Data comprehensive treatment visual application system based on industrial internet

Abstract

The invention relates to the technical field of data visualization, in particular to an industrial internet-based data comprehensive treatment visualization application system, which comprises a node process extraction module, a data link organization module, a communication relation analysis module, an abnormal node treatment module and a comprehensive treatment display module. According to the method, the node sequence is set, the progressive trend is judged based on the time difference, the reverse nodes are screened out, the time sequence consistency is guaranteed, the node time interval continuity is checked in sequence, the node dependency relationship is dynamically modeled, the connection distortion risk is reduced, and the connection weak sections are screened; the method is advantaged in that connectivity abnormity identification efficiency is improved, node timestamp offset and dependence missing frequency are extracted, abnormity classification management is completed according to a standard production takt and a threshold value, abnormal node attribution is refined, traceability and treatment precision is improved, and industrial process data treatment depth, transparency and decision support capability are enhanced.

Description

Technical Field

[0001] The present invention relates to the field of data visualization technology, and in particular to a data comprehensive management visualization application system based on the industrial Internet. Background Art

[0002] The field of data visualization technology includes information processing methods that intuitively display data content and structure through graphics, images, interactive interfaces, etc. The core content is to classify, organize, and analyze large amounts of complex data, and use graphical means to present data relationships and changing trends in an intuitive and easy-to-understand manner. It covers data collection, data integration, data modeling, graphics rendering, interactive design and other aspects. It is mainly used in scenarios such as auxiliary decision analysis, data monitoring, pattern recognition, and trend forecasting, and improves data interpretation efficiency and decision support capabilities through visualization.

[0003] The data comprehensive governance visualization application system based on the Industrial Internet refers to the use of the Industrial Internet platform to aggregate various types of industrial production data, establish a unified data governance framework around technical issues such as data collection, data standardization processing, data fusion and integration, data consistency verification, and data classification management, and realize the visualization of data governance processes and results through the development of graphical interfaces. Specifically, it adopts methods such as data cleaning rule system construction, data fusion logic design, data standard unified modeling, and visual interactive layout design to standardize and structure the heterogeneous data generated in the Industrial Internet environment, and dynamically present the status of each link of data governance and the overall distribution of data through multi-dimensional graphical elements.

[0004] In the existing comprehensive data governance process, node sequence is often only arranged based on static time records, and there is a lack of progressive trend judgment and reverse node elimination steps, which leads to the inability to timely discover time-misaligned nodes, resulting in data stream crosstalk problems. The construction of dependency relationships mostly relies on preset rule matching, and lacks a consistency verification mechanism based on dynamic weights. It is easy for node connection relationships to be abnormal but cannot be accurately identified. Especially when it involves cross-process flows or heterogeneous equipment data, the phenomenon of node connectivity degradation is difficult to be warned in time, resulting in distorted process analysis and hidden connectivity defects. Abnormal node identification usually relies on a single time delay or fault signal threshold judgment, ignoring the compound abnormal characteristics of dependency chain breaks and time drifts, resulting in unclear classification and processing of potential broken chain nodes and deviation nodes, which can easily lead to delays in subsequent governance measures. For example, in a scenario where multiple batches of processes are run alternately, small node drifts and link losses cannot be detected in time, which in turn causes data governance gaps and affects overall production monitoring and decision-making accuracy. Summary of the Invention

[0005] The purpose of this invention is to solve the shortcomings of the existing technology and propose a data comprehensive management visualization application system based on the industrial Internet.

[0006] In order to achieve the above objectives, the present invention adopts the following technical solutions: The data comprehensive management visualization application system based on the industrial Internet includes:

[0007] The node process extraction module obtains the production process node records, sets the node sequence progression direction and calculates the time difference, determines the progression trend, sequentially detects the continuity of the node sequence time interval, and generates a process sequence association set;

[0008] The data link organization module establishes inter-node dependency based on the process sequence association set, compares the consistency of task dependency with the node's preceding node ID node by node, constructs an inter-node connection map and extracts the connection weight of dependency consistency to generate a node connection structure diagram;

[0009] The connectivity relationship analysis module arranges the node relationship pairs in descending order according to the connection weight according to the node connection structure diagram, selects weakly connected nodes and records the weakly connected node numbers, selects insufficiently connected areas according to the completeness of the dependent path between nodes, and generates a list of weakly connected sections;

[0010] The abnormal node governance module extracts the timestamp offset and dependency missing frequency of all nodes in the weak segment according to the list of connected weak segments, classifies and sorts the abnormalities, establishes deviation node groups and broken link node groups, and forms an abnormal node classification matrix.

[0011] As a further solution of the present invention, the process sequence association set includes node time difference judgment results, node progressive trend identification, time interval continuity verification identification, and reverse node number list; the node connection structure diagram includes task dependency mapping, predecessor node matching relationship, connection weight value list, and node connection relationship network; the weak connectivity section list includes a weak connection node number set, a weak connection path identification set, a dependency path integrity index, and a connectivity deficiency area number; the abnormal node classification matrix includes a deviation node number set, a broken link node number set, a timestamp offset statistics table, and a dependency missing frequency statistics table.

[0012] As a further solution of the present invention, the node process extraction module includes:

[0013] The timestamp difference calculation submodule extracts the first node timestamp and the second node timestamp based on the production process node records, calculates the corresponding time difference, performs a sign determination operation on the time difference, marks the time difference as a positive or negative trend, and generates a node progressive trend determination result;

[0014] The node sequence continuity determination submodule identifies the node number sequence according to the node progressive trend determination result, calculates the time interval values ​​in the node sequence in sequence based on the time difference between each pair of adjacent nodes, filters out the node sequences that do not meet the progressive direction continuity requirement, and generates the node time interval deviation;

[0015] The reverse node identification submodule screens and records all deviated node numbers based on the node time interval deviation, determines whether the corresponding time difference is less than 0, identifies the node numbers with reversed time direction, and obtains the process sequence association set.

