Version chain organization and difference tracking method for mooring simulation multi-wheel test data
By creating a version chain and difference tracking method in mooring simulation experiments, the problem of lineage management of multi-round experimental data was solved, achieving efficient difference location and experiment review, and supporting the queryability and reusability of experimental results.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHENZHEN INTELLIGENT OCEAN ENG INNOVATION CENT CO LTD
- Filing Date
- 2026-04-29
- Publication Date
- 2026-07-31
AI Technical Summary
Existing technologies cannot effectively manage the lineage relationships between data from multiple mooring simulation tests, resulting in data silos and inefficient difference analysis, making it difficult to quickly locate the changing objects and time periods.
By creating node information for the current version and associating it with the parent version, calculating object-level difference scores, filtering out abnormal objects, performing time-window-level sliding window comparisons, forming a version chain, and storing difference records, explicit lineage management of multi-round experimental results is achieved.
It achieves a clear version chain organization of multi-round test results, improves the efficiency and accuracy of difference localization, makes test review and experience reuse possible, and reduces the overhead of full comparison.
Smart Images
Figure CN122489518A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the fields of industrial Internet of Things, time-series data management and simulation data organization, and in particular to a method for organizing version chains and tracking differences for multi-round test data in mooring simulation. Background Technology
[0002] In mooring simulation scenarios, a complete test is often not finished after one run, but rather involves continuous parameter tuning, environmental boundary adjustment, model version replacement, and iterative combination of operating conditions around the baseline model. Each test run outputs continuous time-series data with multiple objects, multiple indicators, and long time periods, such as cable tension, hull displacement, attitude changes, wind, wave and current parameters, stage labels, and model version information.
[0003] Current processing methods typically store the results of each round in a separate file, table, or directory. While this method achieves storage, it lacks a natural parent-child relationship between rounds, making it difficult to directly explain which round the changes originated from, what changes were made, or the differences between the previous and current rounds. As the number of experimental rounds increases, data silos easily form.
[0004] In actual analysis, engineers often need to export multiple rounds of complete curves first, and then compare them segment by segment according to object and time period. This approach has three obvious problems: First, the comparison often starts from the full original curves, which incurs significant reading and transmission overhead; second, parameter changes, model changes, and result changes are usually scattered across different records, requiring manual piecing together during review; third, even if it is known that the results of two rounds are "very different," it is difficult to quickly explain which object changed, which indicator changed, and from which time period the change began. In other words, the current processing method cannot directly manage the results of multiple rounds of experiments into version assets with explicit lineage, layer-by-layer comparison, and sustainable traceability. Summary of the Invention
[0005] This invention provides a version chain organization and difference tracking method for multi-round test data in mooring simulation. It addresses the shortcomings of existing technologies that cannot directly manage multi-round test results as version assets with explicit lineage, layer-by-layer comparison, and sustainable traceability. This method enables the formation of a clear version chain upon entry into the database of multi-round test results, and allows for the step-by-step location of differences at the version, object, and time window levels, thereby supporting test review, comparative analysis, and experience reuse.
[0006] This invention provides a method for organizing version chains and tracking differences for multi-round test data in mooring simulation, comprising the following steps.
[0007] For each round of received raw time series data, create the current node information for the current version; according to the preset organization rules, store the raw time series data, current node information, and current summary result information of the current version into the preset time series database; wherein, the raw time series data is the time series data of the mooring simulation test; the current node information includes: the current version identifier, the parent version identifier, and the set of changes of the current version relative to the parent version; If the parent version identifier of the current version is not empty, calculate the object-level difference score of the current version relative to the parent version based on the current summary result information and the parent summary result information; Based on object-level difference scores, abnormal objects are screened to obtain candidate abnormal objects; based on the original time series data, the abnormal time series data corresponding to the candidate abnormal objects are determined. Based on the abnormal time series data, a sliding window comparison at the time window level is performed to obtain the abnormal time window information of the current version relative to the parent version; based on the object-level difference score and the abnormal time window information, the relationship difference record is determined; the original time series data, the current node information, the current summary result information and the relationship difference record are stored in the preset time series database.
[0008] According to the version chain organization and difference tracking method for multi-round test data in mooring simulation provided by the present invention, before calculating the object-level difference score of the current version relative to the parent version, the method further includes: Based on object identifier, indicator identifier and stage label, the current node information and parent node information are aligned to obtain the aligned current node information and aligned parent node information. Based on the sampling frequency, data duration, or stage boundary, if the information of the current node and the information of the parent node are inconsistent, the time series data corresponding to the information of the current node and the time series data corresponding to the information of the parent node are resampled separately; or, the time series data corresponding to the information of the current node and the time series data corresponding to the information of the parent node are aggregated into a unified time base.
