Production data management system of industrial internet

Abstract

The invention discloses a production data management system of an industrial internet, relates to the technical field of industrial internet, and solves the technical problem that the decision accuracy of the industrial internet is affected due to the fact that a large amount of data with potential value is ignored in the prior art. The device comprises a central processing module and a data transfer module connected with the central processing module, the working data of each component module in the target device is analyzed based on the edge computing technology, and once abnormal working data is recognized, the working data is not sent independently, so that the working efficiency is improved. The working data of other component modules related to the abnormal working data are identified according to the incidence relation, the working data are integrated into a first data set, and the first data set is sent to the central processing module for further identification; according to the method, the abnormity is identified from the whole, the identification accuracy is improved, the abnormal data set is further processed, valuable data can be mined from a large amount of data, and the data processing difficulty is reduced.

Description

Technical Field

[0001] This application belongs to the field of industrial Internet and involves data transmission optimization technology, specifically a production data management system for the industrial Internet. Background Art

[0002] The Industrial Internet is a new mode of production. Through the comprehensive interconnection of people, machines and objects, various data and information of industries and enterprises can be processed and analyzed through the platform. It can simplify the management process of production, operation, procurement, sales and even R&D in factories and improve management efficiency. It is also the direction of future industrial development.

[0003] The foundation of the Industrial Internet is various types of target devices. In order to effectively manage these target devices, real-time monitoring is required, which will collect a large amount of data. Most of the time, the target devices are in a normal state, which means that most of the data is not valuable for analysis. The more target devices are connected, the more data that is not valuable for analysis will be collected, which will increase the operating cost of the entire system and reduce the operating efficiency.

[0004] In view of the fact that there is a large amount of worthless data in the industrial Internet, the existing solution proposes to filter the working data of each component module of the target device, and display the filtered working data in a highly concentrated form, so as to reduce resource waste and thus reduce operating costs. This existing solution filters a large amount of data when filtering the working data. Although it can display the working status of the entire industrial zone in a highly concentrated form, it will cause a lot of data with potential value to be ignored, which will affect the decision-making accuracy of the industrial Internet.

[0005] This application provides a production data management system based on the industrial Internet to solve the above-mentioned technical problems. Summary of the invention

[0006] The present application aims to solve at least one of the technical problems existing in the prior art; to this end, the present application proposes a production data management system for the industrial Internet, which is used to solve the technical problem that the prior art ignores a large amount of data with potential value, affecting the accuracy of industrial Internet decision-making.

[0007] To achieve the above-mentioned purpose, the first aspect of the present application provides a production data management system for the industrial Internet, including a central processing module and a data transfer module connected thereto;

[0008] Data transfer module: used to collect working data of each component module of the target device through the set data sensor; preliminarily screen the working data into a first data group and a second data group; wherein the first data group is an abnormal data combination, and the second data group is a normal data combination; and,

[0009] Used to send the first data group and the second data group to a central processing module through a set message queue component; wherein the sending priority of the first data group is higher than that of the second data group;

[0010] A central processing module: used to parse the first data group, extract related data from the received second data group according to the parsing result; integrate the first data group and the corresponding related data into target data, and make decisions by analyzing the target data; and

[0011] Used to store the target data and the second data group except the associated data.

[0012] Preferably, the working data is initially screened into a first data group and a second data group, including:

[0013] Training an artificial intelligence model based on the historical data of each of the component modules to obtain an abnormality recognition model; identifying whether the working data is abnormal through the abnormality recognition model; if so, marking the component module corresponding to the working data as a target component; wherein the artificial intelligence model is constructed based on a long short-term memory network model;

[0014] The working data of the target component and the component modules associated with the abnormal working data are integrated into a first data group, and the working data of other component modules in the target device are integrated into a second data group.

