Industrial internet of things edge gateway system and data processing method thereof

Abstract

The invention discloses an industrial Internet of Things edge gateway system and a data processing method thereof, and the system comprises a data conversion module which is used for converting source JSON data in a specific format into target JSON data in a template format through a predefined template; the data acquisition module uses a configurable data acquisition pipeline as a northbound application node, binds the northbound application node with a southbound equipment node, receives data transmitted by equipment, and dynamically adjusts a data acquisition strategy through configuration management; the northbound AI reasoning module is used for being bound with a southbound equipment node and performing edge side model reasoning according to data transmitted by equipment to obtain an anomaly detection result; and the internal API of the edge gateway system is used for writing in the internal data point of the PLC or sending an information notification to a specified cloud end. By adopting the technical scheme of the invention, the data processing capability is improved, and the intelligent level of the system is enhanced.

Description

Technical Field

[0001] The present invention belongs to the technical field of gateway systems, and in particular relates to an industrial Internet of Things edge gateway system and a data processing method thereof. Background Art

[0002] The edge gateway system is an industrial protocol gateway software designed specifically for the Industrial Internet of Things (IIoT) scenario, which aims to effectively connect and manage industrial equipment, sensors and upper-layer applications. With the development of Industry 4.0 and smart manufacturing, the application of edge computing has gradually become an important means to improve the level and efficiency of factory automation. The edge gateway system collects and processes data on the edge side, and with its ultra-low latency characteristics, it makes data processing more timely, thereby improving the response speed and reliability of the overall industrial automation system.

[0003] The edge gateway system has strong protocol compatibility and can operate efficiently in various complex industrial environments. It supports real-time data collection, data processing, and data forwarding on the edge side, thereby reducing dependence on central cloud computing resources and reducing the pressure on network bandwidth usage. In addition, the edge gateway also has distributed computing capabilities, allowing enterprises to perform data cleaning, preprocessing, and preliminary analysis on the edge side, thereby optimizing the workload of the production control system.

[0004] The current mainstream edge gateway systems, such as NEURON and Rulex, are usually designed with a north-south architecture, which makes data flow and system functions clearly divided, making management and expansion easier.

[0005] ● Southbound node: The southbound node is the interface for device connection and is responsible for obtaining data from the underlying devices in the production process. Southbound nodes usually support a wide range of industrial protocols and can access a variety of industrial sensors, industrial control equipment (such as PLC), and intelligent instruments. These device connection protocols may include Modbus, OPC UA, BACnet, Profibus, etc. to ensure that devices of different types and manufacturers can be smoothly connected to the edge gateway. The goal of the southbound node is to achieve all-round data collection of the production site to ensure the integrity and timeliness of production information.

[0006] ●Northbound node: The northbound node is an interface for upper-layer applications, designed to transmit the collected data to the cloud platform or other upper-layer information systems. The northbound node not only supports data upload, but also supports personalized applications based on specific business scenarios, ensuring that data can be provided to MES systems, ERP systems, data analysis platforms, etc. through REST API, MQTT, HTTP and other methods. The northbound node also supports docking with big data analysis and artificial intelligence models to further explore the value of data and help enterprises with predictive maintenance, production optimization and quality management.

[0007] In addition, edge gateway systems usually have data security mechanisms, such as encryption, access control, and identity authentication, to ensure the security of data during transmission and storage. At the same time, edge gateway systems can also store data locally to cope with temporary network interruptions, ensuring that data is not lost during the production process and that the system is stable.

[0008] In the development of the Industrial Internet of Things, the edge gateway is not just a simple data forwarding tool, but an intelligent node that integrates multiple technologies such as edge computing, protocol conversion, and security management. Its low latency, high reliability, and rich protocol support capabilities make it a vital position in industrial scenarios. By combining edge computing with cloud computing, the edge gateway system can help enterprises optimize resource allocation, improve operational efficiency, and achieve more flexible production processes and more intelligent production decisions.