[0016] As a further solution of the present invention, the data link organization module includes:

[0017] The dependency extraction submodule decomposes the structure of the predecessor relationship marked in each task record based on the process sequence association set, aggregates all predecessor node IDs into the dependency index field of the corresponding node, summarizes the one-to-many dependency chain between nodes based on the index field, clusters and merges multiple tasks that repeatedly point to the same predecessor ID, and obtains node dependency index records;

[0018] The connection mapping submodule compares the dependencies of each node with the preceding node IDs of all nodes based on the node dependency index records, performs a Boolean matching operation based on whether they are completely consistent, calculates the connection count values ​​between the nodes, and sorts them in descending order based on the matching counts to obtain the matching strength data between the nodes;

[0019] The weight extraction and generation submodule selects node pairs that are greater than the matching strength extraction benchmark value to establish connections based on the node matching strength data and the connection count value between each node pair. It then determines the node connection direction based on the connection direction information and assigns a weight identifier to the direction relationship in the connection channel to obtain a node connection structure diagram.

[0020] As a further solution of the present invention, the connectivity relationship analysis module includes

[0021] The weight sorting and screening submodule extracts the connection weights of all connection edges based on the node connection structure graph, sorts all weight values ​​in the set in descending order, sets a benchmark connection strength, and screens out node pairs with weights less than the benchmark connection strength to obtain a low-weight connection pair set;

[0022] The weak connection identification submodule performs a numbering operation on all node pairs whose edge weights are lower than the benchmark connection strength according to the low-weight connection pair set, counts the total number of connections that appear on each node in the set as the connection frequency, calculates the weak connection count value of node i, and filters to obtain a set of weak connection node numbers;

[0023] The path integrity determination submodule takes each node in the weakly connected node number set as the starting point, counts the actual number of paths, constructs a theoretical maximum connection path based on all dependent target nodes, marks structural weak areas, and generates a list of connected weak sections.

[0024] As a further solution of the present invention, the abnormal node management module includes:

[0025] The time offset screening submodule compares the time records of each node with those of the previous node based on the list of weakly connected sections, calculates the corresponding time interval and determines whether there is an overrun situation, marks the time offset abnormal nodes and records their numbers, and obtains a set of time offset abnormal node numbers;

[0026] The dependency missing statistics submodule checks the difference between the number of pre-dependencies of the current node and the actual number of existing connections based on the time offset abnormal node number set, synchronously filters and marks the dependency missing abnormal nodes, and performs a joint analysis with the time offset abnormal results to obtain an abnormal node classification index group;

[0027] The abnormal matrix construction submodule labels the abnormal nodes with classification labels according to the abnormal node classification index group, determines whether the node corresponding label is in a valid state, combines all nodes with the corresponding abnormal labels, and establishes an abnormal node classification matrix.

[0028] As a further embodiment of the present invention, the system further comprises:

[0029] The comprehensive governance display module draws a node flow chart on the visualization platform based on the abnormal node classification matrix and the node connection structure diagram. Normal nodes and normal paths are marked with solid lines, deviation nodes are marked in yellow, broken nodes are marked in red, and weak connection paths are displayed with dotted lines. This module comprehensively reflects the node sequence and connection integrity status and generates a data governance visualization overview diagram.

[0030] The data governance visualization overview diagram includes normal node identification, deviation node identification, broken link node identification, and weak connection path display.

[0031] As a further solution of the present invention, the comprehensive management display module includes:

[0032] The node style setting submodule classifies and sets node styles according to the abnormal node classification matrix and abnormal attributes, identifies normal node sets, deviation node sets, and broken node sets, assigns different graphic identification styles to the three types of nodes, and obtains a node style labeling dataset;

[0033] The path style mapping submodule determines whether there is a weak connection edge attribute between two associated nodes for each connection path in the structure diagram based on the node style labeling dataset and the node connection structure diagram, marks the weak connection path accordingly, integrates the path and style relationship, and obtains the connection path style dataset;

[0034] The graphics drawing output submodule loads the node style labeling dataset and the connection path style dataset into the preset structure of the visualization platform, constructs a graphics coordinate mapping according to the node position information, generates node graphics elements in sequence and assigns corresponding styles, draws path connections, and outputs a data governance visualization overview diagram.

[0035] Compared with the prior art, the advantages and positive effects of the present invention are:

[0036] In the present invention, by setting the node order and determining the progressive trend based on the time difference, the reverse nodes are screened out, the time sequence consistency is guaranteed, the continuity of the node time intervals is checked in sequence, the standardization of the process chain is improved, and the task dependency mapping is constructed based on the node sequence association. By extracting the connection weight and comparing the consistency, the node dependency relationship is dynamically modeled to reduce the risk of connection distortion. The weak connectivity sections are screened by connection weight sorting and path integrity analysis to improve the efficiency of connectivity anomaly identification, the node timestamp offset and the dependency missing frequency are extracted, and the anomaly classification management is completed according to the standard production rhythm and threshold. The attribution of abnormal nodes is refined, the traceability and governance accuracy are improved, and the node verification, link modeling, connectivity analysis and anomaly classification links are connected to enhance the depth, transparency and decision support capabilities of industrial process data governance. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 is a system flow chart of the present invention;

[0038] Figure 2 This is a flowchart of the node process extraction module of the present invention;

[0039] Figure 3 This is a flow chart of the data link organization module of the present invention;

[0040] Figure 4 This is a flow chart of the connectivity analysis module of the present invention;

[0041] Figure 5 This is a flow chart of the abnormal node management module of the present invention;

[0042] Figure 6 This is a flow chart of the comprehensive management display module of the present invention. DETAILED DESCRIPTION

[0043] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0044] In the description of the present invention, it should be understood that the terms "length," "width," "up," "down," "front," "back," "left," "right," "vertical," "horizontal," "top," "bottom," "inside," "outside," and the like, indicating positions or relationships, are based on the positions or relationships shown in the accompanying drawings and are intended only to facilitate the description of the present invention and simplify the description. They do not indicate or imply that the devices or elements referred to must have a specific orientation, be constructed, or operate in a specific orientation. Therefore, they should not be construed as limiting the present invention. Furthermore, in the description of the present invention, "plurality" means two or more, unless otherwise expressly and specifically defined.