[0015] Preferably, training the artificial intelligence model based on the historical data of the component modules includes:

[0016] Extracting historical data of the component modules, and generating a training data set after preprocessing the historical data; wherein the preprocessing includes processing missing values, outliers and noise;

[0017] A long short-term memory network model is constructed, and the long short-term memory network model is trained and evaluated using the training data set, and an anomaly recognition model is obtained after the evaluation is qualified.

[0018] Preferably, the identifying whether the working data is abnormal by using an abnormality identification model includes:

[0019] The working data is pre-processed and input into an abnormality recognition model to obtain reconstructed data;

[0020] Calculate the data difference between the working data and the corresponding reconstructed data, and compare the feature difference with the set threshold range to determine whether the working data is abnormal; wherein the data difference calculation method includes mean square error or cosine similarity, and the threshold range is set according to the statistical distribution of the reconstruction error.

[0021] Preferably, the set threshold range is [μ-2.5σ, μ+2.5σ]; wherein μ is the statistical mean of the reconstruction error, and σ is the standard deviation.

[0022] Preferably, the component modules associated with abnormal working data include:

[0023] Obtaining dependency relationships between component modules in the target device, and establishing a dependency graph of the target device according to the dependency relationships; wherein the dependency relationships include physical connections, data flows, and control flows;

[0024] In the dependency graph, an abnormal propagation path analysis and evaluation is performed starting from the component modules corresponding to the abnormal working data to determine the component modules associated with the abnormal working data.

[0025] Preferably, the message queue component provided includes:

[0026] A message queue is set, and a first data group and a second data group are sent through the message queue, and the first data group is sent first; or,

[0027] A message queue 1 and a message queue 2 are set, wherein the priority of the message queue 1 is higher than that of the message queue 2; the message queue 1 is used to send a first data group, and the message queue 2 is used to send a second data group.

[0028] Preferably, extracting the associated data from the received second data group according to the parsing result includes:

[0029] Identify the association relationship between the target device and other devices; wherein the target device and other devices belong to the jurisdiction of the central processing module, and the association relationship includes a data interaction relationship or a physical connection relationship;

[0030] Based on the association relationship, working data associated with some of the working data in the first data group are extracted from the received second data group, and the extracted working data are used as associated data of the first data group.

[0031] Preferably, after identifying the association relationship between the target device and other devices, the sending order of the second data groups corresponding to the other devices is adjusted according to the association relationship to extract the associated data.

[0032] Preferably, the second data group received refers to the second data group of the target device or other device received by the central processing module.

[0033] Compared with the prior art, the beneficial effects of this application are:

[0034] This application analyzes the working data of each component module in the target device based on edge computing technology. Once abnormal working data is identified, the working data is not sent separately, but the working data of other component modules related to the abnormal working data are identified according to the association relationship, integrated into a first data group, and the first data group is sent to the central processing module for further identification; this application identifies anomalies from a holistic perspective, improves the recognition accuracy, and further processes the abnormal data groups, which helps to mine valuable data from a large amount of data and reduce the difficulty of data processing.

[0035] This application starts from the abnormal working data of a single industrial device, traces back and extracts the working data of other industrial devices in the industrial Internet to form the target data; then the target data is analyzed to identify abnormal problems and make targeted decisions; this embodiment can ensure the accuracy and rationality of the decision. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0037] Figure 1 This is a schematic diagram of the data processing flow of Example 1 of the present application;

[0038] Figure 2 This is a schematic diagram of the system principle of Example 1 of the present application. DETAILED DESCRIPTION

[0039] The technical solution of the present application will be described clearly and completely in conjunction with the embodiments below. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present application.

[0040] As more and more industrial devices are connected to the Industrial Internet, the amount of data generated is also increasing. If all this data is sent to the central processing module for processing, it will not only waste a lot of resources, but also cause the truly problematic data to be unable to be processed in time, affecting the accuracy of the central processing module's decision-making.

