A Cloud Computing-Based Collaborative Processing Method and System for Geological Exploration Data
By constructing a causal evidence chain and a dynamic residual envelope, anomalies can be accurately located and synchronization controlled, solving the problems of inaccurate anomaly location and discontinuous processing in existing technologies, and improving the accuracy and efficiency of collaborative processing of geological exploration data.
Patent Information
- Application Number
- CN202611285012.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-24
- Publication Date
- 2026-11-17
Smart Images

Figure CN123116068A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of distributed data processing technology in geological exploration, and in particular to a cloud computing-based collaborative processing method and system for geological exploration data. Background Technology
[0002] Geological exploration is a technical activity that involves obtaining information on underground geological structures, stratigraphic distribution, lithological characteristics, ore body morphology, and resource occurrence through methods such as drilling, well logging, geophysical exploration, geochemical exploration, remote sensing mapping, and geological logging. During geological exploration, various types of data are typically generated, including borehole column records, well logging curves, geophysical observation data, geochemical detection data, remote sensing images, geological maps, and sample analysis data. Due to the diverse sources, different formats, distinct spatial attributes, and large data volumes of these data, related technologies usually employ distributed data processing methods. Data is categorized and stored according to exploration projects, exploration areas, data categories, spatial coordinates, and acquisition batches, and multiple processing nodes perform data reading, format conversion, coordinate unification, content verification, data calculation, and result aggregation.
[0003] In existing technologies, cloud-based collaborative processing of geological exploration data typically includes processes such as uploading geological exploration data, file type identification, data content parsing, field mapping, coordinate transformation, task allocation, collaborative editing, version saving, and data merging. Specifically, the processing identifies borehole data tables, well logging data files, geophysical observation files, geochemical detection tables, remote sensing image files, and geological map data based on file extensions, header identifiers, field names, and record order. Borehole numbers, measurement point numbers, sample numbers, longitude, latitude, elevation, depth, stratigraphic codes, lithological codes, detection items, and detection values are mapped to preset data fields. Spatial coordinate transformation is performed based on coordinate system name, projection method, central meridian, zone number, and elevation datum. Simultaneously, data processing tasks are divided according to exploration block number, borehole number, survey line number, sampling batch, and processing procedure. Modification information of data records and field content by different participants is recorded. Finally, the processed data is organized and aggregated according to project number, data category, spatial range, and version number.
[0004] However, in the aforementioned distributed collaborative processing, different processing nodes typically determine whether data is abnormal based solely on processed field values, fixed tolerances, or processing logs. Operations such as coordinate unit transformations, axis order adjustments, parameter replacements, precision truncation, and field writing inherently generate varying degrees of normal numerical errors. Using a uniform tolerance makes it difficult to distinguish between normal processing errors and errors in parameter usage, field source, or node transmission. Furthermore, existing processing logs typically only record processing results or modification information, lacking a continuous correlation of inputs and outputs, parameter versions, field sources, and sequential call relationships for each processing action. When subsequent processing actions offset or cover previous anomalies, it becomes difficult to further determine the initial processing node and specific parameter actions where the anomaly occurred, leading to insufficient accuracy in anomaly localization and cause tracing.
[0005] Furthermore, when an anomaly is detected in a record or field, existing collaborative processing methods typically reprocess or halt synchronization as a whole by treating data fragments, files, or copies. This makes it difficult to limit the synchronization scope based on the spatial impact of the anomaly and the actual propagation of anomaly handling actions. This can easily lead to data that has not been affected by the anomaly and has already been verified being stopped from propagating or processed repeatedly. For anomaly fields, there is also a lack of control mechanisms for partial recovery based on trusted historical copies, re-establishing processing relationships, and re-verifying before resuming synchronization. This increases the possibility of anomaly data propagating to subsequent nodes and normal data being repeatedly calculated, affecting the continuity and efficiency of distributed collaborative processing of geological exploration data. Summary of the Invention
[0006] The main objective of this invention is to provide a cloud computing-based collaborative processing method and system for geological exploration data. By constructing a causal evidence chain of input spatial state and parameter actions, and combining dynamic residual envelopes to perform joint location of spatial first difference and causal first difference, the method further determines the first fault action and the scope of anomaly impact through positive sequence replay and counterfactual replay. Furthermore, it adopts field-level synchronization locking and trusted copy recovery mechanisms to solve the problems of inaccurate anomaly location, difficulty in tracing the cause of anomalies, and the tendency of local field anomalies to cause large-scale data synchronization stoppage or duplicate processing in existing distributed geological exploration data collaborative processing.
[0007] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A cloud computing-based collaborative processing method for geological exploration data includes: S1: Read the coordinates of boundary records and geological records within the data slice to be processed, determine the boundary anchor points, internal reference points, source data accuracy and input spatial status, and verify the continuity of node sequence number, copy version and voucher number; S2: Copy the data to be processed into a copy to be verified, break down the coordinate processing into parameter actions, register parameter snapshots and forward and reverse rules, register residual limits, scaling, axis order, and sequence as residual items, register input and output, unit, parameter version, field source, affected field, and previous voucher number as consecutive items, form a causal voucher chain, and write the output into the copy to be verified. S3: Restore the coordinates in reverse order according to the causal evidence chain, check the restored value with the input space state, form a residual envelope according to the source data precision and the residual term, determine the first node and field that exceeds the residual envelope as the spatial first difference node and failure field, determine the node corresponding to the first discontinuity of the continuous term as the causal first difference node, and take the node with the earlier node number as the final first difference node. S4: Replay the causal certificate chain corresponding to the final first difference node, verify the replay output, the output in the continuous item and the value of the copy field to be verified, determine the first failure action and the exception type, form a spatial exception set by the failure field, form a causal impact set by the impact field corresponding to the first failure action and the subsequent action that calls its output, and map it to a synchronous locking set. S5: If the synchronization latch set is empty, then synchronize the copy to be verified; if it is not empty, then stop the propagation of fields within the set and synchronize fields outside the set.
[0008] Preferably, S1 includes: S101. Read the boundary records of the data fragment to be processed and adjacent data fragments, and establish and arrange the boundary anchor points according to the unique record number, geological object number and boundary point number; S102. The geological records within the slice of the data to be processed that have non-common boundaries, adjacent record numbers, and valid topology are determined as the internal reference points. S103. Determine the accuracy of the source data based on the smallest resolution unit corresponding to the significant decimal places of the coordinates and the allowable error unit corresponding to the measurement accuracy identifier; S104. Verify the continuity of the node sequence number, data fragment number, copy parent version number and the previous synchronization certificate number. If continuous, register the coordinates, coordinate units, number of valid digits and field version number of the boundary anchor point and the internal reference point to form the input space state.
[0009] Preferably, S2 includes: S201. Copy the data fragments to be processed to form the copy to be verified. The processing of the coordinates is divided into parameter actions. The parameter actions are divided into unit transformation actions, coordinate axis reassignment actions, coordinate offset actions, parameter replacement actions, precision truncation actions and field writing actions. Parameter snapshots and forward and reverse rules are registered. S202. For the parameter action, the residual limit, scaling, axis sequence and order are registered as the residual item, the input and output, unit, parameter version, field source, affected field and previous voucher number are registered as the consecutive item, and the output of the previous parameter action is used as the input of the next parameter action in the order. S203. Verify the consecutive items of adjacent parameter actions in the order stated, and connect the parameter actions in series according to the preceding certificate number to form the causal certificate chain, and write the output of the last parameter action into the copy to be verified.
[0010] Preferably, S3 includes: S301. Based on the causal evidence chain, the forward and reverse rules are called in reverse order of action to restore the coordinates, and the restored horizontal coordinate, vertical coordinate and elevation are subtracted from the coordinates corresponding to the input spatial state, respectively. S302. Propagate the residuals in each coordinate direction according to the residual limit, scaling, axis order and sequence in the residual item, and synthesize the residual envelope with the source data precision. Determine the first node whose absolute value of the coordinate difference exceeds the corresponding residual envelope as the spatial first difference node, and determine the failure field. S303. According to the node sequence number, check the input and output, unit, parameter version, field source, influencing field and previous voucher number in the continuous item, determine the first non-continuous corresponding node as the causal first difference node, and determine the spatial first difference node and the node sequence number of the causal first difference node that is earlier as the final first difference node.
[0011] Preferably, the residual envelope is formed by propagating the residual limits of each parameter action according to the action sequence of the causal certificate chain. In this process, the preceding residual is converted according to the numerical scaling relationship, the residuals in the horizontal and vertical directions are exchanged according to the axis order mapping relationship, and the residuals generated by the coordinate offset action, the precision truncation action, and the field writing action are accumulated to the corresponding coordinate direction. The resulting accumulated residuals are combined with the precision of the source data to form the residual envelope.