[0045] See also Figure 1 The data comprehensive governance visualization application system based on the Industrial Internet includes:

[0046] The node process extraction module obtains the timestamp, process stage number, and production batch ID from the production process node records, sets the node sequence progression direction, calculates the time difference based on the timestamps of the first and second nodes, determines the progression trend by the time difference sign, sequentially detects the continuity of the node sequence time interval, selects nodes with reverse time differences, records the node numbers, and generates a process sequence association set.

[0047] The data link organization module extracts the task dependencies and predecessor node IDs in the nodes based on the process sequence association set, establishes inter-node dependencies, compares the consistency of task dependencies with the predecessor node IDs of other nodes, constructs an inter-node connection map based on the matching count, extracts the connection weights of the dependency consistency, and generates a node connection structure diagram;

[0048] The connectivity analysis module sorts node pairs in descending order of connection weight based on the node connection structure diagram, selects node relationships with connection weights lower than the benchmark connection strength (determined by the average connection weight of node dependencies in historical data; when the actual connection weight is lower than this value, it is considered a weak connection) and marks them as weakly connected nodes. It also records the weakly connected node numbers and screens out areas with insufficient connectivity based on the completeness of the dependency paths between nodes (obtained by calculating the ratio of the actual number of dependency paths between nodes to the theoretical maximum number of paths, reflecting the integrity of the topological structure) to generate a list of weakly connected sections.

[0049] The abnormal node management module extracts the timestamp offsets and dependency loss frequencies of all nodes in the weak segment based on the list of connected weak segments. It then classifies and sorts out the nodes whose timestamp offsets are greater than the standard production cycle threshold (the standard process interval set according to the production line design parameters, obtained through the equipment process manual or production scheduling system) and whose dependency loss frequencies exceed the preset alarm threshold (the abnormal trigger conditions set according to industrial quantity management standards, usually determined through historical fault data analysis). It then establishes deviation node groups and broken link node groups, and forms an abnormal node classification matrix.

[0050] The comprehensive governance display module draws a node flow chart on the visualization platform based on the abnormal node classification matrix and the node connection structure diagram. Normal nodes and normal paths are marked with solid lines, deviation nodes are marked in yellow, broken nodes are marked in red, and weak connection paths are displayed with dotted lines. It comprehensively reflects the node sequence and connection integrity status, and generates a data governance visualization overview.

[0051] The process sequence association set includes the node time difference judgment results, node progressive trend identification, time interval continuity test identification, and reverse node number list. The node connection structure diagram includes task dependency mapping, predecessor node matching relationship, connection weight value list, and node connection relationship network. The list of weak connectivity sections includes a weak connection node number set, a weak connection path identification set, a dependency path completeness index, and an insufficient connectivity area number. The abnormal node classification matrix includes a deviation node number set, a broken link node number set, a timestamp offset statistics table, and a dependency missing frequency statistics table. The data governance visualization overview diagram includes normal node identification, deviation node identification, broken link node identification, and weak connection path display method.

[0052] See also Figure 2 , the node process extraction module includes:

[0053] The timestamp difference calculation submodule extracts the timestamp of the first node and the timestamp of the second node based on the production process node records, calculates the corresponding time difference, performs a sign determination operation on the time difference, marks the time difference as a positive or negative trend, and generates a node progressive trend determination result;

[0054] Based on the production process node records, the initial data extraction operation is performed on all nodes. The extracted data structure is shown in Table 1, including the node number, the first node timestamp, the second node timestamp, the process stage number and the production batch ID. Among them, nodes N001 to N005 have the first and last timestamps (100, 150), (150, 190), (190, 230), (230, 280) and (280, 320), respectively, belonging to different batches and process stages. When performing the timestamp difference calculation, the first node timestamp is subtracted from the corresponding second node timestamp to obtain the time difference and judge its positive or negative. The calculated differences are -50, -40, respectively. , -40, -50, -40 seconds. Since all time differences are negative, it can be determined that the progressive trends of all nodes are positive. In this process, taking node N001 as an example, the first node timestamp is 100 seconds, the second node timestamp is 150 seconds, and the difference is -50, which means that the node process segment advances backward by 50 seconds. It is applied to the advancement record of a certain product along the process flow to confirm that there is no reverse processing phenomenon. If a certain section in the future nodes, such as N004 to N005, undergoes time reversal, for example, the second node time is earlier than the first node, it will be recorded as a reverse trend node. All time difference directions in this section are negative, and time reversal does not occur, and finally the node progressive trend judgment result is generated.

[0055] Table 1 Production process node record table

[0056] Node number First node timestamp(s) Secondary node timestamp(s) Process stage number Production batch ID N001 100 150 1 B001 N002 150 190 2 B001 N003 190 230 2 B002 N004 230 280 3 B002 N005 280 320 3 B003

[0057] As shown in Table 1, the recorded time of each node is coherent and the time progression direction is consistent.

[0058] The node sequence continuity determination submodule identifies the node number sequence according to the node progressive trend judgment result, calculates the time interval values ​​in the node sequence based on the time difference between each pair of adjacent nodes, filters out the node sequences that do not meet the progressive direction continuity requirements, and generates the node time interval deviation;

[0059] According to the result of node progressive trend judgment, the progressive continuity of the node time series is judged based on the time difference. The specific operation is to calculate the time difference change between adjacent node segments in sequence and process it in the absolute difference method. For example, the N001-N002 segment is 50 seconds, and the N002-N003 segment is 40 seconds. The difference is 10 seconds, which is greater than the set progressive direction continuity benchmark value of 5 seconds. It is determined that there is a progressive break in this segment. Similarly, the difference between the N003-N004 segment and the N004-N005 segment is only 10 seconds, but since the actual set benchmark value is 5 seconds, they are all abnormal and set. The reference of this benchmark value is based on the empirical statistics of the average continuous fluctuation of the actual process segment. After experimental collection and monitoring of 50 batches, the average time fluctuation value between segments was 4.6 seconds, and the standard deviation was 1.2 seconds. Therefore, the continuity benchmark value was set to 5 seconds. In the above example, all segment differences exceeded the 5-second benchmark and were recorded as deviations from the node segment. For those that did not exceed the benchmark, such as the time difference between two segments was 45 seconds and 47 seconds respectively, with a difference of 2 seconds, they were considered continuous. Combined with the above judgment, all segments between nodes were judged, and finally the node segments that deviated from the benchmark trend were extracted and marked to generate the node time interval deviation.