[0041] Some solutions place all data processing on the device side, that is, the industrial equipment data is processed on the device side through edge computing technology and then sent to the central processing module, which reads, displays and stores the data. However, if the working data of a component module in a certain industrial equipment in the production process is abnormal, it may be that there is a problem with the corresponding component module, or there may be a problem with other component modules connected to it in the industrial equipment. If there is a problem with a component module of a certain industrial equipment, it may also be that there is a problem with the component module connected to the abnormal industrial module in other industrial equipment. Therefore, it is necessary not only to solve the problem of large data volume that is difficult to process, but also to ensure that the relevant problem data can be processed in a timely manner.

[0042] Embodiment 1: This embodiment is used to solve the problem that the amount of data to be processed in the Industrial Internet is large and effective mining cannot be performed, which makes data processing difficult and affects the accuracy of decision-making.

[0043] See also Figure 1-Figure 2 , the first aspect of the present application provides a production data management system for the industrial Internet, including: a central processing module, and a data transfer module connected thereto;

[0044] Data transfer module: used to collect working data of each component module of the target device through the set data sensor; preliminarily screen the working data into a first data group and a second data group; wherein the first data group is an abnormal data combination, and the second data group is a normal data combination; and,

[0045] Used to send the first data group and the second data group to the central processing module through a set message queue component; wherein the sending priority of the first data group is higher than that of the second data group;

[0046] Central processing module: used to parse the first data group, extract related data from the received second data group according to the parsing result; integrate the first data group and the corresponding related data into target data, and make decisions by analyzing the target data; and,

[0047] Used to store the target data and a second data group excluding the associated data.

[0048] The central processing module and the data transfer module in this embodiment can be set with reference to the edge computing technology, that is, the data transfer module is an edge computing module with data processing capabilities. The data transfer module is connected to the target device and is used to collect, process and send the working data of the target device. The target device is the production equipment in the industrial Internet, and of course it can also include other devices that need to be monitored.

[0049] Industrial equipment is composed of multiple connected modules, and the modules cooperate with each other to realize the functions of the industrial equipment. Various data sensors are set up in industrial equipment to detect the operating status of the modules and collect working data. Data sensors include pressure sensors, voltage sensors, current sensors, temperature sensors, liquid level sensors, etc.

[0050] The data transfer module in this embodiment collects the working data of each component module in the corresponding target device, and then analyzes and screens the working data, taking the abnormal data and the working data related to the abnormal data as the first data group, and taking other normal data as the second data group. Analyzing whether the working data is abnormal data is actually to determine whether the working state of the component module is normal, that is, to realize the abnormal identification of the target device. When a component module is abnormal, it is determined whether the component modules associated with the component module may be abnormal, and the working data of the abnormal and possibly abnormal component modules are integrated into the first data group, and the working data of other component modules are integrated into the second data group. This scheme detects other component modules of the target device through a single abnormal data expansion, that is, to realize the abnormal detection of the target device. It should be noted that when judging whether other related component modules may be abnormal, it is necessary to extract the working data in combination with the data interaction sequence, that is, to extract the working data that matches the abnormal data in time based on the data stream.

[0051] Next, the first data group and the second data group need to be sent to the central processing module. In this embodiment, the priority of the first data group is greater than that of the second data group, that is, in the same message queue, the first data group will be sent to the central processing module first for further processing by the central processing module; and the second data group is lower in value than the first data group, so it is sufficient to ensure that it can be smoothly transmitted to the central processing module for storage. Of course, if data storage is not required, the second data group may not be sent to the central processing module. It is worth noting that since the value of the second data group is lower, format conversion can be performed for compression when it is sent for easy storage.

[0052] After receiving the first data group, the central processing module analyzes it and extracts the corresponding target device and the abnormal working data therein. At this time, it is necessary to determine whether the abnormality of the target device is related to other target devices based on the association relationship between the devices, and extract the working data of other target devices for further analysis based on the judgment result to determine the specific cause of the abnormality. Then, a decision is made based on the identified specific cause to ensure the rationality and accuracy of the decision.