[0012] Preferably, S4 includes: S401. Starting from the input space state corresponding to the final first difference node, replay the parameter actions in the forward order of the causal voucher chain, check the replay input, replay output, parameter version and action direction of each parameter action with the continuous items, and determine the first parameter action with inconsistent replay output as the first break action. S402. Verify the replay output of the first interruption action with the output in the successive items, the input of the next parameter action, and the value of the copy field to be verified, and determine the anomaly type by combining the parameter version and reverse recovery status. S403. Based on the failure field and record connection relationship, form the spatial anomaly set; based on the impact field of the first interruption action and the subsequent parameter action of calling its output, form the causal impact set; and map the spatial anomaly set and the causal impact set to the synchronization locking set according to the anomaly type.
[0013] Preferably, the initial failure action is replayed with the original parameters and replayed with the alternative parameters. The alternative parameter counterfactual replay only replaces the parameter values, parameter versions, action directions, or execution rules of the initial failure action. When the original parameter replay can reproduce the spatial anomaly of the failure field and the alternative parameter counterfactual replay can eliminate the spatial anomaly without generating a new spatial anomaly, the initial failure action is determined to be a causal anomaly action. When subsequent parameter actions cause the spatial anomaly to be canceled, the causal initial difference node remains unchanged.
[0014] Preferably, S5 includes: S501. When the synchronization latch set is empty, a synchronization license certificate is generated. The synchronization license certificate registers the task number, data fragment number, version number of the copy to be verified, version number of the parent copy, current node number, next node number, previous synchronization certificate number, and the last segment certificate number of the causal certificate chain. S502. Verify the registration content of the synchronization license credential with the copy to be verified, and verify the node transmission relationship represented by the current node number and the next node number. If they match, change the routing status of the copy to be verified from pending downstream verification to allowed synchronization status, and register the synchronization license credential as consumed. S503. When the synchronization locking set is not empty, a synchronization locking certificate is generated. Based on the synchronization locking set, the routing status of the fields within the set is set to stop propagation, and the routing status of the fields outside the set is set to allow synchronization.
[0015] Preferably, when the synchronization lockout set is not empty, the most recent copy with consumed synchronization license credentials before the final first difference node is determined as the trusted starting copy. Fields within the synchronization lockout set are restored in reverse order to their continuous state before the first break action, while fields outside the synchronization lockout set that have passed verification are retained to form a substitute copy. A causal credential chain is re-established for the substitute copy, and coordinates are restored in reverse order based on the re-established causal credential chain to form a residual envelope, determining the spatial first difference node and the causal first difference node. When at least one of the spatial first difference node and the causal first difference node exists, the node with the higher sequence number is determined as the re-determined node. The final first-loss node is determined by replaying the parameter actions in ascending order from the input space state corresponding to the newly determined final first-loss node. The newly determined first-loss action and exception type are determined. A spatial exception set is formed based on the failure field. A causal influence set is formed based on the influence field corresponding to the newly determined first-loss action and the subsequent parameter actions that call its output. The spatial exception set and the causal influence set are mapped to a newly determined synchronization lockout set. When neither the spatial first-loss node nor the causal first-loss node exists, the newly determined synchronization lockout set is determined to be empty. When the newly determined synchronization lockout set is empty, a synchronization license is generated.
[0016] This invention also discloses a cloud computing-based collaborative processing system for geological exploration data, comprising: The input verification module is used to read the boundary records and geological records within the data shards to be processed, determine the boundary anchor points, internal reference points, source data accuracy and input spatial status, and verify the continuity of node sequence number, copy version and voucher number. The certificate chain construction module is used to copy the data to be processed into a copy to be verified, break down coordinate processing into parameter actions, register parameter snapshots, forward and reverse rules, residual terms and continuous terms, form a causal certificate chain and write the output to the copy to be verified. The first difference positioning module is used to recover the coordinates in reverse order based on the causal voucher chain, determine the spatial first difference node and failure field according to the input spatial state, source data precision and residual terms, determine the causal first difference node according to the continuous terms, and determine the final first difference node. The anomaly mapping module is used to replay the causal evidence chain corresponding to the final first difference node, determine the first failure action and the anomaly type, form a spatial anomaly set based on the failure field, form a causal influence set based on the first failure action and the subsequent parameter actions of calling its output, and map it to a synchronous locking set. The synchronization control module is used to synchronize the copy to be verified when the synchronization latch set is empty, and to stop the propagation of fields within the set and synchronize fields outside the set when the synchronization latch set is not empty.
[0017] Compared with the prior art, the present invention has the following beneficial effects: This invention establishes a causal evidence chain of input spatial state and parameter actions simultaneously during distributed geological spatial data processing. It compares the coordinate recovery result with the residual envelope formed by the source data accuracy, single-action residual limits, numerical scaling relationships, axis sequence relationships, and action sequence. Combined with continuity term verification, it determines the spatial first-difference node and the causal first-difference node, thereby obtaining the final first-difference node. Compared to methods that judge anomalies solely based on fixed coordinate tolerances or the final output result, this invention avoids misjudging numerical differences caused by normal quantization, rounding, and precision truncation as anomalies. It also identifies latent anomalies that are partially offset by subsequent erroneous actions and not directly reflected in the final coordinates, thus improving the accuracy and reliability of anomaly localization in distributed geological spatial data.
[0018] This invention breaks down the processing steps, such as unit transformation, coordinate axis reassignment, coordinate offset, parameter replacement, precision truncation, and field writing, into separately executable and traceable parameter actions. Each parameter action is registered with its input / output, parameter version, field source, affected fields, forward / reverse rules, and preceding document number, forming a continuous causal document chain based on actual call relationships. After determining the final first-difference node, the first-difference action is located through forward-sequence replay. The causal relationship between the abnormal action and the spatial anomaly is confirmed by combining the original parameter replay and the counterfactual replay with the substituted parameters. This allows anomalies to be further refined from "data anomaly at a certain node" to specific parameter actions, specific field sources, or specific parameter states, improving the interpretability and verifiability of anomaly cause tracking, processing process verification, and subsequent result dispute tracing.
[0019] This invention forms a spatial anomaly set based on failed fields and spatial connectivity relationships, and a causal impact set based on the initial failure action and subsequent parameter actions of its output. Combined with anomaly type mapping, a synchronization locking set is obtained. Thus, when an anomaly occurs, only the propagation of the affected record fields is stopped, while fields outside the set that have already passed verification continue to synchronize. For locked fields, a new alternative copy can be restored to a continuous state from a trusted initial copy with valid synchronization permission credentials before the final first failure node, and a new causal credential chain is re-established, reverse recovery is performed, and synchronization verification is carried out. This avoids the entire data shard or the entire processing task from stalling due to a single field anomaly, while reducing the risks of anomalous data propagation, duplicate processing, and trusted results being overwritten by anomalous copies, thereby improving the continuity and synchronization efficiency of distributed geological exploration data collaborative processing. Attached Figure Description
[0020] Figure 1 A schematic diagram of the overall process of the cloud computing-based collaborative processing method and system for geological exploration data of this invention; Figure 2This is a schematic diagram illustrating the construction of the data fragmentation space verification object of the present invention; Figure 3 This is a schematic diagram illustrating the joint localization of the spatial first difference node and the causal first difference node in this invention; Figure 4 This is a schematic diagram of the first-break action confirmation and synchronous locking set mapping of the present invention. Detailed Implementation
[0021] The specific embodiments of the present invention will be further described below in conjunction with the technical solution of the present invention. It should be noted that the following embodiments are used to illustrate the cooperative relationship between the various technical features of the present invention and the implementation process of the corresponding technical means, and are not intended to limit the scope of protection of the present invention. Without changing the basic principles of spatial state verification, parameter action causal tracking, anomaly location, and replica synchronization control adopted in the present invention, those skilled in the art can adjust the specific data structure, field names, or storage locations according to the actual distributed computing framework, database type, coordinate data format, and node deployment method used.
[0022] like Figure 1 As shown in the figure, this diagram illustrates the overall processing flow of the present invention, starting from the input of data fragments to be processed, through input space state verification, establishment of causal credential chains, reverse recovery of first difference location, and positive sequence replay anomaly confirmation, and finally controlling the synchronization of the replicas to be verified according to the synchronization latch set.
