Method and system for accessing on an industrial data link

Abstract

The application discloses an industrial data chain access method and system, and relates to the field of industrial data processing.The method comprises the following steps: collecting time sequence monitoring data of industrial equipment, deploying a trend detector on the edge side, dynamically adjusting the anchor frequency on the chain according to the data mutation intensity and abnormal confidence, generating an event snapshot containing a time window and associated point positions when capturing a transient event, preferentially uploading the event snapshot to the chain through an independent channel, and falling back to the normal anchor rhythm after the abnormality subsides.When archiving historical data, a link record containing an inter-volume association index is uploaded to the chain during volume switching, a cross-volume index directory corresponding to the hash root on the chain is maintained, and the anchor time period window on the chain is queried before the archiving system switches the volume to align the boundary.The application solves the problems of transient abnormality tracing blind area and cross-volume disconnection caused by the existing fixed anchor, improves the data tracing continuity and key event capturing capability, and provides reliable support for industrial data auditing and fault tracing.

Description

Technical Field

[0001] This invention relates to the field of industrial data processing, specifically to an industrial data chain access method and system. Background Technology

[0002] This invention relates to a method and system for accessing industrial data chains, applicable to the collection, storage, and traceability of equipment status monitoring data in fields such as intelligent manufacturing and energy and chemical engineering. Existing industrial data chain access technologies suffer from significant bottlenecks: firstly, the use of fixed-time granularity to anchor data makes it easy to miss transient anomalies such as short-term pressure spikes and vibration shocks; furthermore, the static threshold triggering mechanism lags in responding to changes in data characteristics, failing to construct a complete chain of risk evidence; secondly, when historical data is archived in volumes, the volume boundaries often do not align with the on-chain anchoring boundaries, leading to broken traceability across volumes and a lack of continuous evidence for auditing. Existing technologies also lack on-chain / off-chain collaborative mechanisms, resulting in low retrieval efficiency.

[0003] Therefore, this invention proposes an industrial data chain access method and system. Summary of the Invention

[0004] To achieve the above objectives, this application provides a method and system for accessing industrial data chains, the specific technical solution of which is as follows:

[0005] An industrial data chain access method includes:

[0006] S1: Obtain raw time-amplitude data from industrial data link monitoring points, construct a two-dimensional coordinate system based on the raw time-amplitude data; set unit judgment time slot parameters in the two-dimensional coordinate system and input the time slot sliding algorithm to output time slot window data; the time slot window data is subjected to non-maximum suppression to obtain the suppressed anchor sequence.

[0007] S2: The algorithm for determining the type of related segments after suppressing the input of the anchor sequence is used to obtain highly related anchor data. After time interval expansion, hash value extraction, time correction and aggregated indexing, the bookmark master table is obtained.

[0008] S3: Apply the bookmark master table data to perform head-to-tail alignment verification, generate inter-volume connection record data, fill in missing segments of the inter-volume connection record data and arrange them in chronological order to generate complete cross-volume directory data;

[0009] S4: Construct two visualization views based on the complete cross-volume directory data. After generating a continuity certificate based on the two views, input the time interval threshold to detect the gap mark. Review the gap mark to obtain the verification result.

[0010] Preferably, the acquisition of time-amplitude raw data by the industrial data chain monitoring point includes the following steps:

[0011] Real-time time-amplitude raw data is received from industrial data link monitoring points; each time-amplitude raw data consists of a timestamp and the corresponding signal amplitude, and the signal originates from industrial sensors.

[0012] Periodically query sensor data, save the timestamp and corresponding signal amplitude to the raw data stream, forming data points of timestamp and signal amplitude.

[0013] Preferably, constructing a two-dimensional coordinate system based on the original time-amplitude data includes the following steps:

[0014] Using the acquired time-amplitude raw data, a two-dimensional coordinate system with X and Y axes is constructed. In the two-dimensional coordinate system with X and Y axes, the X-axis represents the timestamp and the Y-axis represents the signal amplitude.

[0015] The data points of timestamp and signal amplitude are mapped onto a two-dimensional coordinate system to form a point set.

[0016] The process of setting unit determination time slot parameters in a two-dimensional coordinate system, inputting a time slot sliding algorithm, and then outputting time period window data includes the following steps:

[0017] A time-slot sliding algorithm is constructed. The unit judgment time slot parameter is defined in a two-dimensional coordinate system to generate time-slot window data. At the center of the current time-slot window, a local segment is obtained by sampling with a temporary length.

[0018] The local volatility and event density are calculated based on the time window data. The length of the time window is adjusted by the local volatility and event density to obtain an adaptive time slot.

[0019] Preferably, obtaining the suppressed anchor sequence from the time window data through non-maximum suppression includes the following steps:

[0020] Using an adaptive time slot as the time window length, the time range of the m-th window is generated. Window feature vectors are extracted from local segments to form window data. Anchor point scores are calculated for each window data, and an initial anchor point is generated at the center of the window. Non-maximum suppression is applied to the initial anchor point. Candidate anchor points in the neighborhood are merged using a suppression radius related to the window length. Only the candidate anchor point with the highest score is retained in the neighborhood of the same suppression radius. The candidate anchor points retained after non-maximum suppression constitute the suppressed anchor sequence.

[0021] Preferably, the algorithm for determining the type of associated fragment after suppression of the anchor sequence input to obtain highly associated anchor point data includes the following steps:

[0022] A correlation fragment type determination algorithm is constructed. The input is the suppressed anchor sequence, and the maximum candidate time difference, similarity weight, correlation threshold and type determination threshold group are defined. The correlation fragment type determination algorithm first calculates a lightweight fingerprint in the window feature vector of the suppressed anchor sequence, and then filters candidate anchor pairs with time proximity constraints and score significance constraints, and outputs candidate pairs.

[0023] For each candidate pair, calculate the five-dimensional similarity and weighted synthesize the association score. Construct an undirected graph with the suppressed post-anchor sequence as nodes and anchor pairs with association scores as edges. Perform connected component or modularity-based community detection on the undirected graph to obtain anchor clusters.

[0024] Calculate anchor cluster-level features for each anchor cluster and determine the type of associated fragments, outputting the type and confidence level; output highly associated anchor data for each anchor cluster.

[0025] Preferably, the bookmark master table data obtained by expanding the highly correlated anchor data over time intervals, extracting hash values, time correction, and aggregating indexes includes:

[0026] Highly correlated anchor data is a set of anchor clusters output from anchor clusters, with an adaptive expansion of the original time range for each anchor cluster; the expanded start and end times are then used in the original time-amplitude dataset. Slice the middle section to obtain the expanded fragment of the anchor cluster;

[0027] Each anchor cluster is paired with its corresponding expansion fragment to obtain an expansion fragment set. Strong hash extraction is then performed on the expansion fragment set to obtain a hash chain set corresponding to each anchor cluster.

[0028] A time synchronization model is derived and applied from the set of expanded segments and the suppressed anchor sequence. An affine drift model is used to describe the mapping relationship between the device time and the reference time. The timestamp of each sample in the expanded segment is corrected according to the affine drift model, resulting in a set of block-level tuples indexed by the anchor cluster number and the block number.

