Method and system for constructing dynamic data knowledge graph

Abstract

The invention relates to a construction method and system for a dynamic data knowledge graph, and the construction method comprises the steps: labeling a plurality of original data according to the business logic of a first production scene, and carrying out the classification of the original data, so as to form a first number of types of label data; calling the label data of the first quantity type by using a data calling module to form interactive flow of each type of label data; according to the current calling logic, taking one piece of label data associated with the various types of label data in the label data of the first quantity type as first core key label data, and according to the current calling frequency, confirming the current interaction flow frequency between each piece of label data and the first core key label data, generating a current data knowledge graph; and judging whether a first preset time is spaced from the last generation of the current data knowledge graph or not, and if so, returning to the last step to generate an updated current data knowledge graph. The data knowledge graph generated by the method is real-time and dynamic.

Description

Technical Field

[0001] The present invention relates to the field of data knowledge graphs, and in particular to a method and system for constructing dynamic data knowledge graphs. Background Art

[0002] In today's rapidly developing industrial landscape, industrial big data, with its unique value and potential, is gradually transforming traditional production, management, and decision-making models. However, the practical application of industrial big data still faces challenges. While companies recognize the importance of data, they may currently be stuck in data collection and storage, failing to fully utilize and analyze this data, nor establishing a comprehensive data asset management system. Summary of the Invention

[0003] In order to solve the above technical problems, the purpose of the present invention is to provide a method and system for constructing a dynamic data knowledge graph.

[0004] According to one aspect of the present invention, the present invention provides a method for constructing a dynamic data knowledge graph, the construction method comprising:

[0005] Step A, based on a first production scenario, using multiple data collection systems to collect multiple raw data in real time;

[0006] Step B: labeling and classifying the plurality of raw data according to the business logic of the first production scenario to form a first number of labeled data categories;

[0007] Step C: Based on the management requirements of the first production scenario, using a data calling module to call the first number of categories of label data to form an interactive flow of label data of each category, wherein the calling frequencies of different label data are the same or different;

[0008] Step D: Based on the current calling logic of the data calling module, a piece of label data in the first number of categories that is associated with label data of multiple categories is used as the first core key label data, and based on the current calling frequency of the data calling module, the current interaction flow frequency between each label data and the first core key label data is determined to generate a current data knowledge graph; and

[0009] Step E: Determine whether a first preset time has passed since the last generation of the current data knowledge graph. If so, return to step D to generate an updated current data knowledge graph.

[0010] As an optional technical solution, in step B, according to the business logic of the first production scenario, multiple valid data are screened and extracted from the multiple original data, and the multiple valid data are labeled and classified to form label data of the first quantity category.

[0011] As an optional technical solution, the first quantity category of label data includes basic label data, state label data, process label data and inspection label data.

[0012] As an optional technical solution, the data calling module records the corresponding calling time when calling each tag data to confirm the calling frequency of each tag data.

[0013] As an optional technical solution, the current data knowledge graph includes a first node representing the first core key label data and a second node representing multiple categories of label data associated with the first core key label data. There is a first connection line between the first node and each second node, and the thickness of each first connection line is proportional to the frequency of interaction between the corresponding label data and the first core key label data.

[0014] As an optional technical solution, the current data knowledge graph also includes a third node representing the second core key label data and a fourth node representing multiple category label data associated with the second core label data, as well as a fifth node representing the third core key label data and a sixth node representing multiple category label data associated with the third core label data; there is a second connection between the third node and each fourth node, there is a third connection between the fifth node and each sixth node, the second core key label data is associated with the first core key label data, there is a fourth connection between the third node and the first node, the third core key label data is associated with the first core key label data, there is a fifth connection between the fifth node and the first node, the third core key label data is associated with the second core key label data, and there is a sixth connection between the fifth node and the third node;

[0015] When the first node is abnormal, the abnormal ones among the second nodes connected to it are checked as the first abnormal node. At the same time, the third node and the fifth node connected to it are checked to see if there are critical abnormal ones. If there are critical abnormal ones, the critical abnormal ones are regarded as the second abnormal node. Then, the abnormal ones among the fourth nodes and / or the sixth nodes connected to the second abnormal node are checked as the third abnormal nodes. The joint defense countermeasures are confirmed for the first node, the third node and the fifth node based on the first abnormal node and the third abnormal node; if there are no critical abnormal ones in the third node and the fifth node, the joint defense countermeasures are confirmed for the first node, the third node and the fifth node based on the first abnormal node.