[0023] The data shards to be processed described in this invention refer to data processing units obtained by cloud task nodes according to the block number, spatial partitioning relationship, or predetermined shard capacity of geological spatial data. A data shard to be processed may include multiple geological objects, each of which can be represented by points, lines, surfaces, or spatial objects formed by the sequential connection of multiple boundary points. Duplicate boundary records describing the same common boundary are allowed between different data shards. These duplicate boundary records are not considered data requiring repeated processing, but rather serve as a common spatial basis for spatial continuity verification between adjacent data shards.
[0024] The replica version number is used to identify different data states formed during distributed processing. A processing node receives a replica to be processed that has a corresponding parent replica version number. After completing the parameter actions of the current node, the processing node forms a new replica version to be verified. This replica to be verified is not directly released to subsequent nodes as a formal result replica before passing downstream space verification, causality verification, and synchronization permission verification.
[0025] The parameter action causal credential is not merely a log for recording processing history, but contains data items and rule identifiers required to re-execute and revoke the parameter action. When downstream nodes perform reverse recovery, current nodes perform forward replay, and abnormal recovery nodes generate alternative copies, they directly invoke the parameter action causal credential, without replacing the parameters actually registered in the credential with the current parameters in the runtime parameter library.
[0026] The residuals described in this invention are numerical differences allowed by normal numerical processing, such as parameter significant digits, numerical scaling, field quantization, rounding, truncation, or database write format conversion, when parameter actions are executed correctly according to predetermined rules. These residuals are not used to absorb errors in parameter usage, field source, node transmission, or parameter version. Even if the differences caused by the aforementioned anomalies fall within a certain fixed tolerance range, they are still identified through the continuity of the causal chain of parameter actions.
[0027] In one embodiment of the present invention, step S1 specifically includes the following process.
[0028] The cloud task node reads the project number, source data version number, block number, record unique number, geological object number, boundary point number, original x-coordinate, original y-coordinate, original elevation, coordinate unit, coordinate significant decimal places, measurement accuracy identifier, and field version number from the project source data copy, and forms data fragments to be processed according to the block number and the predetermined fragmentation capacity.
[0029] The cloud-based task nodes configure consecutive node numbers for each distributed processing node according to a predetermined processing flow, and generate a task invocation package when the task starts. The task invocation package includes at least the task number, data shard number, node number, name of the field to be processed, specified parameter version number, source data version number, and target replica area. The same data shard continues to use the same task invocation package after entering subsequent nodes, without reselecting the source data version, node order, or specified parameter version.
[0030] For each coordinate direction, the smallest resolvable unit is determined based on the significant decimal places of the coordinate, and the corresponding permissible error unit is obtained according to the measurement accuracy label. Different units are first converted to the same coordinate unit.
[0031] Let the coordinate direction be The smallest unit of resolution for this direction, determined by the number of significant decimal places, is The permissible error determined by the measurement accuracy specification is When both exist simultaneously, the accuracy of the source data in that direction... for: in, All use the same coordinate units.
[0032] If only one of the criteria exists, the existing one is used as the source data precision; if neither exists, the corresponding record is registered as having missing source precision, and that record is prevented from entering the coordinate processing chain that relies on the residual envelope for judgment. The x-coordinate, y-coordinate, and elevation determine the corresponding source data precision.
[0033] The source data precision only represents the basic precision that the source data itself is allowed to compare. When judging whether the coordinate inverse recovery is consistent, it is also necessary to combine the cumulative residual generated by the actual parameter actions to form the inverse recovery path residual envelope.
[0034] The task node reads the boundary records of the data fragment to be processed and its adjacent data fragments. Each boundary record includes at least a unique record number, a geological object number, a boundary point number, coordinates, adjacent preceding point number, adjacent following point number, and the data fragment number to which it belongs.
[0035] Two boundary records in adjacent data truncations are bound to the same boundary anchor point only if the record unique number, geological object number, and boundary point number are all consistent; if any identifier is inconsistent, the boundary identifier conflict status is registered and no anchor point correspondence is established.
[0036] Boundary anchor points on the same common boundary are arranged according to the connection relationship formed by the previous point number and the subsequent point number. When the subsequent point number of the current anchor point can correspond to the boundary point number of the next anchor point, the two are established in sequence; when there are records that cannot be continued, abnormally many predecessors, or abnormally many successors, the corresponding records are registered as boundary sequence broken, preventing them from entering the subsequent continuous boundary verification.
[0037] For geological records within a data tart that are not located at a common boundary, have adjacent record numbers, and are topologically valid, they are designated as internal reference points. Thus, boundary anchor points are used to verify the spatial continuity between adjacent data tarts, while internal reference points are used to verify the processing status within the same data tart; together, they constitute a set of spatial verification objects.
[0038] like Figure 2 As shown, this diagram represents the boundary anchor chain formed by verifying the consistency and connectivity of common boundary records of adjacent data shards, and determines the valid records of the internal topology of the data shards as internal reference points, thereby jointly forming a set of spatial verification objects.
[0039] After receiving the task call packet, the data fragment to be processed, the packet passed from the previous node, and the previous synchronization certificate, the current node verifies the continuity of node transmission before executing data processing.
[0040] Specifically, the previous node number should be equal to the current node number minus one; the data fragment number in the previous node's transmission packet should be consistent with the currently received data fragment number; the parent version number of the currently pending copy should be consistent with the version number of the copy to be verified output by the previous node; and the currently received previous synchronization credential number should be consistent with the credential number registered in the previous node's transmission packet.
[0041] When all the above conditions are met, the x-coordinate, y-coordinate, elevation, coordinate unit, field data type, field significant digits, and field version number of each boundary anchor point and internal reference point are read to form the input space status of the current node.
[0042] When node numbers are discontinuous, register a node cross-level status; when data shard numbers are inconsistent, register a shard mismatch status; when parent replica versions are discontinuous, register a replica lineage break status; when synchronization documents are discontinuous, register a document transmission break status. If any of the above statuses occur, the current node will refuse to execute parameter actions and update its local replica.
[0043] In another embodiment of the present invention, step S2 is performed after the aforementioned input verification is passed.
[0044] The current processing node replicates the data shards to be processed to form a verification copy, and breaks down the actual coordinate processing into parameter actions that can be executed and traced separately. Parameter actions include at least unit transformation actions, axis reassignment actions, coordinate offset actions, parameter replacement actions, precision truncation actions, and field writing actions. A processing node executes one or more of these actions based on the actual processing content.
[0045] The unit conversion action reads the input field value, input unit, target unit, and actual conversion parameters, calculates the output value according to the forward unit conversion rules, and registers the reverse rules for restoring the output unit to the input unit.
[0046] When the unit conversion relationship is input value multiplied by a conversion factor, the quantization error of the conversion factor and the rounding error of the output field serve as the sources of residuals for this action. Let the input value be , the upper bound of the conversion factor error be , and the upper bound of the output field rounding error be . Then the residual limit of the unit transformation action itself can be determined as: in, All coordinates are converted to the same units used in the output coordinates before calculation.
[0047] The coordinate axis reassignment action reads the axis sequence identifier and the source of the horizontal and vertical coordinate fields, and exchanges or maintains the field correspondence according to the axis sequence identifier. At the same time, it registers the reverse rule to restore the original field source relationship. When this action only changes the field correspondence without numerical calculation, its single action residual limit is zero, but the existing horizontal and vertical coordinate residuals are exchanged synchronously with the field source.
[0048] The coordinate offset action reads the eastward offset parameters, northward offset parameters, vertical offset parameters, and the direction of action of each offset parameter. When the direction of action is increasing, the positive rule executes the corresponding coordinate increase, and the negative rule executes the equivalent subtraction; when the direction of action is subtraction, the opposite processing is performed. The single-action residual limit is determined based on the minimum resolution unit of the offset parameter and the quantization error written into the output coordinate field. After unifying the coordinate units, these two values are taken as the normally permissible residual for this action.
[0049] The parameter replacement action compares the current parameter version with the parameter version specified in the task call package. If they are inconsistent and the current processing requires the use of the specified parameter version, the original parameter number, the replacement parameter number, the original parameter value, the replacement parameter value, the parameter unit, the parameter version, and the reason for replacement are recorded, along with the reverse rule for restoring the original parameter state.
[0050] When the replaced parameter is used in the calculation as an addition or subtraction, the upper bound of the parameter's error is used as the action residual in the corresponding coordinate direction; when it is used in the calculation as a scaling factor, the residual is propagated according to the scaling relationship of the input error to the factor; when other fixed calculation relationships are used, the error propagation relationship of the parameter is pre-registered in the corresponding parameter action rules.