[0029] Multiple types of indexes are constructed using a set of block-level tuples, including time indexes, hash indexes, and anchor context indexes. These three types of indexes are used to obtain a clustered index. Each block is organized into a unified bookmark entry based on the clustered index, and the total bookmark table data is formed based on all bookmark entries.

[0030] Preferably, the bookmark master table data undergoes a beginning-to-end alignment check to generate inter-volume connection record data. After supplementing missing segments in the inter-volume connection record data, it is arranged in chronological order to generate complete cross-volume directory data, including the following steps:

[0031] Using the bookmark master table data, we determine the connection between adjacent volumes for bookmark entries and calculate the consistency score of candidate pairs. For candidate pairs that meet the criteria, we generate an inter-volume connection record. We then combine all inter-volume connection records into a set to generate inter-volume connection record data.

[0032] Preferably, in the highly correlated anchor data, anchor clusters whose expanded time range overlaps with the gap interval are searched, and anchor clusters of type retransmission supplementation or sequential extension are selected as priority candidate anchor clusters; strong hash extraction, time correction and aggregation index are re-executed on the candidate anchor clusters within the gap interval to obtain the bookmark entries for supplementation.

[0033] The newly added bookmark entries are inserted into the bookmark master table data, and then sorted again according to the start time after synchronization. After sorting, a global serial number is assigned to each bookmark entry, and the sequence number of each global serial number is equal to the position number of the bookmark entry corresponding to the global serial number in the bookmark master table data. At the same time, cross-volume consecutive pointers are set for each bookmark entry, that is, the current global serial number is linked with the adjacent global serial numbers before and after it to form complete cross-volume directory data.

[0034] Preferably, the process of constructing two visualization views based on complete cross-volume directory data, generating a continuity certificate based on the two views, detecting gap markers by inputting a time interval threshold, and verifying the gap markers to obtain the verification result includes the following steps:

[0035] The complete cross-volume catalog data leads to a timeline view and a topology view, which are used together to generate continuity certificates and verification records.

[0036] A leaf hash is constructed for each bookmark entry, and these are merged upwards layer by layer to obtain a hash root; the hash root and the time range covered by the complete cross-volume directory are digitally signed using a private key to form a continuity certificate;

[0037] Calculate the adjacent interval between two adjacent bookmark entries in the bookmark master table data, calculate the time interval threshold based on the adjacent interval, and mark the position between the corresponding adjacent bookmark entries as a gap when the adjacent interval is greater than the time interval threshold.

[0038] For each location marked as a gap, a review is conducted according to three scenarios: cases where there is sufficient evidence but a business gap exists, cases where the gap can be repaired by adding bookmark entries, and cases where the connection is questionable; a re-inspection record is generated based on the review results.

[0039] An industrial data chain access system for implementing the aforementioned industrial data chain access method includes: a raw data processing module, a highly correlated anchor point module, a cross-volume directory module, and a visualization view module.

[0040] The raw data processing module acquires time-amplitude raw data from industrial data chain monitoring points, constructs a two-dimensional coordinate system based on the time-amplitude raw data, sets unit judgment time slot parameters in the two-dimensional coordinate system and inputs a time slot sliding algorithm, and outputs time period window data. The time period window data is subjected to non-maximum suppression to obtain a suppressed anchor sequence.

[0041] The highly correlated anchor module suppresses the input of the anchor sequence to obtain highly correlated anchor data through the related fragment type judgment algorithm. The highly correlated anchor data is then expanded over time intervals, hash values ​​are extracted, time is corrected, and aggregated indexes are applied to obtain the bookmark master table data.

[0042] The cross-volume directory module uses the bookmark master table data to perform first-to-last alignment verification, generates inter-volume connection record data, fills in missing segments of the inter-volume connection record data, and then arranges it in chronological order to generate complete cross-volume directory data.

[0043] The visualization view module constructs two visualization views based on the complete cross-volume directory data. After generating a continuity certificate based on the two views, it inputs a time interval threshold to detect gap marks and verifies the gap marks to obtain the verification results.

[0044] The beneficial effects of this invention are as follows:

[0045] This invention constructs structured raw data in a time-amplitude two-dimensional coordinate system, combines it with a time-slot sliding algorithm to accurately segment the window, and then uses non-maximum suppression to remove redundant anchor points, effectively filtering monitoring noise interference, improving the accuracy of the anchor sequence, and laying a reliable data foundation for subsequent screening of highly correlated anchor points.

[0046] This invention uses a correlation fragment judgment algorithm to filter highly correlated anchor points, and then uses time expansion to complete the context, hash value to create a unique identifier, time correction to remove deviation, and aggregated index classification to enhance the correlation and searchability of anchor points, thereby significantly reducing the query cost when accessing cross-volume data.

[0047] This invention identifies cross-volume breakpoints through end-to-end alignment verification, supplements missing data segments, and arranges them by time, solving the problems of disordered connection and missing segments in multi-volume data, ensuring the integrity and continuity of cross-volume directories, and avoiding misjudgments in business logic caused by data breaks.

[0048] This invention constructs a visual view based on cross-volume directories, combines continuity certificates with time threshold detection to automatically mark gaps, reduces false positives during verification, improves the intuitiveness and accuracy of data continuity detection, and reduces the time and effort required for manual verification.

[0049] This invention forms a complete process from raw data processing to anchor point optimization, catalog integration, and quality verification, improving the efficiency of industrial data chains and management, ensuring data accuracy and consistency, and providing high-quality data support for scenarios such as equipment fault tracing and production optimization. Attached Figure Description

[0050] Figure 1 A flowchart of an industrial data chain access method provided in this application;

[0051] Figure 2 The time slot sliding algorithm flowchart provided in this application;

[0052] Figure 3 Flowchart of the algorithm for determining the type of associated fragments provided in this application;

[0053] Figure 4 This application provides a structural diagram of an industrial data chain access system. Detailed Implementation

[0054] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the specific embodiments of this application will be described in detail below with reference to the accompanying drawings.

[0055] Many specific details are set forth in the following description in order to provide a full understanding of this application. However, this application may also be implemented in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of this application. Therefore, this application is not limited to the specific embodiments disclosed below.

[0056] Secondly, the term "one embodiment" or "example" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of this application. An embodiment appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single embodiment or an embodiment selectively mutually exclusive with other embodiments.

[0057] Example 1

[0058] Reference Figures 1 to 3 This is the first embodiment of the present application, such as Figure 1 As shown, an industrial data chain access method is provided.

[0059] Step S1: Obtain raw time-amplitude data from industrial data chain monitoring points, and construct a two-dimensional coordinate system based on the raw time-amplitude data; set the unit judgment time slot parameter in the two-dimensional coordinate system and input the time slot sliding algorithm to output the time slot window data; the time slot window data is subjected to non-maximum suppression to obtain the suppressed anchor sequence.

[0060] Raw time-amplitude data is acquired at monitoring points in the industrial field to ensure judgments are made under unified time and amplitude semantics. Specifically, sensors and data acquisition devices matching the measured object are deployed at each industrial data chain monitoring point. Sensor types can be temperature, pressure, flow, rotational speed, or high-sensitivity vibration accelerometers. The data acquisition device packages the raw measurement values ​​along with the local acquisition time into a data frame and sends it to the nearest data acquisition gateway.