[0053] This embodiment analyzes the working data of each component module in the target device based on edge computing technology. Once abnormal working data is identified, it is not sent separately, but the working data of other component modules related to the abnormal working data are identified based on the association relationship, integrated into a first data group, and the first data group is sent to the central processing module for further identification. Compared with the existing technology, the abnormality is identified from the overall perspective, which improves the recognition accuracy, and the abnormal data group is further processed, which helps to mine valuable data from a large amount of data and reduce the difficulty of data processing.

[0054] Embodiment 2: The difference from Embodiment 1 is that this embodiment provides a method for dividing working data into a first data group and a second data group, so as to improve recognition accuracy and division accuracy.

[0055] The working data is initially screened into a first data group and a second data group, including:

[0056] An artificial intelligence model is trained based on the historical data of each component module to obtain an anomaly recognition model; the anomaly recognition model is used to identify whether the working data is abnormal; if so, the component module corresponding to the working data is marked as a target component; wherein the artificial intelligence model is constructed based on a long short-term memory network model;

[0057] The working data of the target component and the component modules associated with the abnormal working data are integrated into a first data group, and the working data of other component modules in the target device are integrated into a second data group.

[0058] The AI ​​model is trained using historical data from the component modules, including:

[0059] Extract historical data of the component modules and generate training data sets after preprocessing the historical data; the preprocessing includes processing missing values, outliers and noise;

[0060] A long short-term memory network model is constructed, and the long short-term memory network model is trained and evaluated through a training data set. After the evaluation is qualified, an anomaly recognition model is obtained.

[0061] Take an example to illustrate the construction of an artificial intelligence model and the judgment process of working data:

[0062] Step 1: Collect and process data:

[0063] Collect historical data of component modules, mainly time series data;

[0064] The missing values, outliers and noise in the historical data are processed and then feature extraction is performed to obtain feature data;

[0065] The feature data is normalized to obtain the training data set so that the model can learn better.

[0066] Step 2: Define the model structure: Use frameworks such as TensorFlow or Keras to build an LSTM model. The model usually includes an input layer, one or more LSTM layers, and an output layer; select an appropriate loss function and optimizer, and compile the model.

[0067] Step 3: Model training:

[0068] The training data set is divided into a training set and a test set. The training set is used to train the model, and the test set is used to evaluate the model performance.

[0069] Use the training data set to train the LSTM model to obtain the anomaly detection model.

[0070] Step 4: Data detection:

[0071] Reconstruct the working data to be detected through the trained LSTM model to obtain reconstructed data;

[0072] Compare the reconstructed data with the original data and calculate the mean square error;

[0073] Compare the mean square error with the set threshold range to determine whether the working data is abnormal.

[0074] It should be noted that the threshold range is set according to the statistical distribution of the reconstruction error, which can be obtained during the training process. The threshold range can be set to [μ-2.5σ,μ+2.5σ]; where μ is the statistical mean of the reconstruction error and σ is the standard deviation.

[0075] The above-mentioned abnormal identification model can identify abnormal working data, and then obtain the dependency relationship between the component modules in the target device, and establish the dependency graph of the target device based on the dependency relationship; wherein the dependency relationship includes physical connection, data flow and control flow; in the dependency graph, start from the component module corresponding to the abnormal working data to analyze and evaluate the abnormal propagation path, and determine the component module associated with the abnormal working data. It can be understood that when the working data of the component module is abnormal, some component modules may have already appeared abnormal before the abnormality, but they have not been identified, so the component module associated with the abnormal working data is above the data flow.

[0076] In this example, the control flow graph and data flow graph of each component module in the target device are first analyzed to determine the dependency relationship between each component module, and then a dependency graph is constructed. The dependency graph is actually a graph model that combines control dependency and data dependency. The dependency graph can be used to identify the component modules associated with the abnormal working data, determine which working data may be abnormal, and integrate them with the abnormal working data into the first data group.

[0077] The first data group and the second data group are sent in sequence, and a message queue can be set up to send the first data group and the second data group through the message queue, and the first data group is sent first. That is, once the first data group appears (the first data group and the second data group can be distinguished by a mark), the message sending order in the message queue is adjusted, and the first data group is sent first.

