Multi-protocol adaptive communication system and method for industrial Internet of Things

Abstract

The invention provides a multi-protocol adaptive communication system and method for an industrial internet of things, and the system comprises a physical connection layer which collects original data based on a clustering node deployment model, and carries out the preprocessing of the original data; the AI protocol processing layer comprises a protocol identification module and a data fusion module, and the AI protocol processing layer is used for receiving the data preprocessed by the physical connection layer, identifying a protocol type corresponding to the data by utilizing a lightweight deep learning model, and fusing the data by adopting an improved evidence theory weighted fusion algorithm; the fused data is converted into a unified intermediate format; and the intelligent conversion execution layer is used for converting the intermediate format data into a target protocol format in combination with the parameter optimization model and sending the target protocol format data to a corresponding device or system. A three-level architecture is adopted, the advantages of the AI technology and the hierarchical gateway are combined, the adaptive capacity, real-time performance and accuracy of protocol conversion are improved, hardware resource consumption and software complexity are reduced, and the method is suitable for complex scenes of the industrial Internet of Things.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of industrial Internet of Things, and in particular to a multi-protocol adaptive communication system and method for industrial Internet of Things. BACKGROUND

[0002] With the deep development of industrial Internet of Things IIoT, the number of heterogeneous devices accessed in industrial scenarios grows exponentially, covering sensors, controllers, actuators, gateways, servers and other types. Due to differences in manufacturers, application scenarios and functional requirements, the communication protocols used by these devices show a high degree of diversification. The current mainstream industrial Internet of Things protocols can be divided into several categories: Industrial control protocols: such as EtherCAT, PROFINET, Ethernet / IP, which focus on real-time and determinism, and are widely used in production line real-time control scenarios, with data transmission delay requirements usually in the order of microseconds to milliseconds; General monitoring protocols: such as Modbus (including Modbus RTU / ASCII / TCP), DNP3, which are commonly used for industrial device state monitoring and data acquisition, and the protocol design is simple but the compatibility needs to be adapted to different interfaces (such as RS485, Ethernet); Device interconnection protocols: such as OPC UA, which is a cross-platform data interaction standard, supports complex data models and security authentication, and is commonly used for data interconnection between enterprise-level management layer and control layer; Low-power wireless protocols: such as LoRaWAN, Zigbee, Bluetooth Low Energy (BLE), which are suitable for low-bandwidth, long-range or short-range wireless sensor networks, focusing on power optimization and coverage; Message queue protocols: such as MQTT, based on the publish-subscribe model, lightweight and easy to implement, widely used in device and cloud bidirectional communication in cloud-edge collaboration scenarios.

[0003] The diversity of protocols leads to a serious "protocol barrier" problem: different protocols have significant differences in data format, transmission mechanism and interaction logic, for example, EtherCAT uses a master-slave communication mode and a mailbox communication mechanism, while MQTT is based on a client-server architecture and a topic filtering mechanism, which directly leads to difficulties in information exchange between heterogeneous devices. In addition, the performance requirements of different protocols differ greatly - industrial control protocols require microsecond-level real-time performance, low-power protocols require nanowatt-level power control, and general protocols require high compatibility, which further exacerbates the complexity of protocol conversion. Existing multi-protocol conversion technologies have obvious limitations: Multi-protocol stack parallel processing technology: conversion is realized by integrating multiple independent protocol stacks in hardware, but dedicated hardware resources (such as processor cores, memory) need to be allocated for each protocol, resulting in high hardware cost, high power consumption, and data interaction between protocol stacks is prone to conflict; Traditional AI protocol recognition technology: relies on a large amount of labeled data to train deep learning models, although it has certain adaptive ability, but the accuracy of feature extraction for low-bandwidth protocols such as LoRaWAN is insufficient, and the model inference time is relatively long, it is difficult to meet the conversion needs of real-time protocols such as EtherCAT; Hierarchical gateway architecture: although the system logic is simplified through hierarchical design, the data forwarding delay between layers is high, and protocol recognition only relies on fixed rule library, which cannot dynamically adapt to new protocols or protocol variants (such as Modbus Plus, PROFINET IO-Link), and the adaptive ability is insufficient.

[0004] Therefore, there is an urgent need for a multi-protocol adaptive communication technology that can compatible with diversified protocols, balance real-time and adaptability, and reduce software and hardware complexity, to break through the "protocol barrier" of industrial Internet of Things. SUMMARY

[0005] The technical problem to be solved by the present application is to provide a multi-protocol adaptive communication system and method for industrial Internet of Things, to compatible with diversified communication protocols in industrial scenarios, improve the adaptive ability, real-time and accuracy of protocol conversion, and reduce hardware resource consumption and software complexity.