[0009] The main architecture of the edge gateway system is a north-south architecture. The southbound node supports a variety of device connection protocols and is responsible for connecting various sensors, PLCs and other devices; the northbound node is responsible for processing and reporting the data of the corresponding southbound node, such as reporting to the self-built industrial IoT cloud platform, Huawei Cloud IoT cloud platform, cloud database and other cloud services. The data transmission between the northbound and southbound nodes is completed by the message bus inside the system, that is, the data collected by the southbound node will be forwarded to the northbound node after intermediate processing, and then reported to the cloud service by the northbound node. Based on the above architecture, when you want to expand the application scenario based on the capabilities of the northbound node, such as accessing a new third-party cloud platform, a data analysis API of a specific function, etc. (hereinafter collectively referred to as the upper-layer service), you will inevitably face the problem of inconsistency between the data format forwarded by the middle layer to the northbound node and the data format required by the newly connected upper-layer service. The middle layer of the edge gateway software system itself cannot be modified, and the data format required by the upper-layer service is often fixed. At this time, you need to manually write the data format conversion logic, and if you want to access a new upper-layer service, you need to write a new data format conversion logic. This process is often tedious, requires a lot of manual coding, and lacks flexibility to effectively handle complex nested data structures.

[0010] With the development of industrial Internet of Things technology, the data collection and processing of equipment has become increasingly important. In existing technologies, data collection methods often lack flexibility and configurability, and are difficult to adapt to the needs of different devices, especially in edge computing environments. In addition, existing solutions lack the means to provide dynamic collection strategy configuration. Summary of the invention

[0011] The technical problem to be solved by the present invention is to provide an industrial Internet of Things edge gateway system and a data processing method thereof to improve data processing capabilities and enhance the intelligence level of the system.

[0012] To achieve the above object, the present invention adopts the following technical solution:

[0013] An industrial Internet of Things edge gateway system, comprising:

[0014] A data conversion module is used to convert source JSON data in a specific format into target JSON data in a template format through a predefined template;

[0015] The data acquisition module uses a configurable data acquisition pipeline as a northbound application node, binds it to the southbound device node, receives data from the device, and dynamically adjusts the data acquisition strategy through configuration management;

[0016] The northbound AI inference module is used to bind to the southbound device node, perform edge-side model inference based on the data sent by the device, and obtain anomaly detection results; then the internal API of the edge gateway system is used to write to the PLC internal data points or send information notifications to the specified cloud.

[0017] Preferably, the data conversion module includes:

[0018] The acquisition unit is used to acquire two JSON strings, wherein one JSON string is a source JSON document, and the other JSON string is a template JSON document;

[0019] A recursive processing unit, used to traverse each element in the template JSON document through a recursive function;

[0020] The pointer parsing and value replacement unit is used to parse the pointer of the string node containing the JSON Pointer and extract the corresponding value from the source JSON document, and use the extracted value to replace the value of the corresponding node in the template;

[0021] The output unit is used to serialize the modified template JSON document into a string format after traversing and replacing all elements in the template.

[0022] Preferably, the data collection strategy includes: a runtime strategy, a post-processing strategy and a storage strategy.

[0023] The present invention also provides a data processing method for an industrial Internet of Things edge gateway system, comprising:

[0024] Step S1: Convert source JSON data in a specific format into target JSON data in a template format using a predefined template;

[0025] Step S2: Use the configurable data acquisition pipeline as the northbound application node, bind it to the southbound device node, receive data from the device, and dynamically adjust the data acquisition strategy through configuration management;

[0026] Step S3: Bind the northbound AI reasoning module to the southbound device node, perform edge-side model reasoning based on the data sent by the device, and obtain anomaly detection results; then use the internal API of the edge gateway system to write to the PLC internal data points or send information notifications to the specified cloud.

[0027] As a preferred step S1 comprises:

[0028] Get two JSON strings; one JSON string is the source JSON document, and the other JSON string is the template JSON document;

[0029] Traverse each element in the template JSON document through a recursive function;

[0030] For a string node containing a JSON Pointer, parse the pointer and extract the corresponding value from the source JSON document, and use the extracted value to replace the value of the corresponding node in the template;

[0031] After traversing and replacing all elements in the template, the modified template JSON document is serialized into a string format.

[0032] Preferably, the data collection strategy includes: a runtime strategy, a post-processing strategy and a storage strategy.

[0033] The present invention has the following technical effects:

[0034] 1. Flexibility of data compatibility layer: The data compatibility layer of the present invention provides the ability to easily, flexibly and quickly access various cloud platforms and upper-layer applications, so that the edge gateway can be easily integrated with heterogeneous systems to meet different business needs, greatly reducing the access and maintenance costs of the system.