[0009] According to the present invention, a version chain organization and difference tracking method for multi-round test data in mooring simulation is provided, which calculates the object-level difference score of the current version relative to the parent version based on the current summary result information and the parent summary result information, including: Based on the current summary result information and the parent summary result information, for each summary statistics item to be compared in the aligned current data and aligned parent data, calculate the absolute difference of the current version relative to the parent version in the summary statistics item; By using a preset normalization function, the absolute differences of multiple summary statistics are normalized to obtain object-level difference scores.
[0010] According to the present invention, a version chain organization and difference tracking method for multi-round test data in mooring simulation is provided. Based on object-level difference scoring, anomaly objects are screened to obtain candidate anomaly objects, including: If the object-level difference score is not less than the preset object-level difference threshold, the object identifier and indicator identifier corresponding to the object-level difference score are marked as candidate abnormal objects.
[0011] According to the present invention, a method for organizing and tracking version chains and differences in mooring simulation multi-round test data is provided. Based on abnormal time-series data, a sliding window comparison at the time window level is performed to obtain abnormal time window information of the current version relative to the parent version, including: For each abnormal time series data, based on the preset window length and preset step size, the window slides according to the aligned current node information and the aligned parent node information to obtain the window difference score for each window; If the window difference score is not less than the preset window-level difference threshold, the current window is determined to be an abnormal time window; The set of start timestamps, end timestamps, and window difference scores of multiple abnormal time windows is used to determine the abnormal time window information of the current version relative to the parent version.
[0012] According to the present invention, a method for organizing and tracking version chains of multi-round test data for mooring simulation is provided, which stores the original time-series data, current node information, current summary result information, and relationship difference records into a preset time-series database, including: The raw time-series data is stored in a tree-like path that is subdivided level by business theme, current version identifier, generation directory, object identifier, and indicator identifier. Store the current node information in the metadata directory under the business topic and the current version identifier; The current summary results are stored in the summary directory under the corresponding version identifier; organized by object identifier and summary statistics item. Based on the comparison results between the current version identifier and the parent version identifier, the path name is dynamically generated according to the preset naming rules; based on the candidate abnormal objects, relationship difference records, and path names, the relationship difference records are centrally stored in a separate difference analysis directory.
[0013] This invention also provides a version chain organization and difference tracking device for multi-round test data in mooring simulation, comprising the following modules: The version creation module is used to create the current node information of the current version for each round of received raw time series data; according to preset organization rules, the raw time series data, current node information and current summary result information of the current version are stored in a preset time series database; wherein, the raw time series data is the time series data of the mooring simulation test; the current node information includes: current version identifier, parent version identifier and the set of changes of the current version relative to the parent version; The difference calculation module is used to calculate the object-level difference score of the current version relative to the parent version based on the current summary result information and the parent summary result information, provided that the parent version identifier of the current version is not empty; The anomaly filtering module is used to filter out anomaly objects based on the object-level difference score to obtain candidate anomaly objects; and to determine the abnormal time series data corresponding to the candidate anomaly objects based on the original time series data. The difference tracking module is used to perform a sliding window comparison at the time window level based on the abnormal time series data to obtain the abnormal time window information of the current version relative to the parent version; determine the relationship difference record based on the object-level difference score and the abnormal time window information; and store the original time series data, the current node information, the current summary result information and the relationship difference record into the preset time series database.
[0014] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the version chain organization and difference tracking method for mooring simulation multi-round test data as described above.
[0015] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the version chain organization and difference tracking method for mooring simulation multi-round test data as described above.
[0016] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the version chain organization and difference tracking method for mooring simulation multi-round test data as described above.
[0017] The present invention provides a version chain organization and difference tracking method for multi-round test data in mooring simulation. This method involves: creating current node information for each round of received raw time-series data; storing the raw time-series data, current node information, and current summary result information of the current version in a preset time-series database according to preset organization rules; calculating an object-level difference score relative to the parent version based on the current summary result information and the parent summary result information, provided the parent version identifier is not empty; filtering out candidate anomalies based on the object-level difference score; determining the corresponding abnormal time-series data based on the raw time-series data; performing a sliding window comparison at the time window level to obtain the abnormal time window information of the current version relative to the parent version based on the abnormal time-series data; determining relationship difference records based on the object-level difference score and the abnormal time window information; and storing the raw time-series data, current node information, current summary result information, and relationship difference records in the preset time-series database. By creating a current version node containing the parent version node for each round of experiments, the lineage relationship between multiple rounds of experiments is established when the data is entered into the database. This fundamentally solves the problems of data silos and unclear lineage in the background technology, making experiment review and tracing possible. A two-level mechanism, from object-level summary screening to time window-level drill-down, avoids inefficient comparisons starting from the full set of raw data. High-difference objects are first located through lightweight summary calculation, and then fine-grained window analysis is performed on a few abnormal objects, significantly improving the efficiency and accuracy of difference location and solving the problem of high overhead for full-scale comparison. Furthermore, the difference calculation results, i.e., relationship difference records, are associated with and persistently stored with version nodes, making the differences between the current round and its parent round a queryable and reusable data asset, rather than a temporary comparison result, supporting the accumulation and reuse of experience. This achieves a clear version chain formed by the results of multiple rounds of experiments upon database entry, enabling the location of differences at the version, object, and time window levels, supporting experiment review, comparative analysis, and experience reuse. Attached Figure Description
[0018] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.