[0016] As an optional technical solution, according to the call record of the data call module, at the first time node, each tag data and the first core key tag data have a first interactive flow frequency sequence; at the second time node, each tag data and the first core key tag data have a second interactive flow frequency sequence, and the first interactive flow frequency sequence is different from the second interactive flow frequency sequence.

[0017] According to one aspect of the present invention, the present invention provides a system for constructing a dynamic data knowledge graph, comprising:

[0018] A data collection module, the data collection module is used to collect a plurality of raw data using a plurality of data collection systems based on a first production scenario;

[0019] a data classification module, configured to label and classify the plurality of raw data according to the business logic of the first production scenario to form labeled data of a first number of categories; and

[0020] A data calling module, the data calling module being used to call the first number of categories of label data according to the management requirements of the first production scenario to form an interactive flow of the label data of each category, wherein the calling frequencies of different label data are the same or different; and according to the current calling logic of the data calling module, a piece of label data in the first number of categories that is associated with label data of multiple categories is used as the first core key label data, and according to the current calling frequency of the data calling module, the current interactive flow frequency between each label data and the first core key label data is confirmed, and the corresponding current data knowledge graph is generated;

[0021] Among them, every first preset time interval, the first core key tag data and the interactive flow between each tag data and the first key tag data and the current interactive flow frequency are reconfirmed to generate an updated current data knowledge graph.

[0022] As an optional technical solution, the data calling module records the corresponding calling time when calling each tag data to confirm the calling frequency of each tag data.

[0023] As an optional technical solution, according to the call record of the data call module, at the first time node, each label data and the first core key label data have a first interactive flow frequency sequence; at the second time node, each label data and the first core key label data have a second interactive flow frequency sequence, the first interactive flow frequency sequence is different from the second interactive flow frequency sequence, and the updated current data knowledge graph is different from the current data knowledge graph before the update.

[0024] In summary, the construction method and construction system of the dynamic data knowledge graph provided by the present invention uses the data calling module at preset time intervals according to the current calling logic and the current calling frequency to take the label data associated with multiple categories of label data in multiple categories as the first core key label data, and confirm the current interactive flow frequency between each label data and the first core key label data to generate the current data knowledge graph. In this way, the data calling module will generate a data knowledge graph for the corresponding time node at preset time intervals. That is to say, the data knowledge graph generated by the data knowledge graph construction method and system provided by the present invention is real-time and dynamic, and can reflect the flow interactive correlation between each data in the production scene in real time, so that users can make corresponding decisions based on the dynamic data knowledge graph.

[0025] The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments, but this does not limit the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 This is a flowchart of a method for constructing a dynamic data knowledge graph according to an embodiment of the present invention.

[0027] Figure 2 This is a schematic diagram of a data knowledge graph according to an embodiment of the present invention.

[0028] Figure 3 A block diagram of a dynamic data knowledge graph construction system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0029] To facilitate understanding of the technical features, contents, advantages, and achievable effects of the present invention, the present invention will be described in detail below using embodiments in conjunction with the accompanying drawings. The drawings used therein are intended only for illustration and to assist in the description, and may not reflect the actual proportions and precise configurations of the present invention after implementation. Therefore, the proportions and configurations of the attached drawings should not be interpreted to limit the scope of the present invention in actual implementation. This is to be noted in advance.

[0030] See Figure 1-Figure 3 , Figure 1 This is a flowchart of a method for constructing a dynamic data knowledge graph according to an embodiment of the present invention. Figure 2 This is a schematic diagram of a data knowledge graph according to an embodiment of the present invention. Figure 3 A block diagram of a dynamic data knowledge graph construction system according to an embodiment of the present invention.

