to the original open anchor point with the repair anchor point boundary that has the latest confirmation sequence number and has passed verification in the current compensation batch.

[0168] It should be noted that when the edge node updates the confirmed state image, it writes the state snapshot corresponding to the last repair anchor point that has completed the standard replay in the current compensation batch into the latest confirmed state image record corresponding to the current production object, and synchronously updates the corresponding confirmed state image address.

[0169] After the current compensation batch verification fails, the nearest common anchor point is used as the starting point of the local reopening boundary. The reopening range is limited by performing invalidation reset, repair re-entry, and resubmission processing only on the open anchor point sealed area after the nearest common anchor point.

[0170] Furthermore, when the central side determines that the current compensation batch verification fails, the edge node reads the nearest common anchor point identifier in the repair ledger corresponding to the current compensation batch, and uses the nearest common anchor point identifier as the starting point of the partial reopening boundary; it locates the open anchor point sealing interval corresponding to the current compensation batch, and performs partial reopening only on the open anchor point sealing interval after the nearest common anchor point; after completing the partial reopening, it re-executes the process of determining the nearest common anchor point, reverse rollback, standard replay, and generating repair anchor point sealing and repair ledger only on the open anchor point sealing interval after the nearest common anchor point, and re-executes the generation of compensation ledger and hierarchical compensation submission until the compensation batch corresponding to the current partial reopening interval passes the verification.

[0171] It should be noted that when the central side determines that the current compensation batch verification has failed, the edge nodes will not perform a complete restart of all unconfirmed anchor point intervals.

[0172] It should be noted that partial reopening is only performed on open anchor point fragment intervals following the most recent common anchor point. Specifically, the repair anchor point fragments corresponding to the current compensation batch are marked as invalid due to verification failure; the repair target corresponding to the current compensation batch is marked as pending redo; the boundary information of the most recent common anchor point and the closed anchor point fragments already registered in the stable boundary index mirror before the most recent common anchor point remains unchanged; the open anchor point fragments from the most recent common anchor point to the end of the current compensation batch are remarked as open anchor point fragments to be repaired; and the anchor point interval difference fragments corresponding to the most recent common anchor point to the end of the current compensation batch are rewritten into the pending repair queue.

[0173] It should be noted that the open anchor point sealing interval is all open anchor point sealings and corresponding anchor point interval differential pieces from the nearest common anchor point to the end of the current compensation batch.

[0174] It should be noted that partial reopening does not modify the closed anchor fragments that have passed verification before the most recent common anchor, does not delete the stable boundary index mirrors that have been updated before the most recent common anchor, and does not overwrite the confirmed state mirrors that have been confirmed before the most recent common anchor.

[0175] It should be noted that, in order to verify the success rate and stability of the compensation link under different network environments, especially its recovery capability under weak network and network outage conditions, compensation link tests were conducted under different network conditions, such as... Figure 6 As shown, the results indicate that under normal network conditions, the success rate of the compensation link is high, the repair process is relatively smooth, and data synchronization can be completed quickly. However, under weak network conditions, due to the increase in network latency, the success rate of the compensation link fluctuates slightly, and the efficiency of data synchronization decreases. Under network outage conditions, the success rate of the compensation link decreases significantly, and even synchronization failures occur. The repair operation is affected by network interruption and cannot restore data consistency in a timely manner. Figure 6 This reflects the effectiveness of the compensation link under different network conditions and verifies that the present invention can effectively address the problem of production data synchronization under different network conditions through a multi-level compensation mechanism, especially its ability to repair in extreme network environments.

[0176] In summary, this invention achieves unified organization of cross-regional correlation, patch status, and processing boundaries of production data by constructing a causal hypergraph and dividing anchor patch into closed and open anchor patch. Furthermore, by determining the nearest common anchor as the sole repair starting point, it achieves orderly convergence of repair paths and stable closure of compensation links, thereby improving the continuity, accuracy, and controllability of edge collaborative processing of production data in weak network and network outage scenarios.