[0061] The data acquisition gateway first performs time synchronization, unifying the local time of each acquisition device to the same reference clock through a time source (such as network time synchronization, precision time synchronization, or satellite time synchronization); the data acquisition gateway then performs protocol and format normalization, converting measurement values ​​from different sources and with different dimensions into a unified key-value format of timestamp plus signal amplitude value.

[0062] After receiving the data, the platform performs quality verification and cleaning. Out-of-order time records are rearranged in ascending time order; for duplicate timestamps, only the one with the smallest amplitude fluctuation is retained; short-term missing segments are interpolated and filled in based on the trend of adjacent points. Short-term missing segments refer to a short period in the original time-amplitude data where no actual records are found, even if continuous records should have appeared. Occasional extreme values ​​are suppressed according to robustness criteria. Occasional extreme values ​​refer to abnormally large or small values ​​that occur occasionally in the time series and last for a very short period (usually only one or two sampling points). Robustness criteria suppression means not deleting these values. The process involves not only minimizing extreme values ​​but also bringing the values ​​back to a reasonable range. For slowly drifting operating condition components, detrending processing is performed using the rolling median as the baseline. Slowly drifting operating condition components refer to low-frequency, continuous, non-sudden baseline shifts in the signal amplitude caused by slow changes in operating conditions or sensors. Detrending with the rolling median removes the base of these low-frequency, continuous, non-sudden baseline shifts in the signal amplitude caused by slow changes in operating conditions or sensors, leaving only the effective fluctuations around zero. Finally, unit conversion and amplitude normalization are performed according to the monitoring point or measurement channel. After the above process, records are obtained in ascending order of time, each with a timestamp. With a corresponding signal amplitude The original dataset, which is the time-amplitude raw data D1, is used as the direct input for subsequent modeling.

[0063] After obtaining D1, a two-dimensional coordinate system M2 is constructed based on D1 to map timestamps and signal amplitudes to the same geometric space, facilitating sliding analysis and visual verification. The construction of M2 follows these rules: the horizontal axis is the time axis, with the time zero point taken from the earliest timestamp in D1; the vertical axis is the amplitude axis, with the amplitude zero point taken from the cleaned baseline, and the amplitude unit uses the physical quantity unit of each channel or its normalized unit; the scale resolution of the time axis is given by the effective sampling period of D1, ensuring that adjacent time scales cover at least one sampling interval; the upper and lower limits of the amplitude axis scale are determined by the statistical minimum to maximum or corresponding quantile range of D1, ensuring that the vast majority of samples can be represented within M2.

[0064] After setting up the coordinate axes and scales, map each pair of timestamps and signal amplitude values ​​in D1 to a point in M2. When multiple channel amplitudes exist at the same time, use layered layers or color coding to distinguish different channels, and record channel identifiers, unit descriptions, and quality marks in the layer metadata to ensure that subsequent evidence collection of local windows can be traced back to the channel and source. Multi-channel amplitude refers to the set of amplitude values ​​measured by multiple measurement channels at the same time.

[0065] like Figure 2 The diagram shown is a flowchart of the time-slot sliding algorithm for this step. The time-slot sliding algorithm is constructed to adaptively slide the original time-amplitude data D1 through time slots, extract window feature vectors and score them, then perform redundancy removal based on time-varying radius non-maximum suppression, and output the suppressed anchor sequence. The goal is to structure the raw time-amplitude data D1 into computable features, i.e., the suppressed post-anchor sequence. This provides a stable dimension for subsequent determinations.

[0066] The specific steps for constructing a time-slot sliding algorithm include: based on raw time-amplitude data... ,in, For timestamps, For the corresponding amplitude, N is the total number of data points, and k is the index of the number of data points. In the XY axis coordinate system... The definition of the unit determines the time slot parameters, including the reference time slot length. Sliding step size Temporary length Window overlap rate Adaptive Factor and Non-maximum suppression (NMS) radius coefficient and .

[0067] Generate time-limited window data and calculate local volatility. With event density Volatility is measured using the robust metric. The median (median(a)) is defined as... , This represents the median absolute deviation for the m-th time window. This measures the robustness of amplitude data within the window. median(a) is the median amplitude within the window.

[0068]

[0069] in, You can use the global MAD or the recent sliding reference. To prevent division by zero for extremely small positive numbers, Let x be the signal amplitude corresponding to the x-th point. Define the local event density. Choose either the normalized form of the above-threshold change point density or the zero-crossing density, or a weighted summation:

[0070]

[0071] in, The baseline local event density;

[0072] Depend on and By jointly adjusting the window length, an adaptive time slot can be obtained. :

[0073]

[0074] by As the length of the time period window, the center of the time period window =center( Time window center It is used to locate the center time of the m-th unit decision time slot in the two-dimensional coordinate system M2, center( ) is a function used to determine the center time of a time slot. Upper bound of time span Indicates the upper bound of the amplitude change. It is a time scalar. This is a limiting function used to restrict the value of a variable to a preset upper and lower limit. The time range for generating the m-th window, used as the step size. In the center of the current candidate window Around, with temporary length Sampling yields local fragments Temporary length After mapping the raw time-amplitude data D1 to the two-dimensional coordinate system M2, the center time point of each candidate window is... The transitional window length used when performing local statistical forecasting. It is only used to initially cut out a temporary local segment and obtain the result from D1. .

[0075] exist upper extraction window feature vector Window feature vector It includes four basic quantities: energy, gradient, spectrum, and entropy, forming window data. :

[0076]

[0077] in, For energy, For gradient, Let b be the probability of the b-th bucket in the histogram. The magnitude histogram entropy, The main spectral peak. Output window data. , It is a logarithmic function, where max represents the maximum value and f is the frequency. This indicates that a Fourier transform is performed on the data segment. The magnitude 'a' is the reciprocal of time. For the amplitude in a local segment The time is Data points.

[0078] Calculate anchor point scores for each window. and in the center of the window Generate initial anchor point :

[0079]

[0080] in, , , , These are the energy, gradient, and [other parameters] after 0-1 normalization based on global or recent statistics, respectively. and ; The entropy difference between adjacent windows is a possible value. For noise intensity estimation, such as the proportion of high-frequency energy or the inverse of the signal-to-noise ratio, , , , and These are the weighting coefficients for each item. Initial anchor point. , To suppress redundant peaks generated by dense windows, non-maximum suppression (NMS) along the time axis is employed, with a radius of... Adaptive to window length :

[0081]

[0082] exist Only retain within the scope The largest output is the suppressed anchor sequence. ,in, To suppress the time coordinates of the post-anchor sequence, To suppress the scoring of the post-anchor sequence, P is the feature vector of the suppressed anchor sequence, and P is the total number of anchor sequences retained after nonmaximum suppression.

[0083] It should be noted that, This usually corresponds to the amplitude value of the anchor point itself or the energy of the window it is in, and is used to evaluate the intensity of the event. The higher the value, the more the algorithm tends to select points with high energy and large amplitude as anchor points.

[0084] This typically corresponds to the local gradient or rate of change near the anchor point, used to evaluate the suddenness of an event. The higher the value, the more the algorithm tends to select points where the signal changes drastically, as these are often the starting points of events.