[0031] The present invention provides a method for constructing a dynamic data knowledge graph, which is applied to industrial production. The method comprises:

[0032] Step A, based on the first production scenario, uses multiple data collection systems to collect multiple raw data in real time; in one embodiment, the data collection system may include but is not limited to: IoT sensor networks, MES systems (Manufacturing Execution System, Manufacturing Execution System), EPMS systems (Equipment Performance Management System, Equipment Performance Management System) and other data collection terminals, real-time collection of factory operation data and equipment working data. In addition, the IoT sensor network includes but is not limited to temperature sensors, humidity sensors, pressure sensors, flow sensors, vibration sensors and other types of sensors, which are distributed in key areas such as production lines, warehouses, and logistics systems. Each sensor collects data at a predetermined time interval and transmits the data to the data collection system through a low-power wide area network (such as LoRaWAN).

[0033] Step B: Label and classify multiple raw data according to the business logic of the first production scenario to form a first number of types; in one embodiment, the first production scenario is the manufacture of display panels. According to the business logic of manufacturing display panels, the first number of types is, for example, 20 or 30 types, but is not limited to this. In different production scenarios, due to different business logic, the specific value of the first number may be different.

[0034] Step C, according to the management requirements of the first production scenario, use the data calling module 103 to call the label data of the first number category to form an interactive flow of label data of each category, wherein the calling frequencies of different label data are the same or different; in one embodiment, the data calling module 103 has the same calling frequency for multiple label data belonging to the first largest category, the same calling frequency for multiple label data belonging to the second largest category, and the same calling frequency for multiple label data belonging to the third largest category, but the data calling module 103 has a different calling frequency for multiple label data belonging to the first largest category than for multiple label data belonging to the second largest category, the data calling module 103 has a different calling frequency for multiple label data belonging to the first largest category than for multiple label data belonging to the third largest category, and the data calling module 103 has a different calling frequency for multiple label data belonging to the second largest category than for multiple label data belonging to the third largest category, but is not limited to this.

[0035] Step D: Based on the current call logic of the data call module 103, a piece of label data in the first number of categories that is associated with label data of multiple categories is used as the first core key label data. Based on the current call frequency of the data call module 103, the current interaction flow frequency between each piece of label data and the first core key label data is determined to generate a current data knowledge graph.

[0036] Step E: Determine whether a first preset time has passed since the last generation of the current data knowledge graph. If so, return to step D to generate an updated current data knowledge graph. If not, continue timing monitoring and executing the judgment action.

[0037] The method for constructing a dynamic data knowledge graph of the present invention utilizes the data calling module 103 to call the data collected by each data collection system, and confirms the first core key label data and the current interactive flow frequency between each label data and the first core key label data according to the current calling logic to generate the current data knowledge graph. Furthermore, at every first preset time, the new first core key label data and the current interactive flow frequency between each label data and the new first core key label data will be re-confirmed according to the current calling logic to generate an updated current data knowledge graph. In other words, the present invention generates a data knowledge graph based on the interactive flow generated by the call of each label data by the data calling module 103, and continuously updates the knowledge graph based on the call of each label data by the data calling module 103 after each preset time interval to form a dynamic data knowledge graph to reflect the flow interactive correlation between each data in the production scenario in real time, so that users can make corresponding decisions based on the dynamic data knowledge graph.

[0038] In one embodiment, in step B, multiple original data are labeled and classified separately to form labeled data of the first quantity category, instead of directly labeling and classifying all original data, because the original data collected by each data collection system is relatively large, and each original data is considered to need to be collected based on aspects such as data comprehensiveness during the system design phase, but based on the actual business logic of the first production scenario, it can be determined that the frequency of calling some original data will be low (or even not called), and the management effect on the entire data of the specific production scenario of the first production scenario is relatively unobvious. Therefore, before labeling and classifying multiple original data, multiple valid data are first screened and extracted from the multiple original data according to the business logic of the first production scenario, and then the extracted multiple valid data are labeled and classified separately to form labeled data of the first quantity category. The valid data mentioned here refers to the original data with a data call frequency greater than a preset threshold in the first production scenario. For example, in each production cycle (for example, one day, but not limited to this), the original data with a data call frequency greater than 30 times is extracted to form valid data, and the original data with a data call frequency less than or equal to 30 times is regarded as invalid data, and these invalid data will not be labeled and classified (that is, screened out).