[0051] The precision truncation action reads the field value before truncation, the target data type, the number of significant digits to retain, and the rounding method, generates the field value after truncation, and saves the value before truncation, the value after truncation, and the difference between the two.
[0052] For rounding, the maximum rounding error is determined based on the smallest resolution unit of the target field; for direct truncation, the maximum truncation error is determined based on the smallest resolution unit of the target field. The larger of the actual difference before and after truncation and the corresponding upper bound of the quantization error is used as the single-action residual limit for that action.
[0053] Precision truncation is an action that may not be uniquely recoverable through an inverse mathematical function; therefore, it is registered as a tolerance-reversible action. During inverse recovery verification, it is not required to uniquely deduce the original value from the truncated value. Instead, the state before truncation or the recovery reference state saved in the voucher is used for recovery, and the irreversible numerical loss caused by truncation is included in the residual envelope.
[0054] The field write action reads the output of the previous parameter action as its input value and writes it to the target field of the copy to be verified. The field write action also records the action input value, the actual stored value of the target field before writing, the actual stored value after writing, the target copy version number, and the field version number.
[0055] Among them, the action input value is used to maintain the continuity of actions in the causal certificate chain; the actual stored value before the target field is written is used to restore the overwritten data when the copy field needs to be revoked. The two are not treated as the same concept.
[0056] When no data type conversion or precision loss occurs during field writing, the single-action residual can be zero; when data type conversion, decimal place reduction, or field length limitation occurs, the corresponding quantization error is determined as the action residual limit based on the actual numerical conversion and truncation rules of the database.
[0057] Each parameter action forms an executable parameter action causal certificate. The causal certificate at least records a unique number, field name, action number, action sequence, input field source action number, input value, input unit, parameter name, actual parameter value, parameter unit, parameter version number, action direction, forward execution rule identifier, reverse recovery rule identifier, output value, output unit, reversibility type, single action residual limit, action impact record field set, preceding action certificate number, execution node number, and target copy version number.
[0058] Among them, the single-action residual limit, numerical scaling relationship, axis sequence relationship and action sequence constitute the residual item; input and output, input and output unit, parameter version, field source, action-affected field and preceding voucher number constitute the continuous item.
[0059] For the same record and the same processing field, the subsequent parameter action must directly use the output of the previous parameter action voucher as input, without rereading the same intermediate field in the copy to be verified. If the subsequent action requires the use of other external fields, those fields should be pre-registered as external input fields.
[0060] The current node checks the consecutive items of adjacent actions according to the action sequence. If the output of the previous action and the input of the next action exceed the accuracy range of the output of the previous action, register a numerical chain break; if the output unit of the previous action and the input unit of the next action are inconsistent and no unit conversion action is set, register a unit chain break; if the axis sequence is reassigned and the subsequent action calls the source of the field before the exchange without restoring it, register an axis sequence source rollback; if the old parameter version is called without a new parameter replacement certificate after parameter replacement, register a parameter version rollback; if the field used by the next action is neither the output of the previous action nor a registered external input, register an influence relationship chain break.
[0061] Each action is sequentially linked by the credential number of the preceding action to form a causal credential chain. If any chain breaks, no formal synchronization license is generated, but the input space state, the already formed causal credential chain, the broken chain location, and the copy to be verified are retained for subsequent verification.
[0062] After the parameter actions are executed, the node reads the output x-coordinate, output y-coordinate, elevation, unit, and field version number of the boundary anchor point and internal reference point to form the output spatial state, and establishes a field lineage vector for the target field. The field lineage vector records at least the field source copy version number, field parent version number, most recently passed node number, most recently acted credential number, whether the field has been restored, whether the field uses the current copy, and the causal credential end segment number.
[0063] The input space state, output space state, causal credential chain, field lineage vector, node number, data shard number, previous synchronization credential number, and the version number of the copy to be verified are all written into the node transmission packet. The copy to be verified is stored in the synchronization region, and its routing status is set to pending downstream verification.
[0064] In some embodiments of the present invention, step S3 is executed by the downstream node based on the causal certificate chain actually formed by the previous node.
[0065] Downstream nodes start from the last action in the causal certificate chain and call the reverse recovery rules item by item in the reverse order of the actions.
[0066] Field write actions are restored to their input state; when performing actual field rollback, the actual stored value of the target field before writing is restored. Precision truncation actions perform tolerance inverse recovery based on the state before truncation saved in the voucher, and include the unrecoverable portion in the action residual. Parameter replacement actions restore the original parameter values and versions; coordinate offset actions perform equivalent reverse processing according to the original action direction; coordinate axis reassignment actions restore the original field source; unit conversion actions restore the input unit state using the conversion parameters actually registered in the voucher.
[0067] During reverse recovery, the current parameter library is not called to reselect parameters, the action direction is not re-determined, and intermediate values that already exist in the causal certificate are not read from the copy to be verified.
[0068] Downstream nodes simultaneously propagate the motion residuals in each coordinate direction according to the residual term. Residuals involved in the calculation in the same coordinate direction are first unified to the same coordinate unit.
[0069] Let the cumulative residual in the coordinate direction before the i-th parameter takes effect be . The numerical scaling factor of this action on the preceding residual is: The limit value of the single-action residual of this action is The cumulative residual after the action is executed is: in, This is a dimensionless scaling factor.
[0070] For actions that do not change the numerical scale, =1. When the coordinate axis reassignment action is executed, the cumulative residuals of the x and y coordinates are swapped according to the axis order; the coordinate offset action adds the residuals caused by the precision of the offset parameters to the corresponding direction; the precision truncation and field writing actions add the corresponding quantized residuals to the actual field.
[0071] After all action residuals have been propagated, the cumulative residuals of the action chain are combined with the source data precision to form the inverse recovery path residual envelope. For coordinate directions... It can be determined using a conservative accumulation method: in, The three terms have the same dimension: the source data accuracy in the corresponding direction, the cumulative motion residual after propagating all parameter motion residuals, and the cumulative motion residual after propagation.
[0072] Therefore, the residual envelope varies with the actual action chain, parameter precision, field data type, and action order, rather than using the same fixed coordinate tolerance for all nodes.
[0073] After inverse reconstruction is completed, the reconstructed coordinates are compared with the input space state recorded by the previous node. Let the reconstructed coordinates be... The corresponding coordinates in the input space state are The coordinate difference is: in, All coordinates use the same unit corresponding to the coordinate direction.
[0074] when: When: the coordinate item is registered as restored to consistency; when: When an inconsistency is detected, it is recorded as a failure and the direction of failure, difference, residual envelope, and number of the last executed inverse recovery action are saved.
[0075] The spatial verification object passes when the x-coordinate, y-coordinate, and elevation of a boundary anchor point or internal reference point are all restored to consistency; the spatial verification object fails when any coordinate direction is not restored to consistency.
[0076] The system verifies each processing node sequentially according to the node number. When the previous node passes and the current node is the first to have a coordinate difference exceeding the corresponding residual envelope, the current node is identified as the spatial first difference node, and the corresponding coordinate field is identified as the failure field.
[0077] At the same time, the inputs and outputs, units, parameter versions, field sources, affected fields, and preceding document numbers in the causal voucher chain of each node are checked according to the node sequence number. The node where any consecutive item is broken for the first time is determined as the causal first difference node.
[0078] When the causal first difference node is earlier than the spatial first difference node, the causal first difference node is used as the final first difference node and registered as a latent anomaly; when the spatial first difference node is earlier than the causal first difference node, the spatial first difference node is used as the final first difference node; when the two are the same, the node is used as the final first difference node.
[0079] When only a spatial first difference node exists, the spatial first difference node is used as the final first difference node; when only a causal first difference node exists, the causal first difference node is used as the final first difference node; when neither exists, no abnormal final first difference node is formed.
[0080] Once the causal first difference node is determined, even if subsequent parameter actions use error handling in the opposite direction, causing the final coordinate difference to fall back into the residual envelope, the determined causal first difference node will not be revoked.
[0081] Please see Figure 3 This diagram illustrates the process of determining the spatial first difference node by reversing the recovery results and the residual envelope, determining the causal first difference node by the continuous terms of the causal voucher chain, and determining the final first difference node according to the node number.
[0082] In a further implementation, step S4 locates the first failure action based on the final first difference node and the failure field.
[0083] The system uses the final first difference node number, the failure boundary anchor point or internal reference point number, the failure field name, and the version number of the copy to be verified as joint location conditions to read the causal credential chain of the node.