[0035] 2. Intelligent data collection: Through the configurable data collection pipeline, the system can achieve flexible configuration and efficient data processing capabilities, and can realize accurate data collection and management according to the needs of different scenarios, thereby improving the overall data quality.

[0036] 3. Empowerment of edge intelligence (AI) reasoning: Based on edge intelligence (AI), the system has intelligent analysis and control capabilities on the edge side, and can achieve real-time intelligent (AI) reasoning and response close to the data source. This function not only reduces the delay of data uploading to the cloud, but also reduces bandwidth consumption, making the IoT system more intelligent and efficient.

[0037] The industrial Internet of Things edge gateway system of the present invention has the advantages of strong data compatibility, flexible and efficient data collection, and edge intelligence (AI) intelligent empowerment; it simplifies complex processes, improves the intelligence level of edge gateways, and expands the application scenarios of traditional edge gateway systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the embodiments of the present invention 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 embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0039] Figure 1 This is a schematic diagram of the structure of the industrial Internet of Things edge gateway system according to an embodiment of the present invention;

[0040] Figure 2 This is an example diagram of the conversion process of the data conversion module;

[0041] Figure 3 It is a structural diagram of the data acquisition module;

[0042] Figure 4 This is a schematic diagram of the processing flow of the northbound AI inference module. DETAILED DESCRIPTION

[0043] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0044] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0045] Embodiment 1:

[0046] like Figure 1 As shown, an embodiment of the present invention provides an industrial Internet of Things edge gateway system, including:

[0047] A data conversion module is used to convert source JSON data in a specific format into target JSON data in a template format through a predefined template;

[0048] The data acquisition module uses a configurable data acquisition pipeline as a northbound application node, binds it to the southbound device node, receives data from the device, and dynamically adjusts the data acquisition strategy through configuration management;

[0049] The northbound AI inference module is used to bind to the southbound device node, perform edge-side model inference based on the data sent by the device, and obtain anomaly detection results or prediction results; then use the internal API of the edge gateway system to write to the internal data points of the PLC or send information notifications to the specified cloud.

[0050] As an implementation of an embodiment of the present invention, the data conversion module includes:

[0051] The acquisition unit is used to obtain two JSON strings: one is the source JSON document and the other is the template JSON document; these strings are parsed into modifiable and traversable JSON objects. The source JSON document serves as a data repository, while the template JSON document defines the expected structure of the final output.

[0052] The recursive processing unit is used to traverse each element in the template JSON document through a recursive function and process them one by one:

[0053] Ⅰ String node (including JSON Pointer): If the value of the current node is recognized as a JSON Pointer, the system will use it to obtain the corresponding value from the source JSON document and assign the value to the current node; otherwise, it will not be processed.

[0054] Ⅱ Object node: If the current node is an object (Object), the recursive function will continue to traverse each key-value pair in the object until it recurses to a leaf node and goes to step Ⅰ.

[0055] III Array node: If the current node is an array, the recursive function will continue to traverse each element in the array, and continue recursively processing to the leaf node, and go to step I.

[0056] The pointer parsing and value replacement unit is used to parse the pointer and extract the corresponding value from the source JSON document for the string node containing the JSON Pointer, and use the extracted value to replace the value of the corresponding node in the template. Through JSONPointer, it is possible to accurately access the deeply nested fields in the source JSON, ensuring that the output JSON document maintains the structure required by the template, but the data content has been replaced by the actual value.

[0057] The output unit is used to serialize the modified template JSON document into a string format after completing the traversal and replacement of all elements in the template; the converted JSON string represents the final output, its structure conforms to the original template definition, and its content is filled with the source JSON data.

[0058] Specific examples are:

[0059] like Figure 2 As shown in the figure, the first row of rectangles represents the key in the key:value pair, the second row represents the value in the key-value pair, the curved arrow represents getting the key value of a JSON Object, the straight arrow represents getting the element value of a JSONArray, the hollow arrow represents ":" in the key:value pair, and the parallelogram represents the JSON POINTER, which is a String type value in the actual algorithm process. Specifically:

[0060] 1. Receive the source JSON document and the template JSON document, and start recursively traversing all JSON nodes of the template JSON document;

[0061] 2. Recursively reach the leaf node user_name of the template JSON document and determine whether its value / name is a JSONPOINTER. If yes, find the value "Alice" corresponding to the key name in the source JSON document, assign this value to the current leaf node, and terminate the recursive branch; if no, terminate the recursive branch;

[0062] 3. Arrive at the leaf node user_age of the template JSON document and determine whether its value / details / age is a JSONPOINTER. If so, find the value "30" corresponding to the key details->age in the source JSON document, assign this value to the current leaf node, and terminate the recursive branch. If not, terminate the recursive branch.