[0085] This typically corresponds to the reciprocal of the variance or standard deviation of the window containing the anchor point, used to evaluate the regularity of the signal's shape. A stable peak or trough has relatively small fluctuations around it. The higher the value, the more the algorithm tends to select anchor points that are clear in shape and not affected by too much noise.

[0086] Corresponding to It is used to evaluate the specific morphological characteristics of the main frequency. The higher the value, the more the algorithm focuses on finding anchor points that conform to a specific waveform pattern.

[0087] This corresponds to noise intensity estimation, used to evaluate inverse quantities such as the proportion of high-frequency energy or the signal-to-noise ratio. The lower the value, the more the algorithm tends to look for anchor points with low noise intensity.

[0088] Therefore, the time window data of S1 is... Suppressing the anchor sequence, i.e. , which serves as the input to S2.

[0089] Two-factor adaptive time slots simultaneously with volatility and event density Adjusting the window length allows for a wider time slot in the weak steady-state segment and a tighter time slot in the strong change segment, balancing detection rate and false alarm control. The time-varying non-maximum suppression radius varies with... Dynamic adjustment avoids oversuppression or undersuppression of a fixed radius in different dynamic scenarios. The unified score composed of multimodal window feature fusion scoring energy, gradient, entropy, spectrum, and noise inverse term enables the priority selection of verifiable and verifiable anchor points.

[0090] This step addresses the issue of local clock discrepancies at different monitoring points by deploying sensors and data acquisition devices matched to the measured object in the industrial field, combined with a time synchronization source to achieve time synchronization among multiple data acquisition devices. This prevents data analysis misalignment caused by time asynchrony. Simultaneously, protocol format normalization transforms multi-source, dimensionless data into a unified timestamp + amplitude value format, eliminating the risk of data format chaos. Subsequent cleaning processes, including rearranging, deduplication, missing value filling, extreme value suppression, and detrending, along with the construction of a time-amplitude two-dimensional coordinate system, two-factor adaptive time-slot sliding (widening time slots in weak steady-state segments and tightening time slots in strong change segments), and multimodal feature scoring, accurately capture key data from strong equipment changes while avoiding redundant acquisition in weak steady-state segments. It also filters noise and outliers, resulting in a highly accurate and low-redundancy output post-suppression anchor sequence. This provides high-quality, structured raw data support for subsequent high-correlation anchor point selection, effectively reducing the unnecessary computational costs of subsequent analysis.

[0091] Step S2: Input the suppressed anchor sequence into the associated fragment type judgment algorithm to obtain highly associated anchor data, and obtain the bookmark master table data through time interval expansion, hash value extraction, time correction and aggregate indexing.

[0092] like Figure 3 The diagram shown is a flowchart of the associated fragment type determination algorithm of this application. The associated fragment type determination algorithm is constructed based on generating highly associated anchor data from suppressed anchor sequences. Its purpose is to aggregate dense and scattered anchor points into a small number of semantically clear fragment anchor clusters, reducing false associations. Type labels are used to guide hash verification, cross-volume docking, and missing data filling. It both compresses the computational domain and improves interpretability and auditability. It serves as a crucial bridge to the overall bookmark data.

[0093] The specific steps for constructing a fragment type determination algorithm are as follows: input the relevant fragment type determination algorithm... Define the maximum candidate time difference Similarity weights θ1, θ2, θ3, θ4, and θ5 are used. The maximum candidate time difference is a preset time threshold. Only when the time difference between two anchor points is less than or equal to this threshold are they considered a valid candidate pair, and their five-dimensional similarity is then calculated. To avoid incurring rehashing overhead immediately before extracting hash values, the associated fragment type determination algorithm first... Internal computation lightweight fingerprint (For example, performing MinHash or SimHash on the window amplitude symbol string or low-dimensional spectral center sequence) for fast pre-screening.

[0094] Time proximity constraint With score significance constraint Filter candidate anchor pairs and output candidate anchor pairs. , To obtain the minimum value:

[0095]

[0096] in, Let i be the time position of the i-th anchor point. Let j be the time position of the j-th anchor point. It is the score of the i-th anchor point. It is the score of the j-th anchor point. This is the anchor point scoring threshold, used for filtering when constructing candidate anchor point pairs: only when... anchor point pair Only then will it be retained for subsequent similarity calculation and association determination.

[0097] For each candidate pair Calculate five-dimensional similarity and weighted composite association score Time fit ,in, It is an exponential function. For acceptable empirical expectation and dispersion; shape correlation Spectral coherence Anchor context consistency Contextual consistency can be measured, for example, by the relative order of adjacent anchors or the ratio of their scores; lightweight hash similarity. Lightweight hash similarity uses function refers to The similarity coefficient is used to measure the degree of similarity between two objects, with a value ranging from 0 to 1. The larger the value, the more similar the two objects are; 1 means they are completely identical; 0 means they are completely dissimilar.

[0098] Synthetic correlation score ;reserve The opposite side is built.

[0099] by As a node, based on existence Construct an undirected graph G with pairs of nodes as edges; perform connected component or modularity-based community detection on G to obtain anchor clusters C={u}. Summarize each anchor cluster u and select representative anchors. (Usually, scores are taken within the anchor cluster) Largest), preliminary time range Lightweight fingerprint sensor Merging (such as majority voting or bit-level merging), where, , These represent the minimum and maximum values ​​for the initial time range, respectively.

[0100] For each anchor cluster u, calculate the anchor cluster-level features and determine the associated fragment type, then output the type. With confidence level Four verifiable main types and their judgment criteria are given (the θ type is a threshold group):

[0101] In Type-SEQ sequential extended fragments, the time within the anchor cluster is strictly increasing and the intervals are stable. and The rules for determining high scores in both high-order and consistent-order scores are as follows: .in, It refers to the variance of the set of time intervals between adjacent anchor points, which is used to measure whether the time interval is stable; This represents the mean of shape correlation. The mean of spectral coherence; This represents the mean of anchor context consistency. It is the time stability threshold; only when the variance does not exceed it can the time interval be considered stable. It is the shape similarity threshold; the average value must not be lower than it. It is the spectral coherence threshold. This refers to the sequence consistency threshold. The time stability threshold is typically set between 0.08 and 0.12. The shape similarity threshold is typically set between 0.70 and 0.80. The spectral coherence threshold is typically set between 0.70 and 0.80. The sequence consistency threshold is typically set between 0.75 and 0.80.

[0102] The rules for determining whether there is temporal overlap or backtracking within the anchor cluster of retransmitted supplementary recordings (Type-RET), and whether the time series profiles of the supplementary recordings are basically the same, the frequency components and energy distributions are basically the same, and the hash similarity is high are as follows: .

[0103] Anchor clusters in cross-volume join segments (Type-XVL) or Adjacent to the volume boundary (the boundary is confirmed by subsequent verification in S3; this step approximates it with the estimated boundary or periodic boundary) and with obvious sequential extension characteristics: ; This indicates the start or end timestamp of the anchor cluster u; This refers to the boundary timestamp of the data volume; min{˙} refers to the absolute value of the minimum time difference between the start or end timestamp of the computing cluster and the volume boundary timestamp. This represents the threshold at time t of the volume-connected segment. This indicates a sequential extension feature.