[0039] In one embodiment, based on the business logic of the first production scenario, the first quantity category of label data may include four major categories of label data: basic label data, status label data, process label data, and inspection label data. The basic label data can be further divided into multiple subcategories of label data, including, but not limited to, equipment number, equipment type, barcode number, program name, substrate thickness, substrate length, and substrate width. The status label data can be further divided into multiple subcategories of label data, including, but not limited to, alarm codes, heartbeat packets, electrical control signals, temperature alarm flags, and humidity alarm flags. The process label data can be further divided into multiple subcategories of label data, including, but not limited to, blade hardness, demolding speed, printing gap, dripping speed, printing downtime, cleaning starting point, cleaning length, and paper transfer speed. The inspection label data can be further divided into multiple subcategories of label data, including, but not limited to, actual X offset value, actual Y offset value, upper X offset limit, lower X offset limit, upper Y offset limit, and lower Y offset limit.

[0040] In one embodiment, in step D, combined with the label data classification described above, for example, based on the current call logic of the data call module 103, the first core key label data is determined to be the device code, and other associated label data such as scraper hardness, temperature alarm flag, electrical control signal, actual X offset value, and printing gap can be determined. In this way, based on the current call frequency of the data call module 103, the current interaction flow frequency between the label data such as scraper hardness, temperature alarm flag, electrical control signal, actual X offset value, and printing gap and the device code can be determined, and then a data knowledge graph with the device code as one of the cores is generated. In other words, the generated data knowledge graph can have at least two interrelated layers and can include at least one core data, and the aforementioned device code can be one of the core data.

[0041] Furthermore, the calling logic and calling frequency of each tag data adopted by the data calling module 103 are not fixed. Thus, the interaction flow frequency between each tag data and the first core key tag data also changes dynamically. For example, at a first time node, the data calling module 103 adopts the first calling logic and the first calling frequency to call each tag data, and forms a first data knowledge graph based on the first interaction flow frequency between each tag data and the first core key tag data. At a second time node, the second time node may differ from the first time node by N (N is a positive integer greater than 1) times the first preset time. For example, although both are display panels, the specific model of the display panels manufactured has changed. The data calling module 103 instead adopts the second calling logic and the second calling frequency to call each tag data, and forms a second data knowledge graph based on the second interaction flow frequency between each tag data and the first core key tag data. The second data knowledge graph is different from the first data knowledge graph, and can dynamically reflect the parameter changes of each equipment, process, factory environment, etc. in the first production scenario, so that relevant personnel can make relevant decisions based on the latest data knowledge graph.

[0042] As mentioned above, since the calling frequency of label data changes dynamically, as time goes by (for example, the actual business needs of the first production scenario change), when the calling frequency of some label data is too low at a certain moment, it is possible that some label data in the data knowledge graph generated at the first time node (for example, defined as the first data knowledge graph) disappears in the data knowledge graph generated at the second time node (for example, defined as the second data knowledge graph). In this way, unnecessary data that has little effect on current production management can be removed in a timely manner.

[0043] In one embodiment, the calling frequency of each label data can be confirmed by recording the calling time when the data calling module 103 calls each label data, that is, the data calling module 103 will record the corresponding calling time when calling each label data, so as to calculate the number of times the data calling module 103 calls each label data within a preset time period, and then confirm the calling frequency of each label data. Similarly, since the calling frequency of each label data changes dynamically, as time goes by, when the calling frequency of a new label data increases to more than a preset number of times (for example, 30), then correspondingly, new label data associated with the first core key label data will be added to the new data knowledge graph. In this way, the establishment and update of the data knowledge graph of the present invention is not based on the classification and association rules set in advance, but is generated accordingly based on the real-time interactive flow between each label data.