[0063] 4. Arrive at the leaf node city of the template JSON document and determine whether its value / details / city is a JSONPOINTER. If so, find the value "Wonderland" corresponding to the key details->city in the source JSON document, assign this value to the current leaf node, and terminate the recursive branch; if not, terminate the recursive branch;

[0064] 5. Arrive at the leaf node abilities of the template JSON document and determine whether its value / skills is a JSONPOINTER. If so, find the value "Swordsmanship" corresponding to the key skills in the source JSON document, assign this value to the current leaf node, and terminate the recursive branch; if not, terminate the recursive branch;

[0065] 6. All recursive branches terminate and return the template JSON document.

[0066] The advantages of the data conversion module are:

[0067] 1. Dynamic and flexible data mapping: Using JSON Pointer references makes data mapping very dynamic and flexible. There is no need to manually write a lot of code for each new conversion. Especially when the format of an existing data transmission scenario needs to be adjusted, you only need to simply change the template.

[0068] 2. Support for nested data of arbitrary depth: The recursive traversal mechanism ensures that nested structures of arbitrary depth can be properly handled, thereby achieving seamless filling of complex data formats.

[0069] 3. Reduce development burden: Through the automated data mapping process, the present invention reduces the time and effort required to convert data into the target format, and is particularly suitable for application scenarios that require repeated data processing and secondary development.

[0070] As an implementation method of the embodiment of the present invention, Figure 3 As shown, the data collection strategy includes: runtime strategy, post-processing strategy, and storage strategy.

[0071] Further, runtime policies include:

[0072] Data deduplication: Provides a configuration option for whether deduplication is required. When the deduplication function is enabled, the system automatically detects and removes duplicate data to ensure data uniqueness.

[0073] Decimal Places: You can configure the number of decimal places to retain for numerical data to control data accuracy and meet the needs of different application scenarios.

[0074] Dynamic collection interval: supports the function of dynamically adjusting the data collection interval, automatically adjusting the collection frequency according to the changes in device data, thereby optimizing collection efficiency and reducing resource consumption.

[0075] Anomaly detection and processing strategies: Provides algorithm configuration options for detecting and processing anomalies in the data collection process, such as missing data interpolation, outlier filtering, and other strategies to ensure data integrity and validity.

[0076] Edge machine learning models: Users can enable machine learning-based edge models to perform real-time analysis of collected data, such as trend prediction and anomaly detection, so that preliminary data processing can be performed directly during the data collection process to improve the level of intelligence.

[0077] Further, post-processing strategies include:

[0078] Append timestamp: This function supports appending timestamps to each piece of collected data to ensure the temporal consistency of the data in subsequent analysis.

[0079] Data aggregation method: The system supports the configuration of aggregate calculations on the collected data, such as calculating the average, maximum, minimum, etc. by time period, in order to reduce the amount of data and enhance data availability.

[0080] Time series anomaly detection: On the basis of supporting dynamic adjustment of the collection interval, the change frequency of the collected data items is close to the change frequency of the real data. The deep learning time series anomaly detection model can be used to detect anomalies in the device operation status.

[0081] Furthermore, the storage strategy includes: file storage strategy and database storage strategy.

[0082] The file storage strategy includes:

[0083] File path configuration: Users can set the storage path of data acquisition files to specify the storage location, thereby facilitating data management.

[0084] Maximum file size: Users can define the maximum size of data storage files in MB. When the file size exceeds the set value, a new file will be automatically opened to store data to ensure easy manageability of data files.

[0085] Data compression algorithm selection: Users can select different data compression algorithms (such as Gzip, Snappy, etc.) to reduce the volume of data storage and adapt to different storage resource situations.

[0086] The database storage strategy includes:

[0087] Database type selection: Supports storing collected data in various types of databases (such as SQL database, SQLite database). Users can choose the appropriate storage method according to their needs to ensure data persistence and convenient access.