[0104] For disordered segments with low sequential consistency, frequent time rewinding, and inconsistent time series profiles in the supplementary segments, and inconsistent frequency components and energy distribution, the following criteria are used: .in, The anchor context consistency threshold. The shape correlation threshold, The threshold value is the spectral coherence threshold, which is set by those skilled in the art based on experience.

[0105] Confidence This can be obtained by normalizing the support based on the rules. Generate highly correlated anchor data, and output highly correlated anchor data for each anchor cluster u. :

[0106]

[0107] in, For the ID of the anchor cluster, For the selected representative anchor point, This is a preliminary timeframe. For the type of associated fragments output, For confidence level, For lightweight fingerprints.

[0108] Afterwards, S2's time interval expansion, hash value extraction, time correction, and aggregated index can be performed. of Within the cycle, a strong hash (such as SHA-256) and high-precision clock drift estimation are performed on the complete original payload, and bookmark summary table data is generated.

[0109] The five-dimensional fusion association of time, shape, spectrum, context, and hash is used to participate in the determination with lightweight fingerprints before rehashing, significantly reducing the pre-computation; finally, strong hashing is used for confirmation to improve auditability. Graph model clustering plus rule-based interpretable type determination obtains stable anchor clusters, and transparent threshold rules are used to give the fragment type and confidence level, which facilitates the subsequent end-to-end verification in S3 and the visualization review in S4. Forward-looking annotation of cross-volume scenarios is applied in advance with Type-XVL candidate labels before entering S3, providing prior constraints for the generation of inter-volume connection record data and reducing the cost of subsequent backtracking correction.

[0110] Time interval expansion is performed on highly correlated anchor data. This is the set of anchor clusters output from the preceding steps. Each anchor cluster is represented by a record containing: anchor cluster number, representative anchor, time range within the anchor cluster, fragment type, type determination confidence level, and summary information from the lightweight fingerprint. The time range within the anchor cluster refers to the interval between the earliest and latest times of all anchors within that anchor cluster.

[0111] First, an adaptive expansion is performed on the original time range of each anchor cluster. The expansion consists of two parts: the first part equals the median of the time intervals between adjacent anchors within the cluster multiplied by a first expansion coefficient, which amplifies or reduces the contribution of the median time interval to the expansion. The empirical range for the first expansion coefficient is typically between 0.5 and 1.5. The second part equals the median absolute deviation of the time intervals multiplied by a second expansion coefficient, which is also typically between 1.5 and 3.0. To avoid expansion being too small or too large, the sum of the first and second parts is further restricted to a configurable minimum and maximum expansion value. Subsequently, the expanded start and end times are used in the original time-amplitude dataset. A slice is an expanded segment that uses the expanded start and end times as boundaries to precisely extract records within the time period from D1 while maintaining their original order, generating an expanded segment that can be used independently for hashing, time synchronization, and indexing.

[0112] Pair each anchor cluster with its corresponding extended fragment to obtain an extended fragment set. Subsequent strong hash extraction is performed based on this set.

[0113] Use the expanded fragment set obtained in the previous step Perform strong hash extraction. For each expanded segment, divide the segment evenly into several hash blocks according to the sampling rate and the desired granularity of evidence collection. Use the reciprocal of the interval between adjacent timestamps as the sampling rate. Based on the sampling rate of this segment, the desired duration of each hash block can be converted into the approximate number of records that each block should contain, thereby stably controlling the granularity and performance of the hash chain. The desired granularity of evidence collection is the target duration set for the evidence collection system, representing the minimum coverage duration of the evidence units that are expected to be independently generated and verified. The lower limit of the desired granularity of evidence collection should not be less than two effective sampling intervals and should not be less than the time required for the minimum number of statistical samples, so as to avoid too few samples in the block and statistical distortion. The upper limit of the desired granularity of evidence collection should not be greater than a reasonable multiple of the duration of a single typical event, to avoid a block spanning multiple independent events and weakening the resolution of evidence collection.

[0114] To establish an auditable hash chain, a seed block is first generated for the anchor cluster. This seed is obtained by concatenating the lightweight fingerprint of the anchor cluster with its number and then hashing it again. Next, when calculating the hash of the first block, the data of the first block is concatenated with the seed block in sequence, and then hashed to obtain the hash time of the first block. When calculating the hash of the second and subsequent blocks, the data of the current block is concatenated with the hash value of the previous block in sequence, and then hashed. This forms a hash chain where blocks are interconnected. To support subsequent inter-volume interfacing and integrity verification, both the fingerprint of the first block (i.e., the hash time of the first block) and the fingerprint of the last block (i.e., the hash time of the last block) are recorded.

[0115] Obtain the set of hash chains corresponding to each anchor cluster. This includes the hash sequence of each block, as well as the first and last block fingerprints of the anchor cluster.

[0116] For the expanded fragment set Perform time correction, by With the suppression of anchor sequences This leads to the introduction and application of a time synchronization model. It requires data from three sources: an expanded fragment set... Suppressing post-anchor sequences And a time correction reference set. The time correction reference set consists of multiple pairs of device times and reference times, used to estimate the drift relationship of the device clock relative to the reference clock.

[0117] An affine drift model is used to describe the mapping relationship between device time and reference time. The affine drift model can be understood as follows: the corrected time equals the original device time multiplied by a scaling factor plus a fixed offset. The scaling factor and offset are solved using robust regression, ensuring that the overall error is minimized and insensitive to outliers after the device time is mapped to the reference time via the affine drift model.

[0118] After obtaining the affine drift model, each sample timestamp in the expanded segment is corrected using the affine drift model. Simultaneously, the start and end times of each hash block are updated to have corrected time boundaries, resulting in a set of block-level tuples indexed by anchor cluster numbers and block numbers. Each tuple contains: anchor cluster number, block number, the start and end times of the block after time correction, and the corresponding hash value. This set of block-level tuples is used in the next step to construct the aggregated index.

[0119] Use the block-level tuple set obtained in the previous step. To support subsequent retrieval and integration by time, hash, and context, construct three types of indexes to generate a clustered index:

[0120] Time Index: The corrected block start time is used as the key, and the anchor cluster number and block number are used as the values, which facilitates quick positioning on the timeline.

[0121] Hash Index: Using block hash as the key and anchor cluster number and block number as the value, it facilitates integrity verification and integration.

[0122] Anchor context index: Anchor clusters or their adjacent summaries are used as keys, and anchor cluster numbers and related block number sets are used as values, which facilitates backtracking or linked retrieval based on contextual relationships.

[0123] Time indexes are recommended to be implemented using B+ trees or skip lists, hash indexes using hash tables, and anchor context indexes using inverted indexes.

[0124] The three types of indexes from the previous step are used to obtain the clustered index. According to the aggregated index Organize each block into a unified bookmark entry. Each bookmark entry contains: anchor cluster number, block number, start and end times after time synchronization, block hash, fingerprints of the first and last blocks of that anchor cluster, representative anchor, fragment type, and confidence level, etc.; use all bookmark entries and aggregated indexes. .