[0060] The reverse node identification submodule filters and records all deviated node numbers based on the node time interval deviation, determines whether the corresponding time difference is less than 0, identifies the node numbers with reversed time direction, and obtains the process sequence association set;

[0061] Based on the node time interval deviation, all node segments that have been marked as deviating from the trend are re-screened to determine whether time reversal occurs. Specifically, the original time difference data is extracted for sign judgment. If the time difference is positive, the secondary node is earlier than the first node, which is a time reversal. For example, the original difference between nodes N004 and N005 is -40 seconds, maintaining normal progression. However, if the acquisition result is updated to 280 seconds for the first node and 260 seconds for the secondary node, the time difference is +20 seconds, indicating a time reversal. Combined with the process stage number 3 of this node segment and the production batch B003, its sequential validity in the batch process needs to be determined. In actual scenarios, if multiple nodes in the batch are found to be reversed, it can be regarded as a large batch disturbance and should be given a higher recognition priority. Combined with the batch disturbance weight and the process stage influence distribution, the identified reverse node set is numbered and marked, and the node identification results are summarized to finally obtain the process sequence association set.

[0062] See also Figure 3 , the data link organization module includes:

[0063] The dependency extraction submodule decomposes the structure of the predecessor relationship marked in each task record based on the process sequence association set, aggregates all predecessor node IDs into the dependency index field of the corresponding node, summarizes the one-to-many dependency chain between nodes based on the index field, clusters and merges multiple tasks that repeatedly point to the same predecessor ID, and obtains the node dependency index record;

[0064] Based on the process sequence association set, all node IDs are extracted, and their corresponding task dependencies and predecessor node IDs are read one by one. First, taking node N101 as an example, its task dependency is [N100] and its predecessor node ID is N100. This information is constructed into a one-to-one dependency chain. Node N102 is processed continuously, and its dependency is [N101] and its predecessor node ID is also N101. Node N103 has multiple dependencies, namely N101 and N102, forming a one-to-many dependency structure. Multiple dependencies need to be structurally split and expanded into multiple connection paths. At the program processing level, the structure mapping is completed by decomposing the task dependency field array and corresponding it with the key value of the predecessor ID field. Then, nodes N104 and N105 are processed and their dependency structures are established in turn. Finally, the original task dependency structure table is obtained. The specific data structure is shown in Table 2. After structural cleaning of all node records, duplicate dependency paths are removed and the field format is unified. The records are converted into a standardized node-dependency structure dictionary for subsequent mapping and comparison, and finally the node dependency index records are generated.

[0065] Table 2 Example of task dependency structure

[0066] Node ID Task dependencies Predecessor node ID N101 [N100] N100 N102 [N101] N101 N103 [N101,N102] N102 N104 [N103] N103 N105 [N104] N104

[0067] As shown in Table 2, the multi-dependency structure of node N103 reflects a dependent and parallel task relationship, while the remaining nodes are one-to-one sequential dependency relationships.

[0068] The connection mapping submodule compares the dependencies of each node with the IDs of the preceding nodes in all nodes according to the node dependency index records, and performs a Boolean matching operation based on whether they are completely consistent. The formula is:

[0069]

[0070] Calculate the connection count C between nodes i and j ij , sorted in descending order according to the matching count, and obtained the matching strength data between nodes, where A ik represents the k-th dependency encoding value of the i-th node, B jk represents the encoding value of the kth preceding node of the jth node, D k is the frequency coefficient of the k-th dependency in all nodes, T kis the disturbance factor of the process stage where the k-th dependent task is located, and K represents the total number of dependencies;

[0071] After obtaining the node dependency index record, the task dependencies of each node are compared with the predecessor node IDs of other nodes one by one to determine whether there is a mappable connection relationship. For each pair of nodes, the connection strength measurement index is constructed by calculating the numerical matching difference between its dependencies and predecessor IDs. Taking the connection relationship between node N103 (depending on N101 and N102) and node N102 (predecessor node is N102) as an example, the assignment is as follows:

[0072] A i1 =101,A i2 =102 (N103 depends on N101 and N102);

[0073] B j1 =102,B j2 =102 (the preceding node of N102 is N102);

[0074] D1 = 0.2, D2 = 0.3, which means that the frequency of N101 is 20% and that of N102 is 30%;

[0075] T1=1.0, T2=1.2, which means the process disturbance factors of the two dependent items are 1.0 and 1.2 respectively;

[0076] Substitute the above data into the formula and calculate item by item:

[0077] Item 1 (dependency is N101):

[0078]

[0079] Item 2 (dependency is N102):

[0080]

[0081] The summation result is:

[0082] C ij =1.382+0=1.382;

[0083] Therefore, the node connection count value between nodes N103 and N102 is 1.382, which shows that N103 has a partial dependency connection with N102, and the difference is corrected by the frequency and disturbance factors to construct the connection strength ranking between nodes. The lower the value, the higher the connection matching degree.

[0084] The node connection count value is used to quantify the structural connection strength between any two nodes at the task dependency level. Its specific significance lies in measuring the degree of numerical matching between the dependency of one node and the predecessor identifier of another node. This value not only reflects the degree of overlap of the task dependency structure between nodes, but also comprehensively considers the frequency of occurrence of the dependency in all nodes and the degree of disturbance in the process stage in which it is located. The smaller the value, the higher the correspondence between the two nodes in the dependency structure, that is, the node is more likely to serve as the predecessor execution object of another node. The node connection count value can be used as an important basic reference indicator for the subsequent generation of connection channels, the assignment of edge weights and the construction of node graphs. It has a core discriminant significance in actual process analysis and task chain reasoning.