[0177] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for edge collaborative processing of production data based on the Industrial Internet, characterized in that, include: Load the stable boundary index image and the confirmed state image, preprocess the production data accessed by the industrial Internet, generate process anchor points and divide anchor point intervals when the process state of the same production object is detected to change, generate differential slices for each state change within each anchor point interval, and continuously add differential slices during weak network and network outage periods to form an unconfirmed anchor point interval index. Anchor slices are generated based on the unconfirmed anchor interval index. The continuity of the interval is verified by the single-chain predecessor relationship within the anchor interval. The external reference set is extracted by the cross-slice predecessor reference relationship. A causal hypergraph is constructed by combining the external reference relationship, process sequence and anchor inheritance relationship. The anchor slices are divided into closed anchor slices and open anchor slices. Register the closed anchor point patch to the stable boundary index mirror. For the open anchor point patch, backtrack all upstream paths along the causal hypergraph. Determine the nearest common anchor point from the common upstream anchor points already registered in the stable boundary index mirror and use it as the unique repair starting point. Perform reverse rollback and normalized replay on the open anchor point patch based on the confirmed state mirror corresponding to the nearest common anchor point to generate the repair anchor point patch and repair account. By combining the repaired anchor point sharding and the repaired ledger, a compensation ledger including the proof layer and the difference layer is generated. The layer compensation is committed in the order of the proof layer being committed first and the difference layer being committed later. After the verification passes, the stable boundary index mirror and the confirmed state mirror are updated. After the verification fails, only the open anchor point sharding interval after the nearest common anchor point is partially reopened.

2. The edge collaborative processing method for production data based on the Industrial Internet as described in claim 1, characterized in that, The preprocessing of production data accessed by the Industrial Internet includes: Based on data collected from different sources, a unified production object identifier is mapped, and batch information is extracted from the current sampled data, work order data and local cache and bound to the corresponding production object identifier; A unified process status code is mapped, the original field name is mapped to a unified attribute identifier, the original sampling time is uniformly converted, and the attribute value is processed by status coding to form a standardized status record.

3. The edge collaborative processing method for production data based on the Industrial Internet as described in claim 1, characterized in that, The step of generating process anchor points and dividing anchor point intervals when a switch in the process state of the same production object is detected includes: The edge node maintains the current confirmed process status for each production object. When the process status code corresponding to the current sample is inconsistent with the current confirmed process status, the edge node enters the continuous confirmation process, determines that a process status switch has occurred, and generates a new process anchor point. The interval between the anchor point of the previous process and the anchor point of the current process is defined as a complete anchor point interval. The current process anchor point is used as the starting point of the new interval to establish the currently opened anchor point interval. Within the currently opened anchor point interval, when the previous state value and the subsequent state value corresponding to the current attribute are different, the corresponding difference piece is generated.

4. The edge collaborative processing method for production data based on the Industrial Internet as described in claim 1 or 3, characterized in that, The process of continuously adding differential fragments during weak network and network outage periods to form an unconfirmed anchor point interval index includes: Continuously monitor the link status with the central side, and distinguish between normal status, weak network status and network outage status based on the link status. In the weak network status, differential slices are allowed to be written to the local disk first and the confirmation and synchronization with the central side is delayed. In the offline state, the sending of real-time confirmation requests to the central side is stopped and differential shard appending processing is continuously performed locally. The edge node establishes an index of unconfirmed anchor point intervals and writes the index information corresponding to each anchor point interval that has not yet been confirmed by the central side into the unconfirmed anchor point interval index.

5. The edge collaborative processing method for production data based on the Industrial Internet as described in claim 1, characterized in that, The process of generating anchor point patches based on the unconfirmed anchor point interval index includes: Based on the index of the unconfirmed anchor point interval, locate the corresponding process anchor point chain segment and anchor point interval differential slice stack segment, and generate an anchor point seal for each unconfirmed anchor point interval. When the anchor point identifier of the current anchor point interval termination process is empty and the current open status mark is valid, the current anchor point interval is determined as the currently open anchor point interval, and the corresponding anchor point seal with the current open status mark is generated when the seal generation trigger condition is met. When the anchor point identifier of the current anchor point interval termination process is valid, the current anchor point interval is determined as a complete interval that has been closed, and the corresponding anchor point seal is generated first.