[0125] All bookmark entries are stably sorted in ascending order of their post-synchronization start time. When start times are the same, a secondary sort is performed based on the lexicographical order of the anchor cluster number and block number. After sorting, a forward and backward pointer are added to each bookmark entry, pointing to the previous and next block-level hashes of that entry, respectively, for chain-based consistency checks. A master bookmark table is then generated based on all bookmark entries. Bookmark master table data This serves as the direct input for subsequent end-to-end alignment verification.

[0126] This step combines a fragment type identification algorithm with lightweight fingerprint pre-screening. First, low-association anchors are quickly eliminated using low-overhead methods like MinHash, significantly reducing pre-calculation and avoiding the consumption of computing power by invalid data. Then, a weighted similarity score based on five dimensions—time, shape, spectrum, context, and hash—is used to accurately identify high-association anchors, greatly improving anchor association accuracy and reducing subsequent deviations caused by false associations. Subsequent time interval expansion to complete anchor context, SHA-256 strong hashing to ensure data immutability, and affine model correction for clock drift (controlling time deviations to the millisecond level), combined with aggregated indexes such as B+ tree time indexes and hash table integrity indexes, result in a highly structured bookmark master table. This provides clear anchor association criteria for S3 head-to-tail verification and significantly improves the efficiency of subsequent cross-volume data retrieval, greatly reducing the time cost of manually sorting out association relationships.

[0127] Step S3: Apply the bookmark master table data to perform head-to-tail alignment verification, generate inter-volume connection record data, fill in missing segments in the inter-volume connection record data, and then arrange them in chronological order to generate complete cross-volume directory data.

[0128] First and last alignment verification, using bookmark master table data. Each bookmark entry includes a volume identifier, indicating which volume the bookmark entry comes from; optionally, a page number may also be included.

[0129] For each adjacent volume containing bookmarks, a matching score is determined. For the preceding volume, several candidate bookmarks near the end are collected as the tail set; for the following volume, several candidate bookmarks near the beginning are collected as the head set. For any pair of tail and head bookmarks, a matching score is calculated. This score consists of three parts:

[0130] The first part is used to reflect whether the hashes are consistent. If the tail block fingerprint of the tail bookmark entry matches the first block fingerprint of the head bookmark entry, then a score is added to the pair of candidate bookmark entries; otherwise, no score is added.

[0131] The second part is designed to reflect temporal continuity. The smaller the time gap, the higher the score for this part; the larger the time gap, the lower the score for this part.

[0132] The third part reflects the consistency of order. If the relative order of adjacent contexts is consistent with the scoring ratio, the score for this part is higher; otherwise, it is lower.

[0133] After calculating scores for all candidate pairs, the pair with the highest score is selected as the optimal docking point. A configurable threshold is set, taking the highest score from the score distribution as the configurable threshold. When the score of the optimal docking point is higher than a configurable threshold, the candidate docking is considered successful; otherwise, it is marked as questionable.

[0134] For candidate pairs that meet the criteria, generate an inter-volume linkage record. Combine all inter-volume linkage records into a set to generate inter-volume linkage record data.

[0135] Each volume connection record Includes the volume number, the selected tail bookmarks and head bookmarks, the final score, and the success or doubt status.

[0136] Data recorded by inter-volume linking Bookmark summary table data Highly correlated anchor data Leading to the supplementary recording results To determine the docking location marked as questionable, it is necessary to first identify the gap interval at that location on the timeline. Then, in... The system searches for anchor clusters whose expanded time range overlaps with the gap interval, and prioritizes anchor clusters of type retransmission supplementation or sequential extension. Strong hash extraction, time correction, and aggregated indexing are then re-executed on these candidate anchor clusters within the gap interval to obtain bookmark entries for supplementation.

[0137] Consistency checks are performed on supplementary bookmark entries: If a supplementary bookmark entry exists that makes the last fingerprint of the supplementary entry consistent with the first fingerprint of the bookmark entry immediately adjacent to the right side of the gap, or significantly reduces the temporal discontinuity at that point to an acceptable level, then the gap is considered acceptable for supplementation; otherwise, it is considered unacceptable for supplementation. Each supplementary entry result... Indicate the time interval of the gap, the supplementary recording conclusion, and the bookmark entries used for supplementary recording.

[0138] Arranged in chronological order, by With supplementary recording results Leading to Insert all bookmark entries used for supplementary entries into the master bookmark table data, and then re-sort them stably according to the start time after time synchronization. After sorting, assign a global serial number to each bookmark entry. The sequence number of this serial number is equal to the position number of the corresponding bookmark entry in the master bookmark table data. At the same time, set cross-volume consecutive pointers for each bookmark entry, that is, link the current serial number with the adjacent preceding and following serial numbers to form complete cross-volume directory data. Each bookmark entry has a global serial number, original bookmark information or supplementary bookmark information, and a cross-volume consecutive pointer.

[0139] This step, when performing head-to-tail alignment verification using the bookmark master table data, combines three-dimensional scoring based on hash consistency, time continuity, and sequence consistency to determine cross-volume alignment. This accurately identifies misalignments between volumes (such as mismatched fingerprints between the tail and head blocks, or time gaps exceeding thresholds), significantly improving the initial alignment accuracy of cross-volume connection records and avoiding the misattribution errors common in traditional manual alignment. For questionable alignment locations, priority is given to retransmission and supplementary recording, followed by sequential extension of anchor clusters for supplementary recording. After consistency verification confirms the validity of the supplementary recording, the data gap filling rate between volumes can be significantly improved, effectively filling gaps in multi-volume storage of industrial data. Finally, the data is sorted by time after synchronization and a global serial number and continuous pointers are added. The resulting complete cross-volume directory data has a high degree of structure, eliminating the need to search through each volume when retrieving cross-volume data. The target data can be quickly located by the serial number or pointer. Compared to the traditional method of manually organizing cross-volume directories, efficiency is greatly improved, significantly reducing the manpower and time investment in data management.

[0140] Step S4: Construct two visualization views based on the complete cross-volume directory data. After generating a continuity certificate based on the two views, input the time interval threshold to detect the gap mark and verify the gap mark to obtain the verification result.

[0141] From complete cross-volume directory data This introduces the timeline view and the topology view.

[0142] The Vtime timeline view plots the start and end intervals of each bookmark entry on the timeline, following the global serial number sequence. Within each start and end interval, the fragment type, type confidence, and hash digest of the bookmark entry are labeled (only the first few characters are displayed for easy visual verification).

[0143] The Vtopo topology view places different volumes on different layers and pages on columns within the same layer. Each bookmark entry is drawn as a node, and nodes are linked together into chains using consecutive pointers across volumes. If a node originates from a supplement, it is highlighted with a dashed line or a special identifier, and its originating anchor cluster is indicated. The Vtime timeline view and the Vtopo topology view are used together for subsequent generation of continuity certificates and verification.

[0144] From complete cross-volume directory data Generate a continuity certificate. Construct a leaf hash for each bookmark entry. The input to the leaf hash is the following parts concatenated in sequence: global serial number, the block hash of the bookmark entry, and the start and end times after synchronization. Then, merge the hashes layer by layer upwards using a binary concatenation and rehashing method until a hash root is obtained. Combine the hash root with the time range covered by the complete cross-volume directory, and digitally sign the hash root and the time range covered by the complete cross-volume directory using the private key to form a continuity certificate. The continuity certificate proves both the continuous temporal coverage of the directory content and provides proof from the leaf hash to the hash root for any bookmark entry. The continuity certificate includes the hash root, the covered time range, a description of the hashing and aggregation method, and the private key signature.