[0044] Furthermore, the interaction flow frequency between each tag data and the first core key tag data is dynamically changing. The dynamic change not only refers to the high and low changes of each interaction flow frequency itself as described above, but also includes the change in the order between each interaction flow frequency. For example, according to the call record of the data call module 103, at the first time node, each tag data has a first interaction flow frequency order with the first core key tag data; at the second time node, each tag data has a second interaction flow frequency order with the first core key tag data, and the first interaction flow frequency order is different from the second interaction flow frequency order. Taking the first core key tag data as the device code as an example, the first three digits in the first interaction flow frequency order are, for example, scraper hardness, temperature alarm mark, and electric control signal, and the first three digits in the second interaction flow frequency order are, for example, scraper hardness, temperature alarm mark, and actual value of X offset. This makes it easier for relevant personnel to conduct analysis and make targeted decisions based on the latest data knowledge graph.

[0045] It should be noted that the data calling module 103 can be applied to the construction system 1000 of the dynamic data knowledge graph (such as Figure 3 As shown in the figure, which will be described in detail later), the construction system 1000 also includes a data dashboard 104 that is communicatively connected to the data calling module 103. In this embodiment, in step C, while the data calling module 103 is used to call the label data of the first number category to form an interactive flow of label data of each category, the data dashboard 104 can also generate corresponding visualization charts based on the situation of the data calling module 103 calling the label data of the first number category. The visualization chart formed in step C and the data knowledge graph generated in step D can both be presented by the data dashboard 104. In one embodiment, the visualization chart generated in step C can be regarded as a result presentation product. The visualization chart generated in step C mainly reflects some trends of a certain type of label data itself (such as the running curve of a certain device code on the time axis), and cannot be directly associated with other label data. The correlation with other label data needs to be realized through the data knowledge graph generated in step D. For example, when the generated visualization chart shows that the operating curve of a certain device code is abnormal, the corresponding data knowledge graph can be queried to confirm the label data associated with this device code and further make improvement decisions.

[0046] In one embodiment, please refer to Figure 2, the current data knowledge graph includes a first node 201 representing the first core key label data and a second node 202 representing multiple categories of label data associated with the first core key label data. There is a first connection line 301 between the first node 201 and each second node 202. The thickness of each first connection line 301 is proportional to the frequency of interaction between the corresponding label data and the first core key label data. In other words, the thicker a certain first connection line 301 is, the higher the frequency of interaction between the label data connected to the certain first connection line 301 and the first core key label data is, and the thinner a certain first connection line 301 is, the lower the frequency of interaction between the label data connected to the certain first connection line 301 and the first core key label data is. Taking the first core key label data as a device code as an example, if the operation curve corresponding to this device code is abnormal, it can be considered to check each label data in the order of interaction flow frequency from high to low.

[0047] Also, please continue to refer to Figure 2 , the current data knowledge graph includes not only the first node 201 representing the first core key label data, but also the third node 203 representing the second core key label data and the fourth node 204 representing multiple categories of label data associated with the second core label data, as well as the fifth node 205 representing the third core key label data and the sixth node 206 representing multiple categories of label data associated with the third core label data; there is a second connection 302 between the third node 203 and each fourth node 204, there is a third connection 303 between the fifth node 205 and each sixth node 206, the second core key label data is associated with the first core key label data, there is a fourth connection 304 between the third node 203 and the first node 201, the third core key label data is associated with the first core key label data, the fifth node 205 and the first node 201 There is a fifth connection 305 between them, the third core key label data is associated with the second core key label data, and there is a sixth connection 306 between the fifth node 205 and the third node 203; when the first node 201 is abnormal, the abnormal ones in the second nodes 202 connected to it are checked as the first abnormal node, and at the same time, the third nodes 203 and the fifth nodes 205 connected to it are checked to see if there are critical abnormal ones (critical abnormal ones can be regarded as not reaching the level of abnormality, but have reached a level with higher risks). If there are critical abnormal ones, the critical abnormal ones are regarded as the second abnormal node, and then the abnormal ones in the fourth nodes 204 or / and the sixth nodes 206 connected to the second abnormal node are checked as the third abnormal node. According to the first abnormal node and the third abnormal node, the joint defense measures are confirmed for the first node 201, the third node 203 and the fifth node 205.