[0019] Figure 1 This is a flowchart illustrating the version chain organization and difference tracking method for multi-round test data in mooring simulation provided by the present invention.
[0020] Figure 2This is a schematic diagram of the overall architecture of multi-round mooring simulation test data provided by the present invention, which forms a version chain in IoTDB and generates difference results.
[0021] Figure 3 This is a schematic diagram of the relationship structure of version nodes, parent version relationships, and relationship difference results provided by the present invention.
[0022] Figure 4 This invention provides a flowchart for difference tracking, which involves first summarizing and screening, then locating the object, and finally drilling down through a time window.
[0023] Figure 5 This is a schematic diagram showing the organization of the version chain data provided by this invention in the IoTDB, including the original layer, metadata layer, summary layer, and difference layer.
[0024] Figure 6 This is a comparison diagram of the processing paths between the traditional island-style round comparison method and the version chain tracing method of this invention.
[0025] Figure 7 This is a schematic diagram of the version chain organization and difference tracking device for mooring simulation multi-round test data provided by the present invention.
[0026] Figure 8 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation
[0027] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.
[0028] The following is combined Figures 1 to 8 This invention is described.
[0029] Figure 1 This is a flowchart illustrating the version chain organization and difference tracking method for multi-round test data in mooring simulation provided by the present invention, as shown below. Figure 1 As shown, the method includes the following: Step 101: For each round of received raw time series data, create the current node information for the current version; according to the preset organization rules, store the raw time series data, current node information and current summary result information of the current version into the preset time series database; wherein, the raw time series data is the time series data of the mooring simulation test; the current node information includes: the current version identifier, the parent version identifier and the set of changes of the current version relative to the parent version.
[0030] In step 101 above, the raw time series data of the current mooring simulation output is received, and a unique version identifier run_id_k is assigned to the result of the current mooring.
[0031] Based on the source of the test task, the test branch, or the user-specified relationship, record parent_run_id_k for the current version and synchronously record change_set_k for this round; if the current round is the baseline round, set parent_run_id_k to null and mark it as the root version.
[0032] According to the preset organization rules, the target locations of the original time series, version metadata, summary results, and difference results are mapped in the preset time series database.
[0033] In this embodiment of the invention, any time-series database that can support time-series data writing, time-window retrieval, and event triggering or subscription distribution capabilities can be used as a preset time-series database; preset time-series databases include, but are not limited to, IoTDB, InfluxDB, TimescaleDB, and OpenTSDB.
[0034] Extract summary statistics such as max, mean, rms, std, and p2p based on the object and indicator to form the current summary result information Summary_k(o,m).
[0035] Unlike the traditional approach of storing each round of simulation results as an isolated dataset in a separate database, this invention abstracts each round of experiment into a version node run_id_k, explicitly records its parent_run_id_k and change_set_k, and synchronously generates summary difference results and time window difference results relative to the parent version when the version is stored in the database. This allows the pre-set time series database to directly carry the tracking chain of which round this round came from, what was changed in this round, and in which objects and time windows the results have changed significantly.
[0036] This application uniformly registers the results of multiple rounds of experiments as version nodes and maintains parent-child relationships; secondly, it organizes the original layer, metadata layer and summary layer with version nodes as the core, and organizes the difference layer with the relationship between the current version and the parent version; thirdly, it first completes object-level rapid screening based on the summary results; fourthly, it only performs sliding window difference analysis on the screened candidate objects to locate specific abnormal time windows.
[0037] The difference results in this application are not isolated attributes inherent to a particular version, but rather comparison results always generated relative to its parent version or a specified comparison anchor. In other words, this invention emphasizes a difference tracking mechanism on a version chain, rather than simply adding a difference table or performing a temporary comparison.
[0038] like Figure 2 As shown, the method of this invention can be abstracted into a test access module, a version registration and node organization module, a parent version difference calculation module, and a difference traceability output module. For ease of implementation, the version registration and node organization module can be further subdivided into a version registration submodule, a summary generation submodule, and a version node organization submodule. These modules can be deployed in the same processing chain or deployed separately according to data processing responsibilities.
[0039] The test access module is responsible for receiving the data output from each round of mooring simulation; the version registration submodule assigns a run_id to the results of this round and records its parent_run_id and change_set; the summary generation submodule is responsible for generating object-level summary results; the version node organization submodule is responsible for writing the original time series, version metadata, and summary results into the preset time series database according to unified rules; the parent version difference calculation module is responsible for generating object-level difference results based on the relationship between the current version and the parent version; the time window level drill-down module is responsible for continuing to locate the time period with the most significant difference on the candidate abnormal objects; and the difference tracing output module is responsible for outputting the version-level, object-level, and time window-level tracing results and providing a unified entry point for subsequent analysis or review.