[0078] In another preferred embodiment, message queue 1 and message queue 2 are set, and message queue 1 has a higher priority than message queue 2; message queue 1 is used to send the first data group, and message queue 2 is used to send the second data group. Moreover, network resources should be preferentially allocated to message queue 1 to improve sending efficiency.

[0079] The first data group has a great influence on the further analysis of the anomaly, so it can be understood as valuable data, and the second data group is mainly normal data, which can be understood as worthless data. The valuable data is sent first, and the worthless data is sent in sequence according to the situation.

[0080] This embodiment mainly identifies abnormal working data through the constructed abnormal identification model. The abnormal identification model can be sent to the data transfer module for use after the central processing module is constructed; then the working data of other component modules associated with the abnormal working data are identified according to the data flow and control flow, and sent to the central processing module for processing together, which can improve the accuracy of abnormal diagnosis of target equipment, and improve the transmission efficiency by optimizing the transmission order, thereby reducing the data transmission burden of the industrial Internet.

[0081] Embodiment 3: The difference from Embodiment 1 is that a processing flow is given after the central processing module receives the first data group, that is, expanding and extracting more data of related industrial equipment from the first data group to realize the judgment of anomalies in the industrial Internet and improve the accuracy of decision-making.

[0082] Extracting associated data from the received second data group according to the parsing result includes:

[0083] Identify the association relationship between the target device and other devices; wherein the target device and other devices belong to the jurisdiction of the central processing module, and the association relationship includes a data interaction relationship or a physical connection relationship;

[0084] Based on the association relationship, working data associated with some working data in the first data group is extracted from the received second data group, and the extracted working data is used as the associated data of the first data group.

[0085] The target device and other devices are under the jurisdiction of the central processing module, which can also be understood as the target device and other devices are in the same industrial Internet system. After the central processing module receives the first data group, if only the first data group is analyzed, the potential risks of other industrial devices cannot be identified, and the decision made is limited.

[0086] In this embodiment, after receiving the first data group, the central processing module parses the first data group to obtain the working data therein, and then determines the working data associated with the working data in the first data group according to the interaction relationship between the target device (the industrial device corresponding to the first data group) and other devices. The associated working data here mainly refers to the working data of each component module in other industrial devices, and of course, it can also include the normal working data of the target device itself.

[0087] The working data in the first data group may contain abnormal working data. After the central processing module receives the data, it extracts the working data of other industrial equipment related to the abnormal working data, and combines these data into target data for analyzing the abnormalities of the entire industrial Internet, which can improve the accuracy of abnormal analysis. The analysis of target data can be implemented based on artificial intelligence models, and of course, it can also be implemented based on existing data analysis abnormality diagnosis methods.

[0088] In some other preferred embodiments, the central processing module may also extract working data in the first data group of other industrial equipment based on the first data group.

[0089] It should be noted that when the central processing module extracts the working data of other industrial equipment based on the first data group, the working data to be extracted (especially the second data group) may not have been sent to the central processing module. Considering the timeliness of abnormality identification, the sending order of the working data to be extracted but not yet sent to the central processing module is adjusted. Specifically, the sending order of these working data is advanced so as to generate the target data as soon as possible and formulate an abnormality handling plan as soon as possible.

[0090] This embodiment starts from the abnormal working data of a single industrial device, traces back and extracts the working data of other industrial devices in the industrial Internet to form the target data; then the target data is analyzed to identify abnormal problems and make targeted decisions; this embodiment can ensure the accuracy and rationality of the decision.

[0091] The above embodiments are only used to illustrate the technical method of the present application and are not intended to limit it. Although the present application has been described in detail with reference to the preferred embodiments, a person of ordinary skill in the art should understand that the technical method of the present application may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present application.