[0088] Furthermore, the processing flow of data collection includes:

[0089] Step 21: Initialization and configuration loading

[0090] ○ Read the user-configured collection strategy from the configuration management module, including runtime strategy, post-processing strategy and storage strategy.

[0091] ○Initialize the resources needed for collection

[0092] ■If data deduplication is enabled, set the variable pre_data is_duplicated

[0093] ■If data cleaning is enabled, set the variables mean and standard deviation standard_deviation

[0094] ■If decimal point precision is set, set the variable precision

[0095] ■If dynamic collection interval is enabled, set the variables interval min_intervalmax_intervalincrease_factor decrease_factor

[0096] ■If you want to save the data to a file, set the variable file_path

[0097] ■ If you want to set a maximum size for a single file, set the variable max_size

[0098] Step 22: Data Collection Phase (Runtime Strategy)

[0099] ○Data deduplication: Compare the collected data of the current item with pre_data to see if they are repeated

[0100] ○Collection interval adjustment: If the current data is repeated, skip the processing of the current data and assign interval = interval * decrease_factor; if it is not repeated, assign interval = interval * increase_factor, so that the collection interval is close to the actual frequency of data changes inside the device. And the degree of decrease in the collection interval caused by decrease_factor is stronger than the degree of increase in the collection interval caused by increase_factor, ensuring that most of the time, the collection frequency is higher than the frequency of data changes inside the device, ensuring timely data collection.

[0101] ○ Abnormal value cleaning: The format of each data is set to

[0102] {x1,x2,…,x n}

[0103] The mean of previous data is dynamically calculated during data collection according to the following formula:

[0104]

[0105] The standard deviation of previous data is dynamically calculated during data collection according to the following formula:

[0106]

[0107] For each value of the current data, use the following formula to determine the outlier:

[0108] x i >μ+k·σorx i <μ-k·σ

[0109] Among them, k is the threshold of the outlier value, which is set according to the actual situation. If a value in the current data is judged as an outlier, it is marked as outlier data and stored separately as an outlier data set.

[0110] ○ Decimal point precision: For numeric data, retain the corresponding number of decimal places according to precision to ensure that the data meets the user's accuracy requirements.

[0111] Step 23: Data post-processing phase (post-processing strategy)

[0112] ○Add timestamp: Determine whether to add timestamp information to each piece of data based on the configuration to ensure the temporal consistency of the data.

[0113] ○ Data aggregation: Based on user settings, the collected data is aggregated, such as calculating the average, maximum or minimum value over a specified time period, thereby reducing the amount of data and enhancing data availability.

[0114] ○ Time series anomaly detection: On the basis of supporting dynamic adjustment of the collection interval, the change frequency of the collected data items is close to the change frequency of the real data, and the deep learning time series anomaly detection model can be used to detect anomalies in the operation status of the equipment. If the data is marked as abnormal data, it will be saved separately as abnormal data to facilitate subsequent manual analysis.

[0115] Step 24: Data storage phase (storage strategy)

[0116] ○ File storage:

[0117] ■ According to user configuration, the processed data is stored in the file corresponding to file_path.

[0118] ■File size management: When a file reaches the specified file_size, a new file is automatically created for storage to ensure orderly file management.

[0119] ■Data compression: compress the stored files to reduce storage space. Users can choose to use compression algorithms such as Gzip and Snappy.

[0120] ○ Database storage:

[0121] ■If you choose database storage, store the data in the specified database (SQL or NoSQL) to ensure persistent storage and convenient query of the data.

[0122] The advantages of the data acquisition module are:

[0123] 1. High configurability: It provides a variety of customizable configuration items, which greatly improves the flexibility of data collection and adapts to the needs of diversified industrial equipment.

[0124] 2. Data reliability: Ensure the timing accuracy, uniqueness and integrity of collected data through various strategies such as adding timestamps, deduplication, and anomaly detection.

[0125] 3. Automatic adjustment and intelligent analysis: By dynamically adjusting the collection interval and applying edge machine learning models, the collection efficiency is effectively improved, the waste of system resources is reduced, and the preliminary analysis of the collected data is achieved.

[0126] 4. Flexible storage strategies: By supporting multiple storage strategies for files and databases, users can flexibly choose data storage methods according to their needs, ensuring that data management and retrieval are more efficient and convenient.