[0145] Detect the time interval threshold. Use complete cross-volume directory data. For any two adjacent bookmark entries in the sequence, calculate the interval between the time-synchronized start time of the latter bookmark entry and the time-synchronized end time of the former bookmark entry. After collecting all adjacent intervals, calculate the median and median absolute deviation of these intervals. Then multiply the median by a first weight, and the median absolute deviation by a second weight, and add these two terms together to obtain the time interval threshold. The first and second weights are configurable coefficients for the median and median absolute deviation of adjacent time intervals, used to adjust their contribution to the final threshold. When an adjacent time interval exceeds the time interval threshold, the position between the corresponding adjacent bookmark entries is marked as a gap.

[0146] The first and second weights can be adjusted as needed. For example, the first weight can be 2.0 and the second weight can be 3.0; alternatively, the first weight can be 1.5 and the second weight can be 4.0. Different values ​​will change the sensitivity to sudden widening of the time gap.

[0147] Form a set of gap markers Each gap marker set The record includes: the global serial number position of adjacent bookmark entries, the actual interval value at that point, the threshold used at the time, and a Boolean conclusion on whether it was determined to be a gap.

[0148] The set of gap markers is reviewed and conclusions are drawn. Gap marker set Timeline view (Vtime), topology view (Vtopo), and inter-volume connection record data. To introduce.

[0149] For each location marked as a gap, three scenarios are reviewed. The scenario of "sufficient evidence but a business gap" refers to a situation where the last fingerprint of the adjacent bookmark entries matches the first fingerprint and has a high consistency score, but the time interval still exceeds the threshold. This usually means that the connection is not broken, but rather a natural gap in the business layer. The conclusion in this case is "continuous but gapped." A business gap refers to a situation where, within a certain time period, the business process itself does not produce transactions, events, or validly measured data output, but the data link and system operation are normal. A natural gap is a predictable and interpretable subtype of a business gap.

[0150] The situation where bookmarks can be repaired by adding them is when the gap has been marked as available for addition in the addition results. In this case, the bookmark is merged into the directory and the relevant pointers are updated, resulting in the gap being filled.

[0151] A situation where the connection is questionable refers to a situation where the corresponding inter-volume connection position is marked as questionable, and the sequence consistency is low or the hash connection is inconsistent, then the conclusion is that the link is discontinuous. In this case, it is recommended to reconfirm the time synchronization reference, appropriately expand the time interval expansion, or increase the search radius of the retransmission retrieval in order to further investigate the evidence.

[0152] Based on situations where there is sufficient evidence but a business gap exists, situations where bookmark entries are added and repaired, and situations where there are doubts about the integration, a re-verification record is generated. Each verification record includes: the location of the gap, the final conclusion (including one of three categories: continuous but gaps have been filled and discontinuous links), and evidence citations to support the conclusion (such as relevant bookmark entries, docking scores, and certificate proof paths).

[0153] This step uses a timeline view to annotate bookmark entries with synchronization intervals, fragment types, and hash digests. Combined with a topology view, nodes are linked by volume using pointers, allowing staff to intuitively grasp cross-volume data relationships and continuity status without needing to check raw data line by line, significantly improving the efficiency of data continuity visualization and understanding. When generating continuity certificates, a hash root merged from leaf hash binary searches and digitally signed ensures the directory content is tamper-proof, providing traceable evidence for industrial data compliance audits. The time interval threshold detection uses a weighted average of the median and median absolute deviation to calculate the threshold, avoiding misjudging business gaps with fixed thresholds and significantly reducing the gap mislabeling rate. Subsequent reviews are conducted based on gaps, recoverable entries, and discontinuities to accurately distinguish the causes of gaps. The repair rate of recoverable gaps is significantly improved, and the accuracy of locating discontinuity issues is greatly enhanced, significantly reducing the blind spots of manual verification and providing precise support for quality control of the industrial data chain.

[0154] Example 2

[0155] Reference Figure 4 This is the second embodiment of the present application, which provides an industrial data chain access system.

[0156] An industrial data chain access system includes: a raw data processing module, a highly correlated anchor point module, a cross-volume directory module, and a visualization view module.

[0157] The raw data processing module acquires time-amplitude raw data from industrial data chain monitoring points, constructs a two-dimensional coordinate system based on the time-amplitude raw data, sets unit judgment time slot parameters in the two-dimensional coordinate system and inputs the time slot sliding algorithm, and outputs time period window data. The time period window data is subjected to non-maximum suppression to obtain the suppressed anchor sequence.

[0158] The highly correlated anchor module obtains highly correlated anchor data by suppressing the input of the anchor sequence and the associated fragment type judgment algorithm. The highly correlated anchor data is then expanded over time intervals, hash values ​​are extracted, time is corrected, and aggregated indexes are applied to obtain the bookmark master table data.

[0159] The cross-volume directory module uses the bookmark master table data to perform first-to-last alignment verification, generates inter-volume connection record data, fills in missing segments of the inter-volume connection record data, and then arranges them in chronological order to generate complete cross-volume directory data.

[0160] The visualization view module constructs two visualization views based on the complete cross-volume directory data. After generating a continuity certificate based on the two views, it inputs a time interval threshold to detect gap marks and verifies the gap marks to obtain the verification results.

[0161] In the embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the displayed or discussed mutual couplings or direct couplings or communication connections may be through some communication interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

[0162] The embodiments of this application have been described above with reference to the accompanying drawings. However, this application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments under the guidance of this application without departing from the spirit and scope of protection of the claims. All of these variations are within the protection scope of this application.