[0006] To solve the above technical problems, the technical solution adopted by the present application is: The multi-protocol adaptive communication system for industrial Internet of Things comprises a physical connection layer, an AI protocol processing layer and an intelligent conversion execution layer. The physical connection layer is configured with multiple types of interfaces and a clustering node deployment module, and is used to establish physical connection with heterogeneous devices in the industrial Internet of Things through multiple types of interfaces, and collect and preprocess raw data based on the clustering node deployment model; The AI protocol processing layer comprises a protocol recognition module and a data fusion module, the protocol recognition module adopts a lightweight deep learning model to identify the protocol type, the data fusion module adopts an improved evidence theory weighted fusion algorithm, the AI protocol processing layer is used to receive the data preprocessed by the physical connection layer, identify the protocol type corresponding to the data by using the lightweight deep learning model, perform fusion processing on the data by using the improved evidence theory weighted fusion algorithm, and convert the fused data into a unified intermediate format; An intelligent conversion execution layer is configured to receive the intermediate format data converted by the AI protocol processing layer, convert the intermediate format data into a target protocol format according to a target protocol requirement in combination with a parameter optimization model, and send the target protocol format data to a corresponding device or system.

[0007] The multi-type interface of the physical connection layer includes RS485, Ethernet, a wireless module, RS232 and CAN interfaces, and can adapt to access requirements of multiple protocols such as Modbus, EtherCAT, PROFINET, MQTT, LoRaWAN, Zigbee, OPC UA and DNP3. The wireless module includes oRa, Bluetooth and Wi-Fi modules. The clustering node deployment model in the physical connection layer optimizes node positioning by the following formula: ; wherein is a networking prediction value, is a data distribution dimension, and r is a node distance, is a random error.

[0008] In the improved evidence theory weighted fusion algorithm, the evidence normalization composite weight is calculated The data is fused and converted into the same intermediate format, and the evidence normalization composite weight is calculated The calculation formula is: ; ; wherein is a composite weight of the evidence, n is the number of the evidence, is a posterior weight, is a prior weight.

[0009] The parameter optimization model is: ; wherein is a node positioning output characteristic quantity, is an information clustering interval, is a maximum sampling threshold.

[0010] The parameter optimization model is implemented by a fuzzy iterative algorithm, and includes a node balanced scheduling fusion rule: ; wherein is a transmission communication graph parameter of the Internet of Things gateway multi-protocol conversion, and together with constitutes a transmission communication graph set, A distance parameter between a source and a sink node representing multi-protocol conversion of an Internet of Things gateway, used to measure the spatial distance between the source node and the sink node.

[0011] A multi-protocol adaptive communication method for industrial Internet of Things, comprising the following steps: S1: The physical connection layer collects raw data of heterogeneous devices in the industrial Internet of Things through multiple type interfaces, and pre-processes the raw data; S2: The AI protocol processing layer receives the pre-processed data, identifies the protocol type using a lightweight deep learning model, fuses the data using an improved evidence theory weighted fusion algorithm, and converts it into a unified intermediate format; S3: The intelligent conversion execution layer receives the intermediate format data, converts it into a target protocol format according to the requirements of the target protocol and sends it.

[0012] In the above step S1, the pre-processing includes removing electromagnetic interference noise in the raw data using filtering processing, and unifying raw data of different formats into a specific data frame structure, and the raw data is derived from one or more of devices using Modbus, EtherCAT, PROFINET, MQTT, LoRaWAN, Zigbee, OPC UA and DNP3 protocols.

[0013] In the above step S2, when identifying the protocol type, the frame header identifier, check bit format, data length rule, protocol specific field and other characteristics of the data frame are extracted for judgment.

[0014] In the above step S3, the dynamic adjustment of the conversion strategy includes using compression algorithm to reduce data volume for protocols with high real-time requirements; optimizing the transmission frame structure to reduce energy consumption for low-power wide-area network protocols; prioritizing data integrity for general protocols; and adapting QoS level parameters for message queue type protocols.

[0015] The multi-protocol adaptive communication system and method for industrial Internet of Things mentioned in the present application adopts a three-level architecture, combines AI technology and the advantages of layered gateways, improves the adaptive ability, real-time performance and accuracy of protocol conversion, reduces hardware resource consumption and software complexity, and is suitable for complex scenarios of industrial Internet of Things. BRIEF DESCRIPTION OF DRAWINGS

[0016] The present application will be further described below in conjunction with the drawings and examples: Figure 1 is a system architecture diagram of the present application; Figure 2 is a method flowchart of the present application. DETAILED DESCRIPTION

[0017] The technical scheme of the present application will be described in detail below in combination with the drawings and examples.

[0018] The multi-protocol adaptive communication system for the industrial Internet of Things comprises a physical connection layer, an AI protocol processing layer and an intelligent conversion execution layer. The physical connection layer is configured with a multi-type interface and a clustering node deployment module, and is used to establish physical connection with heterogeneous devices in the industrial Internet of Things through the multi-type interface, collect raw data and perform preprocessing based on the clustering node deployment model; The AI protocol processing layer comprises a protocol identification module and a data fusion module, the protocol identification module adopts a lightweight deep learning model to identify the protocol type, the data fusion module adopts an improved evidence theory weighted fusion algorithm, the AI protocol processing layer is used to receive the data preprocessed by the physical connection layer, identify the protocol type corresponding to the data by using the lightweight deep learning model, perform fusion processing on the data by using the improved evidence theory weighted fusion algorithm, and convert the fused data into a unified intermediate format; The intelligent conversion execution layer is used to receive the intermediate format data converted by the AI protocol processing layer, convert the intermediate format data into a target protocol format according to the target protocol requirement in combination with a parameter optimization model, and send the target protocol format data to the corresponding device or system.