[0127] As an implementation method of the embodiment of the present invention, the main architecture of the traditional edge gateway system is a north-south architecture. The southbound node supports a variety of device connection protocols and is responsible for connecting various sensors, PLCs and other devices; the northbound node is responsible for processing and reporting the data of the corresponding southbound node, such as reporting to the self-built industrial Internet of Things cloud platform, Huawei Cloud Internet of Things cloud platform, cloud database and other cloud services. The data transmission between the northbound and southbound nodes is completed by the message bus inside the system, that is, the data collected by the southbound node will be forwarded to the northbound node after intermediate processing, and then reported to the cloud service by the northbound node.

[0128] Based on the above architecture, in addition to connecting to the cloud and reporting data, the northbound node can also expand other richer application scenarios. One direction is to connect the intelligence (AI) capabilities to the edge gateway system based on the northbound node. Based on the capabilities of the northbound node, the embodiment of the present invention proposes a universal edge gateway intelligent (AI) reasoning module that supports the deployment of mainstream data prediction, anomaly detection and other models. Taking the southbound device as an industrial PLC as an example, the northbound node receives the device data as the model input, runs the edge-side model reasoning, obtains the anomaly detection results or prediction results, etc., and then writes the internal data points of the PLC through the internal API of the edge gateway system. In this way, the intelligent control of the industrial production process is realized on the edge side.

[0129] Further, if Figure 4 As shown in the figure, the processing flow of the northbound AI inference module includes:

[0130] Step 31: Data Collection and Processing

[0131] The southbound device nodes of the edge gateway framework communicate with industrial devices (such as sensors and PLCs) through a variety of industrial protocols and are responsible for collecting raw data from industrial devices. The collected raw data is forwarded to the northbound AI reasoning module through the traditional edge gateway framework data bus. The data will first pass through the data preprocessing module for necessary data cleaning, normalization, and feature extraction to ensure that the data input to the intelligent (AI) model is consistent, accurate, and robust.

[0132] Step 32: Intelligent (AI) Reasoning Execution

[0133] After data preprocessing, the data will be transmitted to the northbound AI reasoning module, which supports the deployment of a variety of mainstream intelligent (AI) models, such as time series models for data prediction (such as long short-term memory network LSTM) or classification models for detecting industrial equipment anomalies (such as support vector machine SVM or decision tree). The northbound intelligent (AI) reasoning framework performs reasoning calculations based on the set configuration and intelligent (AI) model to obtain results such as equipment operation status prediction, fault warning, performance evaluation, etc.

[0134] Step 33: Result Feedback and Control Operation

[0135] The inference results generated by the northbound AI inference module are used for further control operations through the internal API interface of the edge gateway. If an abnormal state is detected, the system can dynamically adjust the equipment parameters by automatically writing the internal points of the PLC to prevent failures or reduce equipment losses. In addition, the prediction results can also be used to optimize the production process, such as pre-adjusting the production line speed, automatically allocating production resources, etc., to improve production efficiency and equipment utilization.

[0136] Step 34: Data reporting and system linkage

[0137] After completing the edge-side reasoning, the northbound node reports the relevant data (such as prediction results, abnormal alarm information, etc.) to the cloud platform through the northbound interface. The cloud platform can include a self-built industrial Internet of Things platform, a third-party cloud service platform (such as Huawei Cloud IoT platform), or an enterprise's internal data center. Through the centralized reporting and management of data, global monitoring of the equipment's operating status and further data analysis and mining can be achieved.

[0138] The advantages of the northbound AI inference module are:

[0139] 1. Edge-side reasoning reduces latency: By deploying intelligent (AI) reasoning functions on the edge, the time delay for data transmission to the cloud is significantly reduced, thereby achieving real-time response and control of industrial sites, which is especially suitable for production scenarios that are sensitive to latency.

[0140] 2. Flexible model deployment and hot update: The northbound intelligent (AI) reasoning framework supports flexible deployment and hot update of models, and can quickly deploy or replace intelligent (AI) models according to actual production needs to ensure that the system can continue to adapt to different application scenarios and optimization needs.

[0141] 3. Data privacy and security assurance: Data processing and intelligent (AI) reasoning operations are completed on the edge side, effectively reducing the risk of data privacy leakage caused by uploading large-scale production data to the cloud, and ensuring the security and privacy of industrial production data.