Claims

1. A method for accessing industrial data chains, characterized in that, Includes the following steps: S1: Obtain raw time-amplitude data from industrial data link monitoring points, construct a two-dimensional coordinate system based on the raw time-amplitude data; set unit judgment time slot parameters in the two-dimensional coordinate system and input the time slot sliding algorithm to output time slot window data; the time slot window data is subjected to non-maximum suppression to obtain the suppressed anchor sequence. S2: The algorithm for determining the type of related segments after suppressing the input of the anchor sequence obtains highly related anchor data. The highly related anchor data is then expanded over time intervals, hash values ​​are extracted, time is corrected, and an aggregated index is applied to obtain the bookmark master table data. S3: Apply the bookmark master table data to perform head-to-tail alignment verification, generate inter-volume connection record data, fill in missing segments of the inter-volume connection record data and arrange them in chronological order to generate complete cross-volume directory data; S4: Construct two visualization views based on the complete cross-volume directory data, generate a continuity certificate based on the two views, input the time interval threshold to detect the gap mark, and verify the gap mark to obtain the verification result; The algorithm for determining the type of associated fragments in the suppressed anchor sequence input obtains highly associated anchor point data, including: A correlation fragment type determination algorithm is constructed. The input is the suppressed anchor sequence, and the maximum candidate time difference, similarity weight, correlation threshold and type determination threshold group are defined. The correlation fragment type determination algorithm first calculates a lightweight fingerprint in the window feature vector of the suppressed anchor sequence, and then filters candidate anchor pairs with time proximity constraints and score significance constraints, and outputs candidate pairs. For each candidate pair, calculate the five-dimensional similarity and weighted synthesize the association score. Construct an undirected graph with the suppressed post-anchor sequence as nodes and anchor pairs with association scores as edges. Perform connected component or modularity-based community detection on the undirected graph to obtain anchor clusters. Calculate anchor cluster-level features for each anchor cluster and determine the type of associated fragments, outputting the type and confidence level; output highly associated anchor data for each anchor cluster. The highly correlated anchor data, after time interval expansion, hash value extraction, time correction, and aggregated indexing, yields the bookmark master table data, which includes: Highly correlated anchor data is a set of anchor clusters output by anchor clusters. The original time range of each anchor cluster is adaptively expanded. The expanded start and end times are used to slice the original time-amplitude dataset to obtain the expanded segments of the anchor clusters. Each anchor cluster is paired with its corresponding expansion fragment to obtain an expansion fragment set. Strong hash extraction is then performed on the expansion fragment set to obtain a hash chain set corresponding to each anchor cluster. A time synchronization model is derived and applied from the set of expanded segments and the suppressed anchor sequence. An affine drift model is used to describe the mapping relationship between the device time and the reference time. The timestamp of each sample in the expanded segment is corrected according to the affine drift model, resulting in a set of block-level tuples indexed by the anchor cluster number and the block number. Multi-type indexes are constructed using a set of block-level tuples, including time indexes, hash indexes, and anchor context indexes. After multi-type indexes, a clustered index is obtained. Each block is organized into a unified bookmark entry based on the clustered index, and the total bookmark table data is formed based on all bookmark entries. The bookmark master table data undergoes a beginning-to-end alignment check to generate inter-volume linking record data. After missing segments are added to the inter-volume linking record data, it is arranged in chronological order to generate complete cross-volume directory data, including: Using the bookmark master table data, we determine the connection between adjacent volumes of bookmark entries and calculate the consistency score of candidate pairs. For candidate pairs that meet the criteria, we generate an inter-volume connection record. We then combine all inter-volume connection records into a set to generate inter-volume connection record data. In the highly correlated anchor data, find anchor clusters whose expanded time range overlaps with the gap interval, and select anchor clusters of type retransmission supplementation or sequential extension as priority candidate anchor clusters; re-execute strong hash extraction, time correction and aggregation index on the candidate anchor clusters within the gap interval to obtain bookmark entries for supplementation. The newly added bookmark entries are inserted into the bookmark master table data, and then sorted again according to the start time after synchronization. After sorting, a global serial number is assigned to each bookmark entry, and the sequence number of each global serial number is equal to the position number of the bookmark entry corresponding to the global serial number in the bookmark master table data. At the same time, cross-volume consecutive pointers are set for each bookmark entry, that is, the current global serial number is linked with the adjacent global serial numbers before and after it to form complete cross-volume directory data. Two visualization views are constructed based on the complete cross-volume catalog data. After generating a continuity certificate based on these two views, a gap marker is obtained by inputting a time interval threshold. The gap marker is then reviewed to obtain the verification results, including: The complete cross-volume catalog data leads to a timeline view and a topology view, which are used together to generate continuity certificates and verification records. A leaf hash is constructed for each bookmark entry, and these are merged upwards layer by layer to obtain a hash root; the hash root and the time range covered by the complete cross-volume directory are digitally signed using a private key to form a continuity certificate; Calculate the adjacent interval between two adjacent bookmark entries in the bookmark master table data, calculate the time interval threshold based on the adjacent interval, and mark the position between the corresponding adjacent bookmark entries as a gap when the adjacent interval is greater than the time interval threshold. For each location marked as a gap, a review is conducted according to three scenarios: cases where there is sufficient evidence but a business gap exists, cases where the gap can be repaired by adding bookmark entries, and cases where the connection is questionable; a re-inspection record is generated based on the review results.

2. The industrial data chain access method according to claim 1, characterized in that, The raw time-amplitude data acquired by the industrial data chain monitoring points includes: Real-time time-amplitude raw data is received from industrial data link monitoring points; each time-amplitude raw data consists of a timestamp and the corresponding signal amplitude, and the signal originates from industrial sensors. Periodically query sensor data, save the timestamp and corresponding signal amplitude to the raw data stream, forming data points of timestamp and signal amplitude.

3. The industrial data chain access method according to claim 2, characterized in that, The construction of a two-dimensional coordinate system based on the raw time-amplitude data includes: Using the acquired time-amplitude raw data, a two-dimensional coordinate system with X and Y axes is constructed. In the two-dimensional coordinate system with X and Y axes, the X-axis represents the timestamp and the Y-axis represents the signal amplitude. The data points of timestamp and signal amplitude are mapped onto a two-dimensional coordinate system to form a point set.

4. The industrial data chain access method according to claim 3, characterized in that, The step of setting unit judgment time slot parameters in a two-dimensional coordinate system and inputting a time slot sliding algorithm to output time period window data includes: A time-slot sliding algorithm is constructed. The unit judgment time slot parameter is defined in a two-dimensional coordinate system to generate time-slot window data. At the center of the current time-slot window, a local segment is obtained by sampling with a temporary length. The local volatility and event density are calculated based on the time window data. The length of the time window is adjusted by the local volatility and event density to obtain an adaptive time slot.

5. The industrial data chain access method according to claim 4, characterized in that, The time window data, after non-maximum suppression, yields the suppressed anchor sequence, which includes: Using an adaptive time slot as the time window length, the time range of the m-th window is generated. Window feature vectors are extracted from local segments to form window data. Anchor point scores are calculated for each window data, and an initial anchor point is generated at the center of the window. Non-maximum suppression is applied to the initial anchor point. Candidate anchor points in the neighborhood are merged using a suppression radius related to the window length. Only the candidate anchor point with the highest score is retained in the neighborhood of the same suppression radius. The candidate anchor points retained after non-maximum suppression constitute the suppressed anchor sequence.

6. An industrial data chain access system, used to implement the industrial data chain access method according to any one of claims 1 to 5, characterized in that, include: The module includes a raw data processing module, a highly correlated anchor point module, a cross-volume directory module, and a visualization view module. The raw data processing module acquires time-amplitude raw data from industrial data chain monitoring points, constructs a two-dimensional coordinate system based on the time-amplitude raw data, sets unit judgment time slot parameters in the two-dimensional coordinate system and inputs a time slot sliding algorithm, and outputs time period window data. The time period window data is subjected to non-maximum suppression to obtain a suppressed anchor sequence. The highly correlated anchor module suppresses the input of the anchor sequence to obtain highly correlated anchor data through the related fragment type judgment algorithm. The highly correlated anchor data is then expanded over time intervals, hash values ​​are extracted, time is corrected, and aggregated indexes are applied to obtain the bookmark master table data. The cross-volume directory module uses the bookmark master table data to perform first-to-last alignment verification, generates inter-volume connection record data, fills in missing segments of the inter-volume connection record data, and then arranges it in chronological order to generate complete cross-volume directory data. The visualization view module constructs two visualization views based on the complete cross-volume directory data. After generating a continuity certificate based on the two views, it inputs a time interval threshold to detect gap marks and verifies the gap marks to obtain the verification results.

Images