[0085] The operational logic of the formula aims to comprehensively measure the degree of matching between two nodes in the dependency structure and the intensity of their fluctuations affected by external factors. The dependency coding difference is used as the basic difference indicator in the formula, and the coding difference between nodes is reflected by subtraction. The smaller the difference, the closer the match. The difference is then divided by "1 plus the square root of the dependency frequency" to downgrade the dependencies that appear frequently. That is, the higher the frequency, the weaker the uniqueness of the connection relationship. Therefore, the denominator is taken and then added with one for reduction. At the same time, in order to enhance the consideration of the impact of the process stage, the difference is multiplied by the value of the disturbance factor plus one, indicating that in the high process disturbance stage, the difference between nodes may be more significant, so the matching cost increases. Finally, the absolute value of each calculation result is taken to eliminate the interference caused by the positive and negative directions of the coding difference. The matching items of all dependencies are finally summarized and aggregated to form a node connection count value, thereby realizing the comprehensive superposition of the three factors of coding difference, dependency rarity and process uncertainty.

[0086] The weight extraction and generation submodule selects node pairs with a connection count greater than the matching strength extraction benchmark value based on the node matching strength data and the connection count value between each node pair to establish connections. It then determines the node connection direction based on the connection direction information and assigns a weight identifier to the direction relationship in the connection channel to obtain a node connection structure diagram.

[0087] According to the matching strength data between nodes, the connection strength values between all node pairs are statistically distributed, the mean μ and the standard deviation σ of the overall matching value set are calculated, assuming that the mean is 1.8 and the standard deviation is 0.5, then set the matching strength extraction reference interval as μ+σ=2.3, filter out the node pairs with strength values higher than 2.3 as strong connection channel candidates, then calculate the 85th percentile value of all matching strengths, for example, assuming that the 85th percentile of the matching value set is 2.1, then take it as the strong connection determination threshold, select the node pairs with matching values greater than 2.1 into the connection set, and then determine the connection direction combined with the dependence direction information between nodes, record the connection channel direction as the dependent node pointing to the depended node, for example, N103 depends on N102, then the connection direction is N103→N102, assign a weight value to each connection channel, normalize the corresponding matching strength value by dividing it by the maximum value of the set, assuming that the maximum matching value is 3.0, if the strength value of N103→N102 is 2.4, then the normalized value is 0.8, which is recorded as the weight value of the connection channel, assign weights to all connection edges and identify the direction, and construct a complete network connection structure diagram, finally obtain the node connection structure diagram.

[0088] Please refer to Figure 4 , the connectivity relationship analysis module includes

[0089] The weight sorting and screening submodule extracts the connection weights of all connection edges based on the node connection structure diagram, sorts all weight values in the set in descending order, sets a reference connection strength, and filters out node pairs with weights less than the reference connection strength to obtain a low-weight connection pair set.

[0090] Based on the node connection structure diagram, the connection weights corresponding to all edges in the diagram are extracted, and each connection pair is numbered and sorted. The extracted edge weights are sequentially formed into an initial weight sequence. For example, the connection weights of the node pair N201→N202 are 0.43, N202→N203 are 0.55, N203→N204 are 0.60, N204→N205 are 0.49, and N205→N206 are 0.52. These five groups of connection pairs form the basic data source for sorting. Subsequently, the connection structure of all project nodes in the historical data is collected, and their connection weights are extracted and summarized. The mean μ and standard deviation σ of these historical connection weights are calculated. For example, from 10 The statistics of μ=0.61 and σ=0.05 from 0 historical connection edges are used, so the benchmark connection strength threshold is set to μ-σ=0.56. After setting this value as the screening standard, all connection pairs with weight values ​​less than 0.56 in the current structure are screened. For example, in the above sample, N201→N202(0.43), N202→N203(0.55), N204→N205(0.49), and N205→N206(0.52) are all lower than this threshold and are classified as low-weight connection pairs. All qualified connection relationships are summarized to build a list, and the corresponding node pair numbers and edge directions are recorded to generate a low-weight connection pair set.

[0091] The weak connection identification submodule numbers and marks all node pairs whose edge weights are lower than the benchmark connection strength based on the low-weight connection pair set, and counts the total number of connections that appear in each node in the set as the connection frequency, using the formula:

[0092]

[0093] Calculate the weak link count value W of node i i Then, all nodes are sorted from large to small according to the value, candidate weak connection nodes are screened, and the weak connection node number set is obtained, where R ij is the connection weight between the i-th node and the j-th node, is the average weight of all connections of node i, R max,i is the maximum weight that appears in the connection of node i, Z i is the number of connections that node i appears in the low-weight set, and m is the number of adjacent nodes connected to node i in the low-weight set;

[0094] According to the set of low-weight connection pairs, the starting and ending nodes corresponding to all edges are numbered and identified to form a node set for statistics. The frequency of each node appearing in the low-weight connection set is counted as the connection frequency parameter. At the same time, the average connection weight and maximum connection weight of all connected edges of the node are extracted to form a set of participating variables for weak connection identification. The specific values ​​are then included in the parameter table required for the identification operation. The reference structure of this process is as follows:

[0095] Table 3 Weak link identification parameters

[0096]

[0097]

[0098] As shown in Table 3, each record contains the edge connection characteristics of the node. By substituting the above data into the formula, the node N201 is substituted into the calculation: the node connection weight R ij =0.43, average weight Maximum weight R max,i =0.72, connection frequency Z i =3, the number of adjacent nodes m = 1, substitute into the formula:

[0099]

[0100] In the same way, the parameters of nodes N202 to N205 are substituted and the results are calculated to obtain the weak connection count value of each node. Then all the results are sorted and the top 15% of the nodes are taken. Assuming that the total number of nodes is 50, the top 7 values ​​are retained. Combined with the structure diagram, it is retrieved whether these nodes are in a state of no alternative connection or only isolated connection. If the conditions are met, they are marked as weakly connected nodes. For example, node N201 is only connected to N202 in the graph, and the edge weight is lower than the benchmark strength, then it is a weakly connected endpoint. Finally, its number is screened out to obtain the weakly connected node number set.