Claims

1. The production data management system of the industrial Internet includes a central processing module and a data transfer module connected thereto, and is characterized by: Data transfer module: used to collect working data of each component module of the target device through the set data sensor; preliminarily screen the working data into a first data group and a second data group; wherein the first data group is an abnormal data combination, and the second data group is a normal data combination; and, Used to send the first data group and the second data group to a central processing module through a set message queue component; wherein the sending priority of the first data group is higher than that of the second data group; A central processing module: used to parse the first data group, extract related data from the received second data group according to the parsing result; integrate the first data group and the corresponding related data into target data, and make decisions by analyzing the target data; and Used to store the target data and the second data group except the associated data.

2. The production data management system of the industrial Internet according to claim 1 is characterized in that: Preliminarily screening the working data into a first data group and a second data group includes: Training an artificial intelligence model based on the historical data of each of the component modules to obtain an abnormality recognition model; identifying whether the working data is abnormal through the abnormality recognition model; if so, marking the component module corresponding to the working data as a target component; wherein the artificial intelligence model is constructed based on a long short-term memory network model; The working data of the target component and the component modules associated with the abnormal working data are integrated into a first data group, and the working data of other component modules in the target device are integrated into a second data group.

3. The production data management system of the industrial Internet according to claim 2 is characterized in that: Training an artificial intelligence model based on historical data of the component modules includes: Extracting historical data of the component modules, and generating a training data set after preprocessing the historical data; wherein the preprocessing includes processing missing values, outliers and noise; A long short-term memory network model is constructed, and the long short-term memory network model is trained and evaluated using the training data set, and an anomaly recognition model is obtained after the evaluation is qualified.

4. The production data management system of the industrial Internet according to claim 2 is characterized in that: The identifying whether the working data is abnormal by using the abnormality identification model includes: The working data is pre-processed and input into an abnormality recognition model to obtain reconstructed data; Calculate the data difference between the working data and the corresponding reconstructed data, and compare the feature difference with the set threshold range to determine whether the working data is abnormal; wherein the data difference calculation method includes mean square error or cosine similarity, and the threshold range is set according to the statistical distribution of the reconstruction error.

5. The production data management system of the industrial Internet according to claim 4 is characterized in that: The threshold value is set in the range of [μ-2.5σ, μ+2.5σ], where μ is the statistical mean of the reconstruction error and σ is the standard deviation.

6. The production data management system of the industrial Internet according to claim 2 is characterized in that: The component modules associated with abnormal working data include: Obtaining dependency relationships between component modules in the target device, and establishing a dependency graph of the target device according to the dependency relationships; wherein the dependency relationships include physical connections, data flows, and control flows; In the dependency graph, an abnormal propagation path analysis and evaluation is performed starting from the component modules corresponding to the abnormal working data to determine the component modules associated with the abnormal working data.

7. The production data management system of the industrial Internet according to claim 1 or 6, characterized in that: The message queue component provided includes: A message queue is set, and a first data group and a second data group are sent through the message queue, and the first data group is sent first; or, A message queue 1 and a message queue 2 are set, wherein the priority of the message queue 1 is higher than that of the message queue 2; the message queue 1 is used to send a first data group, and the message queue 2 is used to send a second data group.

8. The production data management system of the industrial Internet according to claim 1, characterized in that: The extracting associated data from the received second data group according to the parsing result includes: Identify the association relationship between the target device and other devices; wherein the target device and other devices belong to the jurisdiction of the central processing module, and the association relationship includes a data interaction relationship or a physical connection relationship; Based on the association relationship, working data associated with some of the working data in the first data group are extracted from the received second data group, and the extracted working data are used as associated data of the first data group.

9. The production data management system of the industrial Internet according to claim 8, characterized in that: After identifying the association relationship between the target device and other devices, the sending order of the second data group corresponding to the other devices is adjusted according to the association relationship to extract the associated data.

10. The production data management system of the industrial Internet according to claim 8, characterized in that: The second data group received refers to the second data group of the target device or other device received by the central processing module.

Images