[0084] Starting from the input space state registered at the final first difference node, the parameter actions are replayed in forward order. The first action uses the corresponding field value in the input space state, and subsequent actions directly use the output of the previous replayed action.
[0085] For each action, check the input of this replay and the input of the voucher registration in sequence, the parameter values and versions used this time, the direction of this action, and the output of the replay and the output of the voucher registration.
[0086] Discrete states, including parameter version number, action direction, field source, and voucher number, should be completely consistent; numerical inputs and outputs should be judged for consistency according to the data type, significant digits, and output precision registered for the action.
[0087] When the replay output of the current action is inconsistent with the voucher registration output for the first time, the action is determined as the first interruption action.
[0088] If the previous action can be correctly reproduced, but the input registered for the next action is not the output registered for the previous action, the exception type is determined to be an action continuity registration exception; if the actual parameter version used is inconsistent with the parameter version registered on the voucher, it is determined to be a parameter status exception; if the input, parameter, version, and direction are all consistent, but the reproduced output is inconsistent with the registered output, it is determined to be a parameter action execution exception.
[0089] If all intermediate actions can be correctly replayed, the final replay output is compared with the actual field values of the copy to be verified. If the replay output matches the voucher registration output but does not match the field values of the copy to be verified, it is determined to be a copy field write anomaly.
[0090] If the values of the forward replay, voucher registration output, and unverified copy fields are consistent, but the downstream reverse recovery still fails, check the corresponding reverse recovery rules and reverse voucher status to determine if the reverse recovery rule is abnormal or the reverse voucher status is abnormal.
[0091] If the internal actions of a node can be correctly replayed, but the replica version, node transmission packet, or synchronization credential does not correspond to the preceding node, it is determined to be a node transmission exception, credential transmission exception, or replica version transmission exception. When the same exception satisfies multiple states simultaneously, a primary exception type and accompanying exception types can be registered, with the exception that first causes the processing chain to lose continuity being designated as the primary exception.
[0092] For the initially determined initial action, the original parameter replay and the alternative parameter counterfactual replay are also performed.
[0093] The original parameter replay maintains the parameter values, parameter versions, action directions, and execution rules in the cause-effect vouchers unchanged, and is used to determine whether the cause-effect state can regenerate the discovered spatial anomalies.
[0094] Counterfactual reenactment of alternative parameters only changes the parameter values, parameter versions, action directions, or execution rules of the initial break action itself; actions before the initial break action and other subsequent actions remain unchanged. Alternative parameters are obtained from the parameter versions specified and verified by the task, the trusted parameter versions valid before the initial break action, or the verified modified parameter versions, and the source of alternative parameters is registered. Parameters that can pass verification are not searched through unconstrained traversal.
[0095] If the original parameter replay can reproduce the spatial anomaly, and the alternative parameter counterfactual replay can make the spatial anomaly disappear, and at the same time, it does not cause new anomalies to other previously passed boundary anchor points or internal reference points, the initial break action is confirmed as a causal anomaly action.
[0096] If the original parameter replay cannot reproduce the actual differences in the replica to be verified, further exception handling or node transmission exception handling is performed according to the replica field. Although the alternative parameters can eliminate the current failure, if they cause new failures on other spatially verified objects, no alternative synchronization permit is generated; only the counterfactual replay result is saved.
[0097] Subsequently, a spatial anomaly set is formed based on the failure fields and record connection relationships. When only one boundary anchor point fails and all surrounding anchor points and internal reference points pass through, only the failure record field of that anchor point is included in the spatial anomaly set; when multiple consecutive anchor points on the same boundary fail, the nearest recoverable passing anchor point is searched forward and backward along the anchor point chain, and the failure fields between the two passing anchor points are included in the set; when an internal reference point fails, the anomaly range is determined in the same way according to the connection relationships between the internal records.
[0098] The spatial anomaly range extends along the valid record connection relationship and stops when it encounters a recovery record, leaves the current valid connection chain, or encounters an invalid topological relationship, thus forming a spatial connection closure.
[0099] The system simultaneously reads the action impact record field set of the initial action and continues searching along the relationship between the output of the initial action and subsequent actions. The records and fields affected by subsequent actions that directly or indirectly use the output of the initial action constitute the causal impact set.
[0100] Based on the anomaly type, the spatial anomaly set and the causal influence set are mapped to the synchronous locking set.
[0101] For parameter action execution anomalies, action continuity registration anomalies, and parameter state anomalies, select the record field that simultaneously has evidence of spatial failure and the influence relationship of the abnormal action, and perform the necessary spatial connection closure according to the record connection relationship to form a synchronization lock set.
[0102] For write exceptions to replica fields, the record fields whose final replay output does not match the actual replica field values will be directly included in the synchronization latch set.
[0103] If the reverse recovery rule is abnormal or the reverse credential status is abnormal, the record fields affected by the corresponding reverse rule and the subsequent unverified copy record fields that depend on the abnormal reverse status will be included in the synchronization latch set.
[0104] In the event of node transmission anomalies, credential delivery interruptions, or parent replica version disconnections, starting from the replica where the interruption occurred, the replicas that have not yet obtained valid subsequent synchronization permissions are retrieved along the replica version relationship, and their affected record fields are used to form the propagation lockout scope.
[0105] When the same boundary anchor point in two adjacent data slices has coordinate differences in opposite directions, the spatial anomaly sets on both sides and the causal influence sets corresponding to axis order, offset direction, and parameter version are obtained respectively, and the partition connection locking range is formed according to the joint coverage relationship of the two sets.
[0106] Therefore, the synchronous blocking set is constrained by both the spatial anomaly range and the actual impact range of the initial interruption action and its propagation action.
[0107] like Figure 4 As shown, this diagram is used to represent the forward replay of the causal evidence chain starting from the final first difference node to determine the first broken action, to confirm the causal abnormal action through the replay of the original parameters and the counterfactual replay of the alternative parameters, and to form a synchronous locking set based on the spatial abnormality range and the actual impact range of the parameter action.
[0108] In a preferred embodiment of the present invention, step S5 performs replica synchronization control and abnormal field recovery based on the synchronization latch set.
[0109] A synchronization license is generated when the synchronization latch set is empty, the spatial state chain is continuous, the causal credential chain is continuous, the forward replay result is consistent with the credential registration output, the final replay output is consistent with the field of the copy to be verified, and the node transmission relationship is continuous.
[0110] The synchronization license certificate must at least register the task number, data shard number, version number of the copy to be verified, version number of the parent copy, current node number, next node number, previous synchronization certificate number, and the certificate number at the end of the causal certificate chain.
[0111] Before performing synchronization, the distributed storage nodes re-verify the above information against the actual state of the replica to be verified. If all information matches, the routing status of the replica to be verified is changed from pending downstream verification to synchronization allowed.
[0112] The state transition and the consumption of the synchronization license credential are completed within the same atomic state commit process. This can be implemented using distributed transactions, atomic updates with version conditions, or compare-and-swap methods. Before committing the state, it is confirmed that the copy to be verified is still in the state awaiting downstream verification, and the synchronization license credential is still in the unconsumed state. After successful commit, the copy's state is changed to "synchronization allowed," and the synchronization license credential is changed to "consumed."
[0113] Consumed synchronization licenses cannot be used to drive repeated synchronization of the same task, data shard, and copy version, thus avoiding repeated execution of coordinate offsets, unit transformations, or parameter replacements when nodes retry or communication resumes.
[0114] When the synchronization lock set is not empty, a synchronization lock certificate is generated. The synchronization lock certificate shall at least register the final first difference node number, the spatial first difference node number, the causal first difference node number, the first break action number, the exception field, the lock data fragment, the synchronization lock set, the trusted rollback copy number, the exception type, and the exception location certificate number.
[0115] The record field-level routing status is established for the replica to be verified. Each record has a unique number and field name, and has at least three states: pending verification, allowed synchronization, and stopped propagation.
[0116] Fields in the synchronization lockout set are set to the stopped propagation state; fields outside the set that have already been verified are set to the allowed synchronization state. When forming the data change set for the next node, only the records in the allowed synchronization state are extracted, while the stopped propagation fields are retained in the current verification area.
[0117] When the underlying distributed storage replicates records or data pages, the synchronization control layer can generate incremental field update packets based on the record field-level routing status and merge the fields with trusted records at the receiving node. This achieves the goal of stopping the propagation of fields within the synchronization lock set while fields outside the set continue to synchronize.
[0118] For each field in the synchronization latch set, look back from the final first-difference node to find the most recent copy that holds a consumed valid synchronization license credential and identify it as the trusted starting copy.