[0101] The node weak connection count value is used to measure the weak connection tendency of a node in the connection network. It is a comprehensive value formed by counting the fluctuation degree, deviation and structural distribution characteristics of the connection weights between the node and all its adjacent nodes. Specifically, the larger the count value, the more it means that the node has multiple edges with connection weights below the average level, and these edges are significantly different from their historical strongest connections, and frequently appear in the low-weight area of ​​the connection graph, indicating that the path stability of the node in the entire topological structure is poor, and it may be at the end of the structure, sparsely connected areas or isolated edges. It is an important basis for judging whether it has weak connection properties in the graph and can be used as a core judgment indicator for subsequent screening, connectivity analysis, topology optimization and other operations.

[0102] The operation logic used in this formula is based on the multi-dimensional structural features required for weak connection identification. The absolute value of the difference between the connection weight and the average connection weight between node i and its adjacent node j is calculated to measure whether there is local attenuation in the connection. The larger the difference, the greater the deviation of the connection from the average connection state. This term uses the difference between the maximum connection weight of node i and the current weight as the attenuation measure of the connection strength, and the Z in the denominatori +1 introduces a connection frequency suppression mechanism, that is, the greater the connection frequency, the weaker the attenuation effect is diluted, thereby controlling the influence range of frequently appearing nodes. The sum of the two reflects the degree of local structural degradation of each connection edge. Finally, the multi-path information is reduced by summing and averaging the results of all adjacent nodes. This mean is used as the expression of the weak connection strength of the node as a whole for comparison and screening. In a structural sense, it comprehensively reflects the coupling relationship between "local deviation of a single edge" and "connection stability within the entire graph."

[0103] The path integrity determination submodule collects all dependent paths in the connection structure graph based on the weak connection node number set, starting from each node in the number set, and counts the actual number of paths. It then constructs a theoretical maximum connection path based on all dependent target nodes, calculates the path ratio to obtain the completeness of the dependent path, compares the completeness threshold to mark the weak structural areas, and generates a list of connected weak sections.

[0104] According to the set of weakly connected node numbers, a dependent path graph structure is constructed for each weakly connected node. The path combinations between it and all connected nodes are collected, and the actual number of connection paths N1 is counted. Then, the theoretical maximum number of paths N2 is calculated based on the theoretically connectable combinations between nodes. Duplicate nodes are prohibited during the path construction process. The maximum connection graph structure is constructed using backtracking logic. According to the formula completeness = N1 ÷ N2, the path completeness index of the node in the structure graph is calculated. For example, node N201 is actually connected to 2 paths, and the theoretical maximum number of paths is 6, so the completeness is 2 ÷ 6 = 0.333. Then, the set of nodes with a completeness less than 0.45 is included, and the structure graph is mapped based on this set. The subgraph area to which it belongs is extracted and the area number is recorded to obtain a list of weakly connected sections.

[0105] See also Figure 5 , the abnormal node governance module includes:

[0106] The time offset screening submodule compares the time records of each node with those of the previous node based on the list of weakly connected segments, calculates the corresponding time intervals and determines whether there are any overruns. It then marks the nodes with abnormal time offsets and records their numbers, obtaining a set of node numbers for abnormal time offsets.

[0107] Based on the list of weakly connected segments, the current timestamp of each node is extracted from its predecessor's timestamp data. The intervals are compared one by one to obtain the actual time difference between the processes experienced by the node. Then, a difference judgment is performed based on the standard cycle time set in the production line process manual. The judgment criterion is whether the actual time interval exceeds the maximum interval tolerance limit defined by the standard cycle time. During the execution process, the timestamp difference of each group of nodes is used as the basic participating item, and each item is compared and analyzed based on the standard cycle time. For example, the current timestamp of node N301 is 1250, and the timestamp of its predecessor is 1120. The calculated process interval is 130. If the standard cycle time is 120, the offset value is 10. If it exceeds the benchmark, the node is determined to be a time offset abnormal node. Nodes N302 to N305 are processed in the same way, and their time differences are compared with the standard cycle time to see if there is any offset. The numbers of all nodes with offset phenomena are counted, and the result set structure is constructed as shown below:

[0108] Table 4 Abnormal node parameter example table

[0109] Node number Current timestamp Predecessor timestamp Standard tempo N301 1250 1120 120 N302 1320 1210 120 N303 1480 1340 120 N304 1600 1500 120 N305 1780 1600 120

[0110] As shown in Table 4, the actual intervals of multiple nodes have exceeded the standard beat, which constitutes the basis for determining the time abnormal nodes and finally obtains the time offset abnormal node number set.

[0111] The dependency loss statistics submodule checks the difference between the number of pre-dependencies of the current node and the actual number of existing connections based on the time offset abnormal node number set, simultaneously filters and marks the dependency loss abnormal nodes, and jointly analyzes them with the time offset abnormal results to obtain the abnormal node classification index group;

[0112] According to the time offset abnormal node number set, its logical dependency information is extracted node by node in the structure diagram, and the number of dependency connections that should theoretically exist in the stage to which it belongs is counted. The actual number of connections is extracted from the current structure diagram. The dependency missing frequency is calculated based on the two values. The actual number of connections of each node is subtracted from the expected number of dependencies as the missing frequency. The missing frequency is compared with the preset alarm threshold in the industrial quality standard for judgment. If the missing frequency is greater than the alarm value, the node is marked as a dependency missing abnormal node. Taking node N301 as an example, its expected dependency is 3, but the actual dependency is only 1, so the missing frequency is 2. If the system sets the threshold to 1, N301 meets the abnormal judgment condition. The judgment of all nodes is completed in sequence, and the extended fields in Table 4 are supplemented as follows:

[0113] Table 5 Supplementary Table

[0114] Node number Required number of dependencies Actual number of dependencies N301 3 1 N302 2 2 N303 3 1 N304 4 2 N305 2 1

[0115] In this way, abnormal nodes with dependent missing frequencies are further screened out, and intersection classification is performed in combination with the time offset results to obtain the abnormal node classification index group.

[0116] The abnormal matrix construction submodule labels abnormal nodes according to the abnormal node classification index group, determines whether the corresponding label of the node is in a valid state, combines all nodes with the corresponding abnormal labels, and establishes the abnormal node classification matrix;

[0117] According to the abnormal node classification index group, the abnormal attributes of the nodes are two-dimensionally identified to construct the abnormal node status matrix. The node number is used as the unique index row, and the abnormal type is used as the dimension column to set the identification mark bit. Each node is judged whether it meets the abnormal standard in two dimensions. If a node is marked as time offset and dependency missing at the same time, it is marked as valid in the corresponding dimension, otherwise it is invalid. After executing the identification construction process, a complete two-dimensional structure table is generated for abnormal aggregation judgment and governance strategy optimization, and finally the abnormal node classification matrix is ​​established.