6. The edge collaborative processing method for production data based on the Industrial Internet as described in claim 1, characterized in that, The construction of a causal hypergraph by combining external reference relationships, process sequence, and anchor point inheritance relationships, and the division of anchor point patches into closed anchor point patches and open anchor point patches, includes: Using each anchor point as a node in the causal hypergraph, data dependency edges are established based on external reference relationships, sequential edges are established based on process sequence, and anchor point inheritance edges are established based on anchor point inheritance relationships to construct the causal hypergraph. By calculating the interval continuity coefficient and the proportion of external references, the anchor point inheritance consistency value is determined and the sealing closure judgment value is calculated. Based on the closure judgment condition, the anchor point sealing is divided into closed anchor point sealing and open anchor point sealing.

7. The edge collaborative processing method for production data based on the Industrial Internet as described in claim 1, characterized in that, The step of determining the nearest common anchor point from the common upstream anchor points already registered in the stable boundary index mirror and using it as the sole repair starting point includes: Starting with the currently unrepaired open anchor point fragment, trace back all upstream paths along the data dependency edges, sequence edges, and anchor point inheritance edges in the causal hypergraph. Each upstream anchor point in the upstream path set is matched with the registered anchor points in the stable boundary index mirror. Common upstream anchor points that appear in all upstream paths and are registered in the stable boundary index mirror are extracted, and the nearest common anchor point is determined. The nearest common anchor point is used as the unique repair starting point.

8. The edge collaborative processing method for production data based on the Industrial Internet as described in claim 7, characterized in that, The process of performing reverse rollback and canonical replay on open anchor point slicing based on the confirmed state image corresponding to the nearest common anchor point includes: A repair work status table is established based on the confirmed status mirror corresponding to the most recent common anchor point. The open anchor point patch from the most recent common anchor point to the current open anchor point patch and the corresponding differential patch constitute the current repair interval. The current repair interval is reversed and rolled back in reverse order from back to front. Starting with the confirmed status corresponding to the most recent common anchor point, perform standard replay on the current interval to be repaired in the order of first anchor point interval, then differential fragment, to generate repair anchor point fragment and repair account.

9. The edge collaborative processing method for production data based on the Industrial Internet as described in claim 1, characterized in that, The step of performing layered compensation submissions in the order of submitting the proof layer first and the difference layer last includes: The repair anchor point sealing and repair accounts are systematically organized to generate proof layers and difference layers; Perform aggregate digest processing on the proof layer and the difference layer respectively to obtain the proof layer digest value and the difference layer digest value corresponding to the current compensation batch. Combine the verification value of the previous completed and verified compensation batch, the current compensation batch identifier, and the encoded content corresponding to the nearest common anchor point identifier to perform concatenation and fixed-length one-way digest operation to obtain the current compensation batch verification value. Submit the proof layer and the current compensation batch identifier to the central side. After the proof layer structure consistency verification is passed, submit the difference layer and the current compensation batch verification value to the central side. The central side regenerates the corresponding compensation batch verification value based on the received proof layer and difference layer, and performs acceptance judgment on the current compensation batch.

10. The edge collaborative processing method for production data based on the Industrial Internet as described in claim 1, characterized in that, The step of updating the stable boundary index mirror and the confirmed state mirror after successful verification, and performing partial reopening only on the open anchor point sharding interval after the most recent common anchor point after failed verification, includes: When the acceptance judgment corresponding to the current compensation batch is verified as passed, the stable boundary index image and the confirmed status image are updated by performing a repair confirmation write-back update on each repair anchor point. When the acceptance judgment corresponding to the current compensation batch is a verification failure, the edge node takes the nearest common anchor point as the starting point of the local reopening boundary, performs local reopening only on the open anchor point sealing interval after the nearest common anchor point, and rewrites the anchor point interval differential fragments corresponding to the end of the current compensation batch after the nearest common anchor point into the queue to be repaired.

Images