[0040] In one embodiment, data of the same version can be organized in the following path or logical organization form: root.mooring.version.run_0008.raw.rope_01.tension represents the original tension sequence of the current version; root.mooring.version.run_0008.meta.parent_run_id represents its parent version identifier; root.mooring.version.run_0008.summary.rope_01.max_tension represents the summary result of this object; `root.mooring.diff.run_0007_to_run_0008.rope_01.tension.score` represents the object-level difference result for comparing the relationship `run_0007 -> run_0008`. The above path is only an example and is not limited to a specific naming format.
[0041] Optionally, before calculating the object-level difference score of the current version relative to the parent version in step 102, steps A1 to A2 are also included: Step A1: Based on the object identifier, indicator identifier, and stage label, align the current node information and the parent node information to obtain the aligned current node information and the aligned parent node information.
[0042] Step A2: Based on the sampling frequency, data duration, or stage boundary, if the current node information and the parent node information are inconsistent, resample the time series data corresponding to the current node information and the time series data corresponding to the parent node information respectively; or, aggregate the time series data corresponding to the current node information and the time series data corresponding to the parent node information to a unified time reference.
[0043] In steps A1 to A2 above, the current version is aligned with the parent version according to the object identifier, indicator identifier, and stage label; when the sampling rate or duration is inconsistent, it is first resampled or aggregated to a unified time base.
[0044] Regarding version alignment, one possible implementation is to first align by stage labels, and then perform resampling at a uniform time step; if stage labels do not exist, window alignment can be performed according to the benchmark time point agreed upon by the business. For two rounds of trials with inconsistent durations, window comparison can be performed only on the overlapping time intervals, or the shorter time interval can be used as the common comparison range.
[0045] Step 102: If the parent version identifier of the current version is not empty, calculate the object-level difference score of the current version relative to the parent version based on the current summary result information and the parent summary result information.
[0046] In step 102 above, if the current version has a parent version, the parent version Summary_p(o,m) is read, DiffScore_k(o,m) is calculated according to the weight model, and candidate abnormal object pairs are screened out; if the parent version does not exist, this step is skipped and the current version is used as the subsequent comparison anchor point.
[0047] Optionally, step 102 includes steps B1 to B2: Step B1: Based on the current summary result information and the parent summary result information, for each summary statistics item to be compared in the aligned current data and aligned parent data, calculate the absolute difference of the current version relative to the parent version in the summary statistics item.
[0048] Step B2: Normalize the absolute differences of multiple summary statistics using a preset normalization function to obtain object-level difference scores.
[0049] Step 103: Based on the object-level difference score, filter out abnormal objects to obtain candidate abnormal objects; based on the original time series data, determine the abnormal time series data corresponding to the candidate abnormal objects.
[0050] Optionally, in step 103, the process of filtering out anomalous objects based on object-level difference scores to obtain candidate anomalous objects includes step C1: Step C1: If the object-level difference score is not less than the preset object-level difference threshold, mark the object identifier and indicator identifier corresponding to the object-level difference score as candidate abnormal objects.
[0051] Step 104: Based on the abnormal time series data, perform a sliding window comparison at the time window level to obtain the abnormal time window information of the current version relative to the parent version; determine the relationship difference record based on the object-level difference score and the abnormal time window information; and store the original time series data, the current node information, the current summary result information, and the relationship difference record into the preset time series database.
[0052] In step 104 above, as Figure 3 As shown, Figure 3 This is a schematic diagram of the relationship structure of version nodes, parent version relationships, and relationship difference results provided by the present invention.
[0053] Perform a sliding window comparison only on candidate abnormal object pairs, calculate WindowDiff_k(o,m,j), and locate the time window with the most significant difference.
[0054] Write object-level difference scores, window-level difference scores, Top N anomaly time windows and their associated metadata into the preset time-series database difference layer.
[0055] A unified comparison query entry is generated for the current version and the parent version, so that subsequent analysis does not need to splice data from multiple places again.
[0056] It outputs version-level, object-level, and time-window-level difference tracking results to the business side, supporting quick location of target segments that need to be reviewed.
[0057] When the next round of testing continues to adjust based on the previous round, the previous round's version node is used as the new parent version, forming a sustainably expanding version chain.
[0058] Optionally, in step 104, based on the abnormal time-series data, a sliding window comparison at the time window level is performed to obtain the abnormal time window information of the current version relative to the parent version, including steps D1 to D3: Step D1: For each abnormal time series data, based on the preset window length and preset step size, slide the window according to the current node information and the parent node information after alignment to obtain the window difference score of each window.
[0059] Step D2: If the window difference score is not less than the preset window-level difference threshold, determine that the current window is an abnormal time window.
[0060] Step D3: Determine the set of start timestamps, end timestamps, and window difference scores of multiple abnormal time windows as the abnormal time window information of the current version relative to the parent version.