[0048] For example, the first node 201, the third node 203 and the fifth node 205 represent the first device, the second device and the third device respectively. The first device, the second device and the third device are three devices of the same type. When an abnormality is found in the first node 201 on the data dashboard 104 (for example, the red light is on the first device), the current data knowledge graph can be used to trace back and check which label data represented by the associated second nodes 202 has problems and what causes the problems. At the same time, the current data knowledge graph can be used to trace back to see if the third node 203 (representing the second device) and the fifth node 205 (representing the fifth device) associated with it are critical anomalies. If the third node 203 and / or the fifth node 205 are critical anomalies, the critical anomaly is used as the second abnormal node. For example, if the third node 203 is a critical anomaly (i.e., the third node 203 is used as the second abnormal node), and the fifth node 205 is not a critical anomaly (i.e., the third device is currently in a normal state), then it can be confirmed that the second device has not yet reached the abnormal level (the red light has not yet turned on). However, there is a high risk (the yellow light has already turned on). Further, the abnormalities in each fourth node 204 connected to the second abnormal node (third node 203) are further checked as third abnormal nodes. The third abnormal node can determine which tag data associated with the second device has problems and what caused the problems. Then, based on the causes of the problems traced back by the first and third abnormal nodes, corresponding solutions are made to confirm joint defense measures for the first node (representing the first device), the third node (representing the second device), and the fifth node (representing the third device). In this way, the abnormal first device can be processed after the abnormality occurs, and the high-risk second device and the normal third device can be prevented beforehand. The management of the equipment can be moved from ex post management to ex ante management (i.e., ex ante prevention).

[0049] It should be noted that the above example only illustrates the case where the third node 203 (i.e., the second device) is a critical anomaly while the fifth node 205 (i.e., the third device) is in a normal state. In other embodiments, the third node 203 (i.e., the second device) may be in a normal state while the fifth node 205 (i.e., the third device) is a critical anomaly. In yet another embodiment, both the third node 203 (i.e., the second device) and the fifth node 205 (i.e., the third device) may be critical anomalies. The processing methods for these two cases are similar to those described above and will not be repeated here. Of course, in yet another embodiment, both the third node 203 (i.e., the second device) and the fifth node 205 (i.e., the third device) may be in a normal state, that is, there is no critical anomaly. In this case, it is only necessary to check for anomalies among the second nodes 202 connected to the first node 201, and treat the abnormal second node 202 as the first abnormal node. Then, based on the first abnormal node, a joint defense strategy is determined for the first node 201, the third node 203, and the fifth node 205.

[0050] like Figure 2 As shown, the first node 201, the third node 203, and the fifth node 205 are connected. In another embodiment, not limited to this, the first node 201, the third node 203, and the fifth node 205 are not connected to each other, but are all connected to a key data tag node (for example, defined as the seventh node). When the running curve corresponding to the seventh node shown in the data dashboard 104 is abnormal, the relevant cause can be checked through the data knowledge graph, and the joint defense countermeasures can be confirmed based on the comprehensive investigation of the first node 201 (and its associated second nodes 202), the third node 203 (and its associated fourth nodes 204), and the fifth node 205 (and its associated sixth nodes 206).

[0051] Furthermore, as mentioned above, “when the first node is abnormal, the abnormal ones of the second nodes connected to it are checked as the first abnormal nodes”, during this checking process, the checking order can be confirmed according to the thickness of the first connection line between the first node and the abnormal second nodes. Because the thicker the first connection line, the higher the frequency of communication and interaction between the corresponding second node and the first node, the more important the label data corresponding to the second node is. When an abnormality occurs in this second node, it can be checked from this second node first, thereby shortening the time to find the cause of the problem. In addition, the present invention, with the help of real-time dynamic data knowledge graph, no longer checks for abnormalities at a single point (such as a second node) for single-point processing, but makes a comprehensive decision based on the correlation between the nodes in the data knowledge graph, so as to solve the current abnormality while taking comprehensive preventive measures.