[0142] 4. Modularity and scalability: The system adopts a modular design to facilitate functional expansion. The northbound reasoning framework and the southbound device nodes are relatively independent, but they communicate efficiently through the message bus, which allows for the flexible addition of new devices or the introduction of new intelligent (AI) reasoning models.

[0143] Embodiment 2:

[0144] The embodiment of the present invention also provides a data processing method of an industrial Internet of Things edge gateway system, including:

[0145] Step S1: Convert source JSON data in a specific format into target JSON data in a template format using a predefined template;

[0146] Step S2: Use the configurable data acquisition pipeline as the northbound application node, bind it to the southbound device node, receive data from the device, and dynamically adjust the data acquisition strategy through configuration management;

[0147] Step S3: Bind the northbound AI reasoning module to the southbound device node, perform edge-side model reasoning based on the data sent by the device, and obtain anomaly detection results; then use the internal API of the edge gateway system to write to the PLC internal data points or send information notifications to the specified cloud.

[0148] Preferably, step S1 comprises:

[0149] Get two JSON strings; one JSON string is the source JSON document, and the other JSON string is the template JSON document;

[0150] Traverse each element in the template JSON document through a recursive function;

[0151] For a string node containing a JSON Pointer, parse the pointer and extract the corresponding value from the source JSON document, and use the extracted value to replace the value of the corresponding node in the template;

[0152] After traversing and replacing all elements in the template, the modified template JSON document is serialized into a string format.

[0153] Preferably, the data collection strategy includes: a runtime strategy, a post-processing strategy and a storage strategy.

[0154] The embodiments described above are only descriptions of the preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the protection scope determined by the claims of the present invention.

Claims

1. An industrial Internet of Things edge gateway system, characterized in that: include: A data conversion module is used to convert source JSON data in a specific format into target JSON data in a template format through a predefined template; The data acquisition module uses a configurable data acquisition pipeline as a northbound application node, binds it to the southbound device node, receives data from the device, and dynamically adjusts the data acquisition strategy through configuration management; The northbound AI inference module is used to bind to the southbound device node, perform edge-side model inference based on the data sent by the device, and obtain anomaly detection results; then the internal API of the edge gateway system is used to write to the internal data points of the PLC or send information notifications to the specified cloud.

2. The industrial Internet of Things edge gateway system according to claim 1, characterized in that: The data conversion module includes: The acquisition unit is used to acquire two JSON strings, wherein one JSON string is a source JSON document, and the other JSON string is a template JSON document; A recursive processing unit, used to traverse each element in the template JSON document through a recursive function; The pointer parsing and value replacement unit is used to parse the pointer of the string node containing the JSON Pointer and extract the corresponding value from the source JSON document, and use the extracted value to replace the value of the corresponding node in the template; The output unit is used to serialize the modified template JSON document into a string format after traversing and replacing all elements in the template.

3. The industrial Internet of Things edge gateway system as claimed in claim 2, characterized in that: Data collection strategies include: runtime strategy, post-processing strategy and storage strategy.

4. A data processing method for an industrial Internet of Things edge gateway system, characterized in that: include: Step S1: Convert source JSON data in a specific format into target JSON data in a template format using a predefined template; Step S2: Use the configurable data acquisition pipeline as the northbound application node, bind it to the southbound device node, receive data from the device, and dynamically adjust the data acquisition strategy through configuration management; Step S3: Bind the northbound AI reasoning module to the southbound device node, perform edge-side model reasoning based on the data sent by the device, and obtain anomaly detection results; then use the internal API of the edge gateway system to write to the PLC internal data points or send information notifications to the specified cloud.

5. The data processing method of the industrial Internet of Things edge gateway system according to claim 4, characterized in that: Step S1 includes: Get two JSON strings; one JSON string is the source JSON document, and the other JSON string is the template JSON document; Traverse each element in the template JSON document through a recursive function; For a string node containing a JSON Pointer, parse the pointer and extract the corresponding value from the source JSON document, and use the extracted value to replace the value of the corresponding node in the template; After traversing and replacing all elements in the template, the modified template JSON document is serialized into a string format.

6. The data processing method of the industrial Internet of Things edge gateway system according to claim 5, characterized in that: Data collection strategies include: runtime strategy, post-processing strategy and storage strategy.

Images