[0061] Optionally, in step 104, storing the original time-series data, current node information, current summary result information, and relation difference records into a preset time-series database includes steps E1 to E4: Step E1: Store the raw time-series data in a tree-like path that is subdivided level by business theme, current version identifier, generation directory, object identifier, and indicator identifier.
[0062] Step E2: Store the current node information in the metadata directory under the business topic and the current version identifier.
[0063] Step E3: Store the current summary results information in the summary directory under the corresponding version identifier; organize them by object identifier and summary statistics.
[0064] Step E4: Based on the comparison results between the current version identifier and the parent version identifier, dynamically generate path names according to preset naming rules; based on candidate anomaly objects, relationship difference records, and path names, centrally store relationship difference records in an independent difference analysis directory.
[0065] Specifically, such as Figure 4 and Figure 5 As shown, Figure 4 This invention provides a flowchart for difference tracking, which involves first summarizing and screening, then locating the object, and finally drilling down through a time window. Figure 5 This is a schematic diagram showing the organization of the version chain data provided by this invention in the IoTDB, including the original layer, metadata layer, summary layer, and difference layer.
[0066] Assuming the version node corresponding to the k-th round of testing is Run_k, then we can define: Run_k={run_id_k, parent_run_id_k, ChangeSet_k, raw_ref_k, summary_ref_k, view_ref_k}; Where `run_id_k` represents the current version identifier, `parent_run_id_k` represents the parent version identifier, `ChangeSet_k` represents the set of changes relative to the parent version in this round, and `raw_ref_k`, `summary_ref_k`, and `view_ref_k` represent references to the raw data, summary results, and comparison query entry points, respectively. If `parent_run_id_k` is empty, the current version can be used as the baseline root version, generating only the raw layer, metadata layer, and summary layer, without generating difference results relative to the parent version.
[0067] RelDiff_k = {parent_run_id_k, run_id_k, diff_ref_k}; The Relationship Difference Record RelDiff_k is used to represent the comparison relationship from the parent version to the current version, where diff_ref_k points to the object-level difference result, window-level difference result, and the output position of the abnormal time window on this relationship.
[0068] For object o and index m, the summary result for the current version is defined as: Summary_k(o,m) = {max_k, min_k, mean_k, rms_k, std_k, p2p_k}; The summary results set can be constructed by selecting statistical items such as maximum value, minimum value, mean, root mean square, standard deviation, and peak-to-peak value, according to business needs. Let p = parent(k) represent the parent version of the current version, then the object-level difference score can be defined as: DiffScore_k(o,m) = Sum_{q in Q}( w_q * norm( |Summary_k^q(o,m) -Summary_p^q(o,m)| ) ) Here, norm(x) represents the normalization of the difference x. In one embodiment, a relative difference normalization method of x / (|baseline_q| + epsilon) can be used; in another embodiment, a min-max normalization method based on historical statistical range can also be used. The purpose of normalization is to eliminate scoring bias caused by inconsistencies in the dimensions of different abstract items.
[0069] Before performing object-level and time-window-level comparisons, version alignment needs to be completed first. Specifically, alignment can be performed by object identifier o, indicator identifier m, and stage label or condition label; when the sampling frequency, duration, or stage boundary of the current version is inconsistent with that of the parent version, the two can be resampled or aggregated to a unified time base before proceeding to subsequent difference calculations.
[0070] When DiffScore_k(o,m) is greater than the object-level difference threshold Theta_obj, object o and index m are marked as candidate outlier object pairs. Subsequently, time window-level drill-down is performed only on candidate outlier object pairs. Let the window difference score corresponding to the j-th sliding window on the aligned unified time base be: WindowDiff_k(o,m,j) = lambda1 * norm(|max_k^j - max_p^j|) + lambda2 *norm(|mean_k^j - mean_p^j|) + lambda3 * norm(|rms_k^j - rms_p^j|) When WindowDiff_k(o,m,j) is greater than the window-level difference threshold Theta_win, the window can be identified as an abnormal time window that needs to be reviewed in detail. This forms a two-level difference tracking mechanism of first summarizing and then drilling down into the window.
[0071] The meanings of the above symbols and parameters are shown in Table 1.
[0072] Table 1. Symbol and Parameter Description
[0073] The following is a specific implementation example, in which the summary metric set Q is initialized to max, mean, rms, std, and p2p; the object-level difference threshold Theta_obj can be set to a normalized score of 0.20 to 0.35; the sliding window length L_w can be set to 60 seconds to 300 seconds depending on the response period of the simulated object; the window step size Delta_w can be set to 10 seconds to 30 seconds; and the number of window-level outputs TopN can be set to 3 to 10.
[0074] For scenarios where cable tension and hull displacement are the primary analytical objects, the weights of maximum value, root mean square, and peak-to-peak value in the object-level difference score can be appropriately increased; for scenarios that focus more on trend changes, the weights of mean and standard deviation can be appropriately increased. If there are many objects in a single round of testing, object-aligned queries can be performed first at the summary level before proceeding to window-level drill-down.