[0119] The system starts from the last action in the final lead node action chain and restores the synchronization latch fields in reverse order to the last continuous state before the lead break action. Field write actions restore the field values before writing; precision truncation actions obtain the undisturbed field values from a trusted start copy or trusted field lineage source; parameter replacement actions restore the original parameter state; coordinate offset actions cancel the corresponding offset; coordinate axis reassignment actions restore the original field source; and unit transformation actions restore the original field units.
[0120] When the synchronization latch set points to a single record field, only that field is restored; when it corresponds to a range of boundary records, only fields within that range that belong to the causal influence set are restored. Fields outside the synchronization latch set that have already been verified retain their values from the current copy to be verified.
[0121] The recovered data forms a new replacement copy without overwriting the original copy to be verified. The fields in the replacement copy respectively record the field source version number, field parent version number, field value before recovery, field value after recovery, recovery action number, trusted starting copy number, final first difference node number, and the causal voucher end segment number corresponding to the field.
[0122] The source of the abnormal field points to the trusted starting replica or the corresponding trusted field status; the unrecovered field is registered as using the field value of the current replica to be verified. The replacement replica obtains a new replica version number and establishes a new replica version relationship.
[0123] The replacement copy is resent to the processing location corresponding to the final primary node, without reusing the original synchronization permission status, cause / effect credential end number, or original credential consumption status.
[0124] The processing node re-executes the parameter actions from the trusted field state of the replacement replica, establishes a new causal credential chain using the verified parameter values, parameter units, parameter versions, and execution rules, and redetermines the single-action residual limit and action path residual envelope.
[0125] Subsequently, reverse order recovery, spatial first difference location, and causal first difference location are re-executed. If a spatial first difference node or a causal first difference node exists after re-verification, the node with the former index is determined as the newly determined final first difference node. Starting from the input spatial state corresponding to the newly determined final first difference node, forward order action replay, first-break action determination, anomaly type judgment, and mapping of spatial anomaly set and causal influence set are performed to form a newly determined synchronization locking set. If neither the spatial first difference node nor the causal first difference node exists after re-verification, a newly determined final first difference node is not formed, and the newly determined synchronization locking set is set to empty.
[0126] When the newly determined synchronization lock set is empty, a new synchronization permission certificate is generated for the alternative copy, and the local lock on the corresponding record field is released. If the alternative copy still has anomalies, the original synchronization lock state is retained, and the newly determined first difference node, first failure action, and counterfactual replay result are added to the anomaly location certificate, without expanding to the data fragments and fields that have already passed verification.
[0127] When a node task retryes, it first checks the processing status of existing tasks and data shards. Copies of already acquired and consumed synchronization licenses are not written again; abnormal copies of already formed synchronization latches are not propagated to subsequent nodes. During node migration or communication recovery, execution resumes from the most recently consumed synchronization license; parameter actions that have already been completed and verified are not executed again.
[0128] When an erroneous action causes an abnormal coordinate change, and a subsequent erroneous action cancels out this change in the opposite direction, the causal first difference node corresponding to the initial break point is maintained, provided that there has been a break in the input / output, parameter version, field source, influencing relationship, or preceding document relationship. Subsequent cancellation actions are registered as abnormal propagation actions and added to the causal influence set, without gaining synchronization permission even if the final coordinates fall back into the residual envelope.
[0129] During the results aggregation phase, the results aggregation node reads the task number, data shard number, trusted replica version number, parent replica version number, end node number, causal credential end segment number, and credential consumption status from the synchronization license credential of each data shard. Data shards with valid synchronization licenses and matching replica version relationships are written to the formal results area.
[0130] Data fragments in the synchronization lockout state are not marked as completed, nor are missing or abnormal data used to overwrite the final results; other data fragments that have passed synchronization verification continue to complete the result aggregation.
[0131] In the event of a dispute over subsequent results, the corresponding boundary anchor point or internal reference point can be located by the unique record number, field name, and copy version number. Then, the spatial first difference node, causal first difference node, final first difference node, and first break action can be obtained in sequence. The input value, actual parameter value, parameter version, forward execution rule, reverse recovery rule, voucher registration output, replay output, actual copy field value, action impact set, and synchronization lockout set of the first break action can be obtained, thereby forming a continuous abnormal handling evidence relationship.
[0132] In another embodiment of the present invention, a distributed geological spatial data anomaly location and copy synchronization control system is also provided for performing the aforementioned method.
[0133] The system includes at least an input verification module, a credential chain construction module, a first difference positioning module, an anomaly mapping module, and a synchronization control module.
[0134] The input verification module is used to read the boundary records and geological record coordinates within the data fragments to be processed. It establishes boundary anchor points based on the unique record number, geological object number, and boundary point number. It determines internal reference points based on the connection relationship of records within the fragment. It determines the source data accuracy based on the effective decimal places of the coordinates and the measurement accuracy identifier. It also verifies the node sequence number, data fragment number, copy parent version number, and the previous synchronization certificate number to form the input space status.
[0135] The certificate chain construction module is used to copy the data to be processed into fragments to form a copy to be verified. It breaks down coordinate processing into parameter actions, registers parameter snapshots, forward execution rules, reverse recovery rules, residual items and continuous items, so that the output of the previous action can be directly used as the input of the next action. It forms a causal certificate chain through the certificate number of the preceding action, and writes the output of the last action into the copy to be verified.
[0136] The first difference positioning module is used to recover coordinates in reverse order according to the causal voucher chain. It forms the cumulative residual of the action path based on the single action residual limit, scaling relationship, axis sequence relationship and action sequence, and combines it with the source data accuracy to form the reverse recovery path residual envelope. Based on the recovery result, it determines the spatial first difference node and failure field, determines the causal first difference node based on the continuous items, and determines the final first difference node according to the node sequence number.
[0137] The anomaly mapping module is used to replay the causal evidence chain in ascending order, starting from the spatial state input by the final first-failure node. It verifies the replay input, replay output, parameter version, action direction, and field values of the copy to be verified to determine the first failure action and anomaly type. It forms a spatial anomaly set based on the failure field and record connection relationship, forms a causal influence set based on the first failure action and the subsequent parameter actions that call its output, and forms a synchronization locking set according to the anomaly type mapping. It can also perform original parameter replay and alternative parameter counterfactual replay.
[0138] The synchronization control module is used to generate synchronization permission credentials and control the copy to be verified to enter the synchronization allowed state when the synchronization lock set is empty, generate synchronization lock credentials when the synchronization lock set is not empty, stop the propagation of fields within the set and synchronize fields outside the set according to the record field-level routing status; it is also used to recover abnormal fields based on the trusted starting copy and field lineage, form alternative copies and make the alternative copies re-enter the causal credential establishment, reverse recovery, first difference location, action replay and synchronization verification process.
[0139] Through the above structure, data fragments, input spatial status, output spatial status, causal credential chain, residual envelope, field lineage information, and synchronization credentials are sequentially transmitted between distributed processing nodes. This enables the spatial anomaly location results to be further mapped to specific nodes and specific parameter actions, and controls the synchronization of record field-level copies based on the spatial anomaly range and the impact range of parameter actions.
[0140] The present invention is further disclosed below with reference to specific embodiments: In a specific embodiment of the present invention, taking the common boundary data processing of adjacent data shards FA and FB as an example, the abnormal location and replica synchronization control process when the coordinate field source is abnormal during distributed processing is described.
[0141] Please refer to Figure 2The cloud processing task number is TASK-20260807-01. The data shard to be processed is numbered FA, and the adjacent data shard is numbered FB. The units for the x-axis, y-axis, and elevation are all meters. The common boundary between data shards FA and FB includes records R101, R102, and R103, with corresponding boundary points P101, P102, and P103 respectively. The unique record number, geological object number, and boundary point number of the three boundary points can correspond between FA and FB, and the preceding and following point numbers are consecutive, thus forming a boundary anchor chain. Record R201 in data shard FA is not located on the common boundary, but the numbers of its preceding and following adjacent records exist and are topologically valid; therefore, it is used as an internal reference point.
[0142] This embodiment uses the anomaly recorded R102 as an example. The input coordinates of R102 are X=512.364m, Y=489.217m, Z=126.438m. The coordinate fields are rounded to three decimal places, corresponding to a minimum resolution of 0.001m. The measurement accuracy indicator corresponds to an allowable error of 0.005m. Therefore, the accuracy of its source data is: The current processing node is N2, the previous processing node is N1, the current data shard number is FA, the parent replica version number is V-FA-101, and the previous synchronization certificate number is SC-N1-FA-01. Upon verification, the node sequence number, data shard number, parent replica version, and previous synchronization certificate are all consecutive; therefore, the above coordinates are fixed as the input space state of N2.