[0118] See also Figure 6 , the comprehensive governance display module includes:

[0119] The node style setting submodule classifies and sets node styles based on the abnormal node classification matrix and abnormal attributes, identifies normal node sets, deviation node sets, and broken node sets, assigns different graphic identification styles to the three types of nodes, and obtains a node style labeling dataset;

[0120] According to the abnormal node classification matrix, the abnormal status of all nodes is extracted, and the node numbers and their corresponding flags are read one by one to establish classification rules: nodes without any abnormality are defined as normal nodes, nodes with time offset marks are defined as deviation nodes, and nodes with dependency missing marks are defined as broken nodes. Different types of nodes are given differentiated graphic identifiers, among which normal nodes use the default gray solid circle icon, deviation nodes are set as yellow circular elements, and broken nodes are set as red circular elements. During the execution process, the node numbers and node types are mapped one by one to construct a style dictionary and output as a node style configuration file. For example, nodes N001 and N004 are classified as normal, N002 and N005 are classified as deviations, and N003 is classified as broken links. The node style archiving results are as follows:

[0121] Table 6 Node type identification table

[0122] Node number Node Type Graphic Identity N001 normal Gray circle N002 deviation Yellow circle N003 Broken Chain Red circle N004 normal Gray circle N005 deviation Yellow circle

[0123] As shown in Table 6, the pairing configuration between node type and graphic identifier has been completed, which is used as the node drawing input structure to finally obtain the node style labeling dataset.

[0124] The path style mapping submodule uses the node style labeling dataset and the node connection structure diagram to determine whether there is a weak connection edge attribute between the two associated nodes for each connection path in the structure diagram, mark the weak connection path accordingly, integrate the path and style relationship, and obtain the connection path style dataset;

[0125] According to the node style labeling dataset and the path connection information extracted from the node connection structure diagram, the starting point and end point of each path are analyzed. The display style corresponding to the path is determined according to the type combination and connection attributes of the connected nodes. If the two nodes connected by the path are both normal nodes and the connection strength is higher than the system setting benchmark, it is marked as a normal path and set as the default solid line; if the connection involves a deviation node but the connection strength does not meet the weak connection standard, it is set as a bold solid line; if the path involves a weak connection condition (such as the edge weight is lower than the threshold), regardless of the status of the start and end nodes, it is set as a dotted line. Combined with the sample data, such as N001 to N002 connecting normal and deviation nodes, it is set as a solid line; N002 to N003 connecting deviation and broken link nodes, it is set as a bold solid line; N003 to N004 connecting broken link and normal nodes, and the path weight is relatively low, it is set as a dotted path. The complete path style is summarized as follows:

[0126] Table 7 Path style mapping table

[0127] Path start Path End Starting point type Endpoint type Path Style N001 N002 normal deviation Solid line N002 N003 deviation Broken Chain bold solid line N003 N004 Broken Chain normal dotted line N004 N005 normal deviation Solid line N005 N001 deviation normal dotted line

[0128] As shown in Table 7, different paths have completed style classification based on the combination of connection nodes and edge characteristics, forming the basic structure of path primitives, and finally obtaining the connection path style dataset.

[0129] The graphics drawing output submodule loads the node style labeling dataset and the connection path style dataset into the preset structure of the visualization platform, builds a graphics coordinate mapping according to the node position information, generates node primitives in sequence and assigns corresponding styles, draws path connections, and outputs a data governance visualization overview map;

[0130] Based on the node style labeling dataset and the connection path style dataset, the node number, graphic identifier, and path style information are imported into the visualization platform drawing engine. First, node graphics are sequentially established in the graphic container according to the node number and type. The nodes are uniformly distributed using fixed-pitch logical coordinates. Each node position is identified with a corresponding color and graphic. After the node style is loaded, the path data records are traversed, the start and end node numbers are obtained one by one, and the corresponding path style attributes are referenced. The path segment is drawn and the two previous nodes are connected. The path color, line type, and line width are set. The uniqueness of the node number and the order of the path direction are guaranteed throughout the process to avoid cross-connections or omissions. Combining the above node and path configuration information, the complete drawing structure is output as follows:

[0131] Table 8 Visualization drawing parameters

[0132] Node number Node Type Starting point path End Path Path Style N001 normal N005 N002 dotted line N002 deviation N001 N003 Solid line N003 Broken Chain N002 N004 bold solid line N004 normal N003 N005 dotted line N005 deviation N004 N001 Solid line

[0133] As shown in Table 8, the style data of the nodes and paths in the diagram have been fully configured. After loading the data, the platform automatically completes the construction and rendering of the flowchart, and finally establishes a visual overview of data governance.

[0134] The above are merely preferred embodiments of the present invention and do not limit the present invention in any other form. Any technician familiar with the profession may use the technical content disclosed above to change or modify it into an equivalent embodiment with equivalent changes and apply it to other fields. However, any simple modification, equivalent change and modification made to the above embodiment based on the technical essence of the present invention without departing from the content of the technical solution of the present invention shall still fall within the scope of protection of the technical solution of the present invention.

Claims

1. The data comprehensive management visualization application system based on the industrial Internet is characterized by: The system comprises: The node process extraction module obtains the production process node records, sets the node sequence progression direction and calculates the time difference, determines the progression trend, sequentially detects the continuity of the node sequence time interval, and generates a process sequence association set; The data link organization module establishes inter-node dependency based on the process sequence association set, compares the consistency of task dependency with the node's preceding node ID node by node, constructs an inter-node connection map and extracts the connection weight of dependency consistency to generate a node connection structure diagram; The connectivity relationship analysis module arranges the node relationship pairs in descending order according to the connection weight according to the node connection structure diagram, selects weakly connected nodes and records the weakly connected node numbers, selects insufficiently connected areas according to the completeness of the dependent path between nodes, and generates a list of weakly connected sections; The abnormal node governance module extracts the timestamp offset and dependency missing frequency of all nodes in the weak segment according to the list of connected weak segments, classifies and sorts the abnormalities, establishes deviation node groups and broken link node groups, and forms an abnormal node classification matrix.