[0052] Accordingly, the present invention also provides a system 1000 for constructing a dynamic data knowledge graph, comprising: a data acquisition module 101, a data classification module 102, and a data calling module 103 connected in communication, wherein the data acquisition module 101 is used to collect a plurality of raw data using a plurality of data collection systems based on a first production scenario; the data classification module 101 is used to label and classify a plurality of raw data according to the business logic of the first production scenario to form a first number of categories of labeled data; the data calling module 103 is used to call the first number of categories of labeled data according to the management requirements of the first production scenario to form a first number of categories of labeled data. Interactive flow, wherein the calling frequencies of different label data are the same or different; and according to the current calling logic of the data calling module 103, a label data in the first number of categories that is associated with label data of multiple categories is used as the first core key label data, and according to the current calling frequency of the data calling module 103, the current interactive flow frequency between each label data and the first core key label data is confirmed to generate the corresponding current data knowledge graph; wherein, at every first preset time interval, the interactive flow and the current interactive flow frequency between each label data and the first core key label data are reconfirmed to generate an updated current data knowledge graph. Furthermore, the construction system 1000 of the dynamic data knowledge graph may also include a data dashboard 104, which is in communication with the data calling module 103, and the data knowledge graph generated above can be presented by the data dashboard 104. It should be noted that the above description of the construction method of the dynamic data knowledge graph is also applicable to the construction system 1000 of the corresponding dynamic data knowledge graph and will not be repeated.

[0053] In summary, the construction method and construction system of the dynamic data knowledge graph provided by the present invention uses the data calling module at preset time intervals according to the current calling logic and the current calling frequency to take the label data associated with multiple categories of label data in multiple categories as the first core key label data, and confirm the current interactive flow frequency between each label data and the first core key label data to generate the current data knowledge graph. In this way, the data calling module will generate a data knowledge graph for the corresponding time node at preset time intervals. That is to say, the data knowledge graph generated by the data knowledge graph construction method and system provided by the present invention is real-time and dynamic, and can reflect the flow interactive correlation between each data in the production scene in real time, so that users can make corresponding decisions based on the dynamic data knowledge graph.

[0054] The embodiments described above are merely illustrative of the technical concepts and features of the present invention. Their purpose is to enable persons skilled in the art to understand the contents of the present invention and implement them accordingly. They are not intended to limit the patent scope of the present invention. In other words, any equivalent changes or modifications made in accordance with the spirit disclosed in the present invention should still be covered by the patent scope of the present invention.

Claims

1. A method for constructing a dynamic data knowledge graph, applied to industrial production, characterized in that: The construction method includes: Step A, based on a first production scenario, using multiple data collection systems to collect multiple raw data in real time; Step B: labeling and classifying the plurality of raw data according to the business logic of the first production scenario to form a first number of labeled data categories; Step C: Based on the management requirements of the first production scenario, using a data calling module to call the first number of categories of label data to form an interactive flow of label data of each category, wherein the calling frequencies of different label data are the same or different; Step D: Based on the current calling logic of the data calling module, a piece of label data in the first number of categories that is associated with label data of multiple categories is used as the first core key label data, and based on the current calling frequency of the data calling module, the current interaction flow frequency between each label data and the first core key label data is determined to generate a current data knowledge graph; and Step E: Determine whether a first preset time has passed since the last generation of the current data knowledge graph. If so, return to step D to generate an updated current data knowledge graph.

2. The method for constructing a dynamic data knowledge graph according to claim 1, characterized in that: In step B, according to the business logic of the first production scenario, a plurality of valid data are screened and extracted from the plurality of original data, and the plurality of valid data are labeled and classified respectively to form the labeled data of the first quantity category.