[0143] Since the axis sequence identifier of this batch of data is YX, N2 sequentially executes the coordinate axis reassignment action A201, coordinate offset action A202, precision truncation action A203, and field writing action A204.
[0144] A201 swaps the source of the x and y coordinate fields, and its correct output is X=489.217m and Y=512.364m. A202 uses parameter version PV-2026.04 and performs an eastward offset of +1.2504m on the reassigned x coordinate. Therefore, when the action chain is normal and continuous, A202 should directly call the 489.217m output by A201, resulting in: X=489.217+1.2504=490.4674.
[0145] After retaining 3 decimal places in A203, we should get 490.467m, which is then written to the copy to be verified by A204.
[0146] During actual execution, A202 did not call the output of A201, 489.217m, but instead reread the original x-coordinate of 512.364m before the axis sequence swap. Therefore, the actual output of A202 is: X = 512.364 + 1.2504 = 513.6144 m.
[0147] After processing by A203, 513.614m is obtained, and then written to the copy to be verified by A204. At this time, the coordinates of R102 in the copy to be verified are X=513.614m, Y=512.364m, Z=126.438m.
[0148] N2 forms a causal evidence chain of A201→A202→A203→A204. Among them, the horizontal coordinate output registered by A201 is 489.217m, while the input registered by A202 is 512.364m. The two are inconsistent, and A202 did not register the action of restoring the field source before the axis sequence exchange. Therefore, A202 is marked as an axis sequence source rollback action.
[0149] In this action chain, A201 only changes the field correspondence, and its single-action residual limit is 0; A202 uses an offset parameter with 4 decimal places, and its parameter quantization error upper bound is 0.00005m; A203 uses rounding with 3 decimal places, and its single-action residual limit is 0.0005m; A204 does not generate further quantization loss, and its single-action residual limit is 0. The residuals generated by A202 and A203 are applied to the X field after reassignment by A201, corresponding to the Y direction of the input space after inverse recovery of the axis sequence mapping through A201. Therefore, the cumulative residual of the action path corresponding to the Y direction of the input space is: E Y =0.00005+0.0005=0.00055m.
[0150] Based on the source data precision in the Y direction, Y = 0.005m, the residual envelope of the inverse recovery path in the Y direction is obtained: B Y =0.005+0.00055=0.00555m.
[0151] Downstream nodes perform reverse recovery in the order of A204, A203, A202, and A201. A204 is first recovered to its action input value of 513.614m. A203 is then recovered to 513.6144m based on the pre-truncation state saved in the voucher. Subsequently, the 1.2504m eastward offset of A202 is cancelled, resulting in 512.364m. Then, the horizontal and vertical coordinates are recovered according to the original field source relationship registered by A201, resulting in: Where and are the x-coordinate and y-coordinate after inverse recovery, respectively.
[0152] Compared with the N2 input space state, the difference in the horizontal axis is 0, and the difference in the vertical axis is: ∣512.364−489.217∣=23.147 m.
[0153] Since 23.147 is greater than the residual envelope of the corresponding Y-direction inverse recovery path, the ordinate recovery of R102 fails. In this embodiment, the field source backtracking of A202 only occurs in R102. R101, R103, and the internal reference point R201 all use the output of A201 as the input for subsequent corresponding parameter actions. Their causal evidence chains are continuous, and the coordinate differences obtained according to the same inverse recovery rule do not exceed their respective residual envelopes. Therefore, N2 is determined as the spatial first difference node.
[0154] Meanwhile, in the causal continuity check, the output of A201 (489.217m) and the input of A202 (512.364m) are the first discontinuities, therefore N2 is also the first causal node. Since the spatial first node is the same as the causal first node, N2 is the final first node.
[0155] The system begins its forward replay from the input space state of N2. After re-executing A201, X=489.217m and Y=512.364m are obtained, consistent with the voucher registration output. When continuing to replay A202, the output 489.217m from A201 should be used directly, therefore... ; The output of document A202 registration is 513.6144m, and the difference between the two is: =∣490.4674−513.6144∣=23.1470 m.
[0156] A202 thus became the first parameter action whose replay result was inconsistent with the registration status, and it was identified as the first interruption action. The anomaly type was continuous action registration anomaly, specifically manifested as the field source rollback after the coordinate axis was reassigned.
[0157] During further causal confirmation, the original parameter replay maintains the parameter values, versions, action directions, and execution rules of A202 at the time of the anomaly. Following this execution rule, 512.364m is retrieved from the original x-coordinate field before the axis sequence swap as input, yielding 513.6144m again, ultimately reproducing the 513.614m in the copy to be verified. In the alternative parameter counterfactual replay, only the rule used to determine the source of the action input field in A202's execution rule is modified to directly call the 489.217m output by A201, keeping the offset parameter, version, action direction, and other parameters unchanged. A202 then outputs 490.4674m, and the final written value is 490.467m. This alternative execution rule eliminates the spatial anomaly of R102 without causing new spatial anomalies in other records that had already passed spatial verification; therefore, A202 is confirmed as a causal anomaly action.
[0158] The failure of reverse recovery in R102 manifests in the Y direction of the input spatial state; since A201 performed an XY axis sequence swap, this Y direction corresponds to the X field in the current copy to be verified. After mapping along the field lineage, the spatial anomaly field points to the current X field of R102. A202 and its subsequent A203 and A204 also act on this X field, therefore the spatial anomaly range and the causal influence range both point to the X field of record R102, forming a synchronization latch set: Therefore, only the X field of R102 is set to the stopped propagation state; the Y and Z fields of R102, as well as the fields in R101, R103 and R201 that have already been verified, continue to be allowed to synchronize.
[0159] The system determines V-FA-101 as the trusted starting copy from the most recent copy that has obtained and consumed a valid synchronization license credential before N2, and obtains the trusted state of R102 from it: X=512.364m, Y=489.217m, Z=126.438m.
[0160] When N2 processing is re-executed, after swapping the axis sequence, A201 obtains X=489.217m and Y=512.364m. A202 executes according to the correct field source after verification: X=489.217+1.2504=490.4674 m.
[0161] After processing by A203 and A204, the coordinates of R102 in the replacement copy are X=490.467m, Y=512.364m, and Z=126.438m. This replacement copy obtains a new copy version number V-FA-102-R1 and re-establishes the causal certificate chain from A201-R1 to A204-R1.
[0162] Downstream nodes perform reverse recovery on the replacement replica again. A204-R1 first recovers to its action input value of 490.467m, then A203-R1 recovers to 490.4674m based on the pre-truncation state saved in the voucher. Subsequently, A202-R1 cancels the 1.2504m offset, obtaining 489.217m, and then restores the axis sequence relationship of A201-R1, resulting in: The difference between its horizontal and vertical coordinates is 0, and the path residual envelope does not exceed 0.00555m; in the new causal voucher chain, the input and output of each action, the parameter version, the field source and the relationship of the preceding voucher are also continuous.
[0163] Therefore, the redefined set of synchronous locking mechanisms is empty.
[0164] The system generates a new synchronization license credential SC-N2-FA-R1 for the replacement copy V-FA-102-R1. After verifying that the copy version, parent version, node number, previous synchronization credential, and the last segment number of the causal credential are consistent, the system changes the replacement copy from the pending downstream verification state to the allowed synchronization state, and at the same time registers the synchronization license credential as consumed.
[0165] In this embodiment, the anomaly is first located at node N2, and further located to parameter action A202, limiting the scope of synchronization locking to the X field of record R102; after field-level recovery and the formation of a replacement copy, the subsequent copy synchronization of the field is restored by re-verifying the spatial state and causal state.
[0166] Finally, it should be understood that the embodiments described in this specification are merely illustrative of the principles of the embodiments described herein. Other variations may also fall within the scope of this specification. Therefore, alternative configurations of the embodiments described herein are intended to be illustrative rather than limiting, and should be considered consistent with the teachings of this specification. Accordingly, the embodiments described herein are not limited to those explicitly introduced and described herein.