2. The data comprehensive management visualization application system based on the industrial Internet according to claim 1 is characterized in that: The process sequence association set includes the node time difference judgment result, the node progressive trend identification, the time interval continuity verification identification, and the reverse node number list; the node connection structure diagram includes the task dependency relationship mapping, the predecessor node matching relationship, the connection weight value list, and the node connection relationship network; the weak connectivity section list includes the weak connection node number set, the weak connection path identification set, the dependency path integrity index, and the connectivity deficiency area number; the abnormal node classification matrix includes the deviation node number set, the broken link node number set, the timestamp offset statistics table, and the dependency missing frequency statistics table.

3. The data comprehensive management visualization application system based on the industrial Internet according to claim 1 is characterized in that: The node process extraction module includes: The timestamp difference calculation submodule extracts the first node timestamp and the second node timestamp based on the production process node records, calculates the corresponding time difference, performs a sign determination operation on the time difference, marks the time difference as a positive or negative trend, and generates a node progressive trend determination result; The node sequence continuity determination submodule identifies the node number sequence according to the node progressive trend determination result, calculates the time interval values ​​in the node sequence in sequence based on the time difference between each pair of adjacent nodes, filters out the node sequences that do not meet the progressive direction continuity requirement, and generates the node time interval deviation; The reverse node identification submodule screens and records all deviated node numbers based on the node time interval deviation, determines whether the corresponding time difference is less than 0, identifies the node numbers with reversed time direction, and obtains the process sequence association set.

4. The data comprehensive management visualization application system based on the industrial Internet according to claim 1 is characterized in that: The data link organization module includes: The dependency extraction submodule decomposes the structure of the predecessor relationship marked in each task record based on the process sequence association set, aggregates all predecessor node IDs into the dependency index field of the corresponding node, summarizes the one-to-many dependency chain between nodes based on the index field, clusters and merges multiple tasks that repeatedly point to the same predecessor ID, and obtains node dependency index records; The connection mapping submodule compares the dependencies of each node with the preceding node IDs of all nodes based on the node dependency index records, performs a Boolean matching operation based on whether they are completely consistent, calculates the connection count values ​​between the nodes, and sorts them in descending order based on the matching counts to obtain the matching strength data between the nodes; The weight extraction and generation submodule selects node pairs that are greater than the matching strength extraction benchmark value to establish connections based on the node matching strength data and the connection count value between each node pair. It then determines the node connection direction based on the connection direction information and assigns a weight identifier to the direction relationship in the connection channel to obtain a node connection structure diagram.

5. The data comprehensive management visualization application system based on the industrial Internet according to claim 1 is characterized in that: The connectivity relationship analysis module includes The weight sorting and screening submodule extracts the connection weights of all connection edges based on the node connection structure graph, sorts all weight values ​​in the set in descending order, sets a benchmark connection strength, and screens out node pairs with weights less than the benchmark connection strength to obtain a low-weight connection pair set; The weak connection identification submodule performs a numbering operation on all node pairs whose edge weights are lower than the benchmark connection strength according to the low-weight connection pair set, counts the total number of connections that appear on each node in the set as the connection frequency, calculates the weak connection count value of node i, and filters to obtain a set of weak connection node numbers; The path integrity determination submodule takes each node in the weakly connected node number set as the starting point, counts the actual number of paths, constructs a theoretical maximum connection path based on all dependent target nodes, marks structural weak areas, and generates a list of connected weak sections.

6. The data comprehensive management visualization application system based on the industrial Internet according to claim 1 is characterized in that: The abnormal node management module includes: The time offset screening submodule compares the time records of each node with those of the previous node based on the list of weakly connected sections, calculates the corresponding time interval and determines whether there is an overrun situation, marks the time offset abnormal nodes and records their numbers, and obtains a set of time offset abnormal node numbers; The dependency missing statistics submodule checks the difference between the number of pre-dependencies of the current node and the actual number of existing connections based on the time offset abnormal node number set, synchronously filters and marks the dependency missing abnormal nodes, and performs a joint analysis with the time offset abnormal results to obtain an abnormal node classification index group; The abnormal matrix construction submodule labels the abnormal nodes with classification labels according to the abnormal node classification index group, determines whether the node corresponding label is in a valid state, combines all nodes with the corresponding abnormal labels, and establishes an abnormal node classification matrix.

7. The data comprehensive management visualization application system based on the industrial Internet according to claim 1 is characterized in that: The system further comprises: The comprehensive governance display module draws a node flow chart on the visualization platform based on the abnormal node classification matrix and the node connection structure diagram. Normal nodes and normal paths are marked with solid lines, deviation nodes are marked in yellow, broken nodes are marked in red, and weak connection paths are displayed with dotted lines. This module comprehensively reflects the node sequence and connection integrity status and generates a data governance visualization overview diagram. The data governance visualization overview diagram includes normal node identification, deviation node identification, broken link node identification, and weak connection path display.

8. The data comprehensive management visualization application system based on the industrial Internet according to claim 7 is characterized in that: The comprehensive management display module includes: The node style setting submodule classifies and sets node styles according to the abnormal node classification matrix and abnormal attributes, identifies normal node sets, deviation node sets, and broken node sets, assigns different graphic identification styles to the three types of nodes, and obtains a node style labeling dataset; The path style mapping submodule determines whether there is a weak connection edge attribute between two associated nodes for each connection path in the structure diagram based on the node style labeling dataset and the node connection structure diagram, marks the weak connection path accordingly, integrates the path and style relationship, and obtains the connection path style dataset; The graphics drawing output submodule loads the node style labeling dataset and the connection path style dataset into the preset structure of the visualization platform, constructs a graphics coordinate mapping according to the node position information, generates node graphics elements in sequence and assigns corresponding styles, draws path connections, and outputs a data governance visualization overview diagram.

Images