[0075] The above parameters are merely suggestions for the embodiments. In actual applications, they can be adjusted according to the sampling frequency, ship type, mooring configuration, environmental conditions, and performance constraints of the downstream analysis module. However, none of these adjustments affect the overall technical solution of the present invention, which organizes multiple rounds of testing using a version chain and performs parent version difference tracking.
[0076] The pseudocode description of this invention is given below.
[0077] Step A: Initialize version registration rules, change set recording rules, summary metric set Q, object-level difference threshold Theta_obj, window length L_w, and window step size Delta_w.
[0078] Step B: Receive the current round of test data, create a run_id_k for it, and bind parent_run_id_k and change_set_k; if it is the baseline round, set parent_run_id_k to null.
[0079] Step C: First, align the current version with the parent version by object, metric, and stage label, and resample to a unified time base if necessary.
[0080] Step D: Extract Summary_k(o,m) by object and metric; if the current version has a parent version, then further read the parent version's Summary_p(o,m).
[0081] Step E: If the current version does not have a parent version, only the original data, version metadata, and summary results are saved; if a parent version exists, DiffScore_k(o,m) is calculated. When DiffScore_k(o,m) is less than Theta_obj, window-level difference fixing is skipped; when DiffScore_k(o,m) is greater than or equal to Theta_obj, it is marked as a candidate anomalous object pair.
[0082] Step F: Perform a sliding window comparison on the candidate abnormal object pairs, calculate WindowDiff_k(o,m,j), output the TopN abnormal windows, and generate diff_ref_k on the relationship record of "parent version -> current version".
[0083] Step G: Write the original layer, metadata layer, summary layer, and relational difference layer into IoTDB, and open a unified entry point for version comparison and difference tracking.
[0084] This invention can also be implemented in other optional ways, as follows: The original time series, summary results, and difference results are organized in a tree structure of version-object-indicator, which is suitable for integration with existing time series path models.
[0085] Alternatively, a version metadata table, a summary results table, and a window difference table can be created separately, and the relationships can be established through run_id and parent_run_id, which is suitable for interfacing with structured analysis systems.
[0086] And / or, only the summary layer retains all version results for a long time, while the original layer adopts a hierarchical retention or external archiving and lookup strategy, which is suitable for storage cost-sensitive scenarios.
[0087] And / or, while keeping the parent version chain unchanged, allow the current version to generate additional difference results relative to a specified comparison anchor point to support cross-layer comparisons relative to the current version.
[0088] like Figure 6 As shown, Figure 6 This is a comparison diagram of the processing paths between the traditional island-style round comparison method and the version chain tracing method of this invention.
[0089] Compared to the traditional method of storing multiple simulation results separately and then manually exporting and comparing them, this invention first establishes a clear version chain at the data organization level, so that each round of results has a clear source and background of modification, reducing the problem of information breakage in experimental review.
[0090] Compared to directly comparing the entire original curves, this invention uses an object-level summary screening mechanism to prioritize high-discrepancy objects, and then uses a time window-level drill-down mechanism to narrow down the analysis scope, reducing invalid reads, invalid comparisons, and invalid replays.
[0091] Compared to methods that can only answer the question of whether the two rounds are different, this invention can further answer which round changed relative to which round, which type of object changed significantly, and which time window is most worth reviewing, thus making the difference tracking interpretable and operable.
[0092] Without altering the original business semantics of mooring simulation data, this invention expands the preset time-series database from a simple time-series storage base into an organizational base capable of supporting multiple rounds of experimental version chains and relational difference assets. This facilitates the accumulation of subsequent experimental experience, parameter optimization review, and reuse of simulation results.
[0093] This invention provides a version chain organization and difference tracking method for multi-round test data in mooring simulation. For each round of received raw time-series data, the method creates current node information for the current version; according to preset organization rules, it stores the raw time-series data, current node information, and current summary result information of the current version into a preset time-series database; if the parent version identifier of the current version is not empty, it calculates an object-level difference score relative to the parent version based on the current summary result information and the parent summary result information; based on the object-level difference score, it filters out abnormal objects to obtain candidate abnormal objects; based on the raw time-series data, it determines the abnormal time-series data corresponding to the candidate abnormal objects; based on the abnormal time-series data, it performs a time-window-level sliding window comparison to obtain the abnormal time window information of the current version relative to the parent version; based on the object-level difference score and the abnormal time window information, it determines the relationship difference record; and it stores the raw time-series data, current node information, current summary result information, and relationship difference record into the preset time-series database. By creating a current version node containing the parent version node for each round of experiments, the lineage relationship between multiple rounds of experiments is established when the data is entered into the database. This fundamentally solves the problems of data silos and unclear lineage in the background technology, making experiment review and tracing possible. A two-level mechanism, from object-level summary screening to time window-level drill-down, avoids inefficient comparisons starting from the full set of raw data. High-difference objects are first located through lightweight summary calculation, and then fine-grained window analysis is performed on a few abnormal objects, significantly improving the efficiency and accuracy of difference location and solving the problem of high overhead for full-scale comparison. Furthermore, the difference calculation results, i.e., relationship difference records, are associated with and persistently stored with version nodes, making the differences between the current round and its parent round a queryable and reusable data asset, rather than a temporary comparison result, supporting the accumulation and reuse of experience. This achieves a clear version chain formed by the results of multiple rounds of experiments upon database entry, enabling the location of differences at the version, object, and time window levels, supporting experiment review, comparative analysis, and experience reuse.