Claims
1. A cloud computing-based collaborative processing method for geological exploration data, characterized in that, include: S1: Read the coordinates of boundary records and geological records within the data slice to be processed, determine the boundary anchor points, internal reference points, source data accuracy and input spatial status, and verify the continuity of node sequence number, copy version and voucher number; S2: Copy the data to be processed into a copy to be verified, break down the coordinate processing into parameter actions, register parameter snapshots and forward and reverse rules, register residual limits, scaling, axis order, and sequence as residual items, register input and output, unit, parameter version, field source, affected field, and previous voucher number as consecutive items, form a causal voucher chain, and write the output into the copy to be verified. S3: Restore the coordinates in reverse order according to the causal evidence chain, check the restored value with the input space state, form a residual envelope according to the source data precision and the residual term, determine the first node and field that exceeds the residual envelope as the spatial first difference node and failure field, determine the node corresponding to the first discontinuity of the continuous term as the causal first difference node, and take the node with the earlier node number as the final first difference node. S4: Replay the causal certificate chain corresponding to the final first difference node, verify the replay output, the output in the continuous item and the value of the copy field to be verified, determine the first failure action and the exception type, form a spatial exception set by the failure field, form a causal impact set by the impact field corresponding to the first failure action and the subsequent action that calls its output, and map it to a synchronous locking set. S5: If the synchronization latch set is empty, then synchronize the copy to be verified; if it is not empty, then stop the propagation of fields within the set and synchronize fields outside the set.
2. The method according to claim 1, characterized in that, S1 includes: S101. Read the boundary records of the data fragment to be processed and adjacent data fragments, and establish and arrange the boundary anchor points according to the unique record number, geological object number and boundary point number; S102. The geological records within the slice of the data to be processed that have non-common boundaries, adjacent record numbers, and valid topology are determined as the internal reference points. S103. Determine the accuracy of the source data based on the smallest resolution unit corresponding to the significant decimal places of the coordinates and the allowable error unit corresponding to the measurement accuracy identifier; S104. Verify the continuity of the node sequence number, data fragment number, copy parent version number and the previous synchronization certificate number. If continuous, register the coordinates, coordinate units, number of valid digits and field version number of the boundary anchor point and the internal reference point to form the input space state.
3. The method according to claim 1, characterized in that, S2 includes: S201. Copy the data fragments to be processed to form the copy to be verified. The processing of the coordinates is divided into parameter actions. The parameter actions are divided into unit transformation actions, coordinate axis reassignment actions, coordinate offset actions, parameter replacement actions, precision truncation actions and field writing actions. Parameter snapshots and forward and reverse rules are registered. S202. For the parameter action, the residual limit, scaling, axis sequence and order are registered as the residual item, the input and output, unit, parameter version, field source, affected field and previous voucher number are registered as the consecutive item, and the output of the previous parameter action is used as the input of the next parameter action in the order. S203. Verify the consecutive items of adjacent parameter actions in the order stated, and connect the parameter actions in series according to the preceding certificate number to form the causal certificate chain, and write the output of the last parameter action into the copy to be verified.
4. The method according to claim 3, characterized in that, S3 includes: S301. Based on the causal evidence chain, the forward and reverse rules are called in reverse order of action to restore the coordinates, and the restored horizontal coordinate, vertical coordinate and elevation are subtracted from the coordinates corresponding to the input spatial state, respectively. S302. Propagate the residuals in each coordinate direction according to the residual limit, scaling, axis order and sequence in the residual item, and synthesize the residual envelope with the source data precision. Determine the first node whose absolute value of the coordinate difference exceeds the corresponding residual envelope as the spatial first difference node, and determine the failure field. S303. According to the node sequence number, check the input and output, unit, parameter version, field source, influencing field and previous voucher number in the continuous item, determine the first non-continuous corresponding node as the causal first difference node, and determine the spatial first difference node and the node sequence number of the causal first difference node that is earlier as the final first difference node.
5. The method according to claim 4, characterized in that, The residual envelope is formed by propagating the residual limits of each parameter action according to the action sequence of the causal certificate chain. Specifically, the preceding residual is converted according to the numerical scaling relationship, the residuals in the horizontal and vertical directions are exchanged according to the axis order mapping relationship, and the residuals generated by the coordinate offset action, the precision truncation action, and the field writing action are accumulated to the corresponding coordinate direction. The resulting accumulated residuals are combined with the precision of the source data to form the residual envelope.
6. The method according to claim 1, characterized in that, S4 includes: S401. Starting from the input space state corresponding to the final first difference node, replay the parameter actions in the forward order of the causal voucher chain, check the replay input, replay output, parameter version and action direction of each parameter action with the continuous items, and determine the first parameter action with inconsistent replay output as the first break action. S402. Verify the replay output of the first interruption action with the output in the successive items, the input of the next parameter action, and the value of the copy field to be verified, and determine the anomaly type by combining the parameter version and reverse recovery status. S403. Based on the failure field and record connection relationship, form the spatial anomaly set; based on the impact field of the first interruption action and the subsequent parameter action of calling its output, form the causal impact set; and map the spatial anomaly set and the causal impact set to the synchronization locking set according to the anomaly type.
7. The method according to claim 6, characterized in that, For the initial failure action, perform original parameter replay and alternative parameter counterfactual replay respectively. The alternative parameter counterfactual replay only replaces the parameter value, parameter version, action direction, or execution rule of the initial failure action. When the original parameter replay can reproduce the spatial anomaly of the failure field, and the alternative parameter counterfactual replay can eliminate the spatial anomaly without generating a new spatial anomaly, the initial failure action is determined to be a causal anomaly action. When subsequent parameter actions cause the spatial anomaly to be canceled, the causal initial difference node remains unchanged.
8. The method according to claim 5, characterized in that, S5 includes: S501. When the synchronization latch set is empty, a synchronization license certificate is generated. The synchronization license certificate registers the task number, data fragment number, version number of the copy to be verified, version number of the parent copy, current node number, next node number, previous synchronization certificate number, and the last segment certificate number of the causal certificate chain. S502. Verify the registration content of the synchronization license credential with the copy to be verified, and verify the node transmission relationship represented by the current node number and the next node number. If they match, change the routing status of the copy to be verified from pending downstream verification to allowed synchronization status, and register the synchronization license credential as consumed. S503. When the synchronization locking set is not empty, a synchronization locking certificate is generated. Based on the synchronization locking set, the routing status of the fields within the set is set to stop propagation, and the routing status of the fields outside the set is set to allow synchronization.
9. The method according to claim 8, characterized in that, When the synchronization lock set is not empty, the most recent copy with consumed synchronization license credentials before the final first difference node is determined as the trusted starting copy. The fields in the synchronization lock set are restored in reverse order to the continuous state before the first break action, and the fields outside the synchronization lock set that have passed verification are retained to form a substitute copy. A causal credential chain is re-established for the substitute copy. The coordinates are restored in reverse order based on the re-established causal credential chain to form a residual envelope, and the spatial first difference node and the causal first difference node are determined. When at least one of the spatial first difference node and the causal first difference node exists, the node with the higher index is determined as the re-determined final first difference node. The parameter actions are replayed in forward order starting from the input spatial state corresponding to the re-determined final first difference node to determine the re-determined first break action and the exception type. A spatial exception set is formed based on the failure field. A causal influence set is formed based on the influence fields corresponding to the re-determined first break action and the subsequent parameter actions that call its output. The spatial exception set and the causal influence set are mapped to the re-determined synchronization lock set. When neither the spatial first difference node nor the causal first difference node exists, the redefined synchronization locking set is set to empty; when the redefined synchronization locking set is empty, a synchronization license is generated.
10. A cloud computing-based collaborative processing system for geological exploration data, characterized in that, For performing the method according to any one of claims 1-9, comprising: The input verification module is used to read the boundary records and geological records within the data shards to be processed, determine the boundary anchor points, internal reference points, source data accuracy and input spatial status, and verify the continuity of node sequence number, copy version and voucher number. The certificate chain construction module is used to copy the data to be processed into a copy to be verified, break down coordinate processing into parameter actions, register parameter snapshots, forward and reverse rules, residual terms and continuous terms, form a causal certificate chain and write the output to the copy to be verified. The first difference positioning module is used to recover the coordinates in reverse order based on the causal voucher chain, determine the spatial first difference node and failure field according to the input spatial state, source data precision and residual terms, determine the causal first difference node according to the continuous terms, and determine the final first difference node. The anomaly mapping module is used to replay the causal evidence chain corresponding to the final first difference node, determine the first failure action and the anomaly type, form a spatial anomaly set based on the failure field, form a causal influence set based on the first failure action and the subsequent parameter actions of calling its output, and map it to a synchronous locking set. The synchronization control module is used to synchronize the copy to be verified when the synchronization latch set is empty, and to stop the propagation of fields within the set and synchronize fields outside the set when the synchronization latch set is not empty.