3. The method for constructing a dynamic data knowledge graph according to claim 2, characterized in that: The first number of categories of label data include basic label data, state label data, process label data and inspection label data.

4. The method for constructing a dynamic data knowledge graph according to claim 1, characterized in that: The data calling module records the corresponding calling time when calling each tag data to confirm the calling frequency of each tag data.

5. The method for constructing a dynamic data knowledge graph according to claim 1, characterized in that: The current data knowledge graph includes a first node representing the first core key label data and a second node representing multiple categories of label data associated with the first core key label data. There is a first connection line between the first node and each second node, and the thickness of each first connection line is proportional to the frequency of interaction between the corresponding label data and the first core key label data.

6. The method for constructing a dynamic data knowledge graph according to claim 5, characterized in that: The current data knowledge graph also includes a third node representing the second core key label data and a fourth node representing multiple category label data associated with the second core label data, as well as a fifth node representing the third core key label data and a sixth node representing multiple category label data associated with the third core label data; a second connection is provided between the third node and each fourth node, a third connection is provided between the fifth node and each sixth node, the second core key label data is associated with the first core key label data, a fourth connection is provided between the third node and the first node, the third core key label data is associated with the first core key label data, a fifth connection is provided between the fifth node and the first node, the third core key label data is associated with the second core key label data, and a sixth connection is provided between the fifth node and the third node; When the first node is abnormal, the abnormal ones among the second nodes connected to it are checked as the first abnormal nodes. At the same time, the third node and the fifth node connected to it are checked to see if there are critical abnormal ones. If there are critical abnormal ones, the critical abnormal ones are regarded as the second abnormal nodes. Then, the abnormal ones among the fourth nodes and / or the sixth nodes connected to the second abnormal node are checked as the third abnormal nodes. Based on the first abnormal nodes and the third abnormal nodes, joint defense countermeasures are confirmed for the first node, the third node and the fifth node. If neither the third node nor the fifth node has a critical abnormality, a joint defense measure is confirmed for the first node, the third node, and the fifth node according to the first abnormal node.

7. The method for constructing a dynamic data knowledge graph according to claim 1, characterized in that: According to the call records of the data call module, at the first time node, there is a first interactive flow frequency sequence between each tag data and the first core key tag data; at the second time node, there is a second interactive flow frequency sequence between each tag data and the first core key tag data, and the first interactive flow frequency sequence is different from the second interactive flow frequency sequence.

8. A system for constructing a dynamic data knowledge graph, characterized in that: include: A data collection module, the data collection module is used to collect a plurality of raw data using a plurality of data collection systems based on a first production scenario; a data classification module, configured to label and classify the plurality of raw data according to the business logic of the first production scenario to form labeled data of a first number of categories; and A data calling module, the data calling module being used to call the first number of categories of label data according to the management requirements of the first production scenario to form an interactive flow of the label data of each category, wherein the calling frequencies of different label data are the same or different; and according to the current calling logic of the data calling module, a piece of label data in the first number of categories that is associated with label data of multiple categories is used as the first core key label data, and according to the current calling frequency of the data calling module, the current interactive flow frequency between each label data and the first core key label data is confirmed, and the corresponding current data knowledge graph is generated; Among them, every first preset time interval, the first core key tag data and the interactive flow between each tag data and the first key tag data and the current interactive flow frequency are reconfirmed to generate an updated current data knowledge graph.

9. The system for constructing a dynamic data knowledge graph according to claim 8, characterized in that: The data calling module records the corresponding calling time when calling each tag data to confirm the calling frequency of each tag data.

10. The system for constructing a dynamic data knowledge graph according to claim 8, characterized in that: According to the call records of the data call module, at the first time node, each label data and the first core key label data have a first interactive flow frequency sequence; at the second time node, each label data and the first core key label data have a second interactive flow frequency sequence, the first interactive flow frequency sequence is different from the second interactive flow frequency sequence, and the updated current data knowledge graph is different from the current data knowledge graph before the update.

Images