[0094] The following describes the version chain organization and difference tracking device for mooring simulation multi-round test data provided by the present invention. The version chain organization and difference tracking device for mooring simulation multi-round test data described below can be referred to in correspondence with the version chain organization and difference tracking method for mooring simulation multi-round test data described above.
[0095] Figure 7 This is a flowchart illustrating the version chain organization and difference tracking device for multi-round test data in mooring simulation provided by the present invention, as shown below. Figure 7 As shown, the device includes the following: The version creation module 701 is used to create current node information for each round of received raw time series data; and to store the raw time series data, current node information and current summary result information of the current version into a preset time series database according to preset organization rules; wherein, the raw time series data is the time series data of the mooring simulation test; the current node information includes: current version identifier, parent version identifier and the set of changes of the current version relative to the parent version.
[0096] The difference calculation module 702 is used to calculate the object-level difference score of the current version relative to the parent version based on the current summary result information and the parent summary result information, provided that the parent version identifier of the current version is not empty.
[0097] The anomaly screening module 703 is used to screen anomaly objects based on the object-level difference score to obtain candidate anomaly objects; and to determine the anomaly time series data corresponding to the candidate anomaly objects based on the original time series data.
[0098] The difference tracking module 704 is used to perform a sliding window comparison at the time window level based on the abnormal time series data to obtain the abnormal time window information of the current version relative to the parent version; determine the relationship difference record based on the object-level difference score and the abnormal time window information; and store the original time series data, the current node information, the current summary result information and the relationship difference record into the preset time series database.
[0099] This invention provides a version chain organization and difference tracking device for multi-round test data in mooring simulation. For each round of received raw time-series data, it creates current node information for the current version; according to preset organization rules, it stores the raw time-series data, current node information, and current summary result information of the current version into a preset time-series database; if the parent version identifier of the current version is not empty, it calculates an object-level difference score relative to the parent version based on the current summary result information and the parent summary result information; based on the object-level difference score, it filters out abnormal objects to obtain candidate abnormal objects; based on the raw time-series data, it determines the abnormal time-series data corresponding to the candidate abnormal objects; based on the abnormal time-series data, it performs a time-window-level sliding window comparison to obtain the abnormal time window information of the current version relative to the parent version; based on the object-level difference score and the abnormal time window information, it determines the relationship difference record; and it stores the raw time-series data, current node information, current summary result information, and relationship difference record into the preset time-series database. By creating a current version node containing the parent version node for each round of experiments, the lineage relationship between multiple rounds of experiments is established when the data is entered into the database. This fundamentally solves the problems of data silos and unclear lineage in the background technology, making experiment review and tracing possible. A two-level mechanism, from object-level summary screening to time window-level drill-down, avoids inefficient comparisons starting from the full set of raw data. High-difference objects are first located through lightweight summary calculation, and then fine-grained window analysis is performed on a few abnormal objects, significantly improving the efficiency and accuracy of difference location and solving the problem of high overhead for full-scale comparison. Furthermore, the difference calculation results, i.e., relationship difference records, are associated with and persistently stored with version nodes, making the differences between the current round and its parent round a queryable and reusable data asset, rather than a temporary comparison result, supporting the accumulation and reuse of experience. This achieves a clear version chain formed by the results of multiple rounds of experiments upon database entry, enabling the location of differences at the version, object, and time window levels, supporting experiment review, comparative analysis, and experience reuse.
[0100] Figure 8 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 8 As shown, the electronic device may include a processor 810, a communications interface 820, a memory 830, and a communication bus 840. The processor 810, communications interface 820, and memory 830 communicate with each other via the communication bus 840. The processor 810 can call logical instructions from the memory 830 to execute a version chain organization and difference tracking method for mooring simulation multi-round test data.
[0101] Furthermore, the logical instructions in the aforementioned memory 830 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.
[0102] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer is able to execute the version chain organization and difference tracking method for mooring simulation multi-round test data provided by the above methods.
[0103] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, is implemented to perform the version chain organization and difference tracking method for mooring simulation multi-round test data provided by the methods described above.
[0104] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.
[0105] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.
[0106] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A method for organizing and tracking version chains and differences in multi-round test data for mooring simulation, characterized in that, include: For each round of received raw time-series data, create the current node information for the current version; According to preset organization rules, the original time-series data, current node information, and current summary result information of the current version are stored in a preset time-series database; wherein, the original time-series data is the time-series data of the mooring simulation test; the current node information includes: the current version identifier, the parent version identifier, and the set of changes of the current version relative to the parent version; If the parent version identifier of the current version is not empty, calculate the object-level difference score of the current version relative to the parent version based on the current summary result information and the parent summary result information; Based on the object-level difference score, abnormal objects are screened to obtain candidate abnormal objects; based on the original time series data, the abnormal time series data corresponding to the candidate abnormal objects are determined. Based on the abnormal time series data, a sliding window comparison at the time window level is performed to obtain the abnormal time window information of the current version relative to the parent version; based on the object-level difference score and the abnormal time window information, the relationship difference record is determined; the original time series data, the current node information, the current summary result information, and the relationship difference record are stored in the preset time series database.
2. The method for organizing and tracking version chains and differences for multi-round test data in mooring simulation according to claim 1, characterized in that, Before calculating the object-level difference score of the current version relative to the parent version, the following is also included: Based on object identifier, indicator identifier and stage label, the current node information and parent node information are aligned to obtain the aligned current node information and aligned parent node information. Based on the sampling frequency, data duration, or stage boundary, if it is determined that the current node information and the parent node information are inconsistent, the time series data corresponding to the current node information and the time series data corresponding to the parent node information are resampled respectively; or, the time series data corresponding to the current node information and the time series data corresponding to the parent node information are aggregated to a unified time reference.
3. The method for organizing and tracking version chains and differences for multi-round test data in mooring simulation according to claim 2, characterized in that, The calculation of the object-level difference score between the current version and the parent version based on the current summary result information and the parent summary result information includes: Based on the current summary result information and the parent summary result information, for each summary statistical item to be compared in the aligned current data and the aligned parent data, calculate the absolute difference between the current version and the parent version in the summary statistical item; By using a preset normalization function, the absolute differences of multiple summary statistics are normalized to obtain object-level difference scores.
4. The method for organizing and tracking version chains and differences for multi-round test data in mooring simulation according to claim 1, characterized in that, The step of filtering out anomalous objects based on the object-level difference score to obtain candidate anomalous objects includes: If the object-level difference score is not less than the preset object-level difference threshold, the object identifier and indicator identifier corresponding to the object-level difference score are marked as candidate abnormal objects.
5. The method for organizing and tracking version chains and differences for multi-round test data in mooring simulation according to claim 2, characterized in that, The step of performing a time-window-level sliding window comparison based on the abnormal time-series data to obtain the abnormal time-window information of the current version relative to the parent version includes: For each of the abnormal time series data, based on the preset window length and preset step size, the window is slid based on the aligned current node information and the aligned parent node information to obtain the window difference score for each window; If the window difference score is not less than a preset window-level difference threshold, the current window is determined to be an abnormal time window; The set of start timestamps, end timestamps, and window difference scores of multiple abnormal time windows is used to determine the abnormal time window information of the current version relative to the parent version.
6. The method for organizing and tracking version chains and differences for multi-round test data in mooring simulation according to claim 2, characterized in that, The step of storing the original time-series data, the current node information, the current summary result information, and the relationship difference record into the preset time-series database includes: The original time-series data is stored in a tree-like path that is progressively subdivided by business theme, current version identifier, generation directory, object identifier, and indicator identifier. The current node information is stored in the metadata directory under the business topic and the current version identifier; The current summary results information is stored in the summary directory under the corresponding version identifier; and organized by object identifier and summary statistics item. Based on the comparison result between the current version identifier and the parent version identifier, a path name is dynamically generated according to a preset naming rule; based on the candidate anomaly object, the relationship difference record, and the path name, the relationship difference record is centrally stored in an independent difference analysis directory.
7. A version chain organization and difference tracking device for multi-round test data in mooring simulation, characterized in that, include: The version creation module is used to create the current node information for the current version for each round of received raw time series data; According to preset organization rules, the original time-series data, current node information, and current summary result information of the current version are stored in a preset time-series database; wherein, the original time-series data is the time-series data of the mooring simulation test; the current node information includes: the current version identifier, the parent version identifier, and the set of changes of the current version relative to the parent version; The difference calculation module is used to calculate the object-level difference score of the current version relative to the parent version based on the current summary result information and the parent summary result information, provided that the parent version identifier of the current version is not empty; The anomaly filtering module is used to filter out anomaly objects based on the object-level difference score to obtain candidate anomaly objects; and to determine the abnormal time series data corresponding to the candidate anomaly objects based on the original time series data. The difference tracking module is used to perform a sliding window comparison at the time window level based on the abnormal time series data to obtain the abnormal time window information of the current version relative to the parent version; determine the relationship difference record based on the object-level difference score and the abnormal time window information; and store the original time series data, the current node information, the current summary result information and the relationship difference record into the preset time series database.
8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the version chain organization and difference tracking method for mooring simulation multi-round test data as described in any one of claims 1 to 6.
9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the version chain organization and difference tracking method for mooring simulation multi-round test data as described in any one of claims 1 to 6.
10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the version chain organization and difference tracking method for mooring simulation multi-round test data as described in any one of claims 1 to 6.