[0019] The multi-type interface of the physical connection layer comprises RS485, Ethernet, a wireless module, RS232 and CAN interface, and can adapt to the access requirements of multiple protocols such as Modbus, EtherCAT, PROFINET, MQTT, LoRaWAN, Zigbee, OPC UA and DNP3, and the wireless module comprises oRa, Bluetooth and Wi-Fi module. The clustering node deployment model in the physical connection layer realizes node positioning optimization through the following formula: ; Wherein is a network prediction value, is a data distribution dimension, and r is a node distance, is a random error.

[0020] In the improved evidence theory weighted fusion algorithm, the evidence normalization composite weight is calculated to perform fusion processing on the data and convert the data into the same intermediate format, and the evidence normalization composite weight is calculated. ; ; Wherein, is a composite weight of evidence, n is the number of evidence, is a posterior weight, The prior weight is determined.

[0021] The parameter optimization model is: ; wherein, is a node positioning output feature quantity, is an information clustering interval, is a maximum sampling threshold.

[0022] The parameter optimization model is realized through a fuzzy iteration algorithm, and contains a node balanced scheduling fusion rule: ; wherein, is a transmission communication graph parameter of the Internet of Things gateway multi-protocol conversion, and together with etc. constitutes a transmission communication graph set, indicates a distance parameter between the Source and Sink nodes of the Internet of Things gateway multi-protocol conversion, and is used to measure the spatial distance between the source node and the sink node, and is one of the key parameters of node balanced scheduling and routing planning.

[0023] A multi-protocol adaptive communication method for industrial Internet of Things, comprising the following steps: S1: The physical connection layer collects the original data of heterogeneous devices in the industrial Internet of Things through multiple type interfaces, and pre-processes the original data; S2: The AI protocol processing layer receives the pre-processed data, identifies the protocol type using a lightweight deep learning model, fuses the data using an improved evidence theory weighted fusion algorithm, and converts it into a unified intermediate format; S3: The intelligent conversion execution layer receives the intermediate format data, converts it into a target protocol format according to the target protocol requirements and sends it.

[0024] In the above step S1, the pre-processing includes removing electromagnetic interference noise in the original data using filtering processing, and unifying different formats of original data into a specific data frame structure. The original data is derived from one or several of the devices using Modbus, EtherCAT, PROFINET, MQTT, LoRaWAN, Zigbee, OPC UA and DNP3 protocols.

[0025] In the above step S2, when identifying the protocol type, the frame header identifier, check bit format, data length rule, protocol specific field (such as MQTT topic field, LoRaWAN DevEUI field) and other characteristics of the data frame are extracted for judgment.

[0026] In step S3, the dynamic adjustment of the conversion strategy includes using a compression algorithm to reduce data volume for protocols with high real-time requirements (such as EtherCAT and PROFINET), optimizing the transmission frame structure to reduce energy consumption for Low Power Wide Area Network protocols (such as LoRaWAN), prioritizing data integrity for general-purpose protocols, and adapting QoS level parameters for message queue protocols (such as MQTT).

[0027] Embodiment 1: System composition: the physical connection layer uses a multi-interface adaptation circuit to connect industrial devices of different protocols, including: connecting Modbus RTU temperature and humidity sensors through RS485 interfaces, connecting PROFINET controllers and EtherCAT servo motors through Ethernet interfaces, connecting LoRaWAN smoke detectors through LoRa wireless modules, connecting MQTT smart meters through Wi-Fi modules, and connecting DNP3 smart circuit breakers through CAN interfaces. The data preprocessing module filters and processes the frame structure of the collected data (such as anti-interference filtering for LoRaWAN data and pulse noise filtering for Modbus data).

[0028] The protocol intelligent identification module of the AI protocol processing layer uses a trained lightweight CNN-LSTM fusion model to extract features of different protocols: extracting synchronization signal frame headers and process data object (PDO) fields for EtherCAT data, extracting fixed headers and topic fields for MQTT data, and extracting DevEUI and FCtrl control fields for LoRaWAN data, achieving accurate identification of protocol types (with an accuracy of 99.3%). The data fusion and conversion module uses an improved evidence theory weighted fusion algorithm to perform time and space alignment and weight distribution on multi-protocol data (such as temperature sensor data, motor operation data, and smoke data), and then converts it to JSON intermediate format.

[0029] The adaptive conversion engine of the intelligent conversion execution layer dynamically adjusts the strategy according to the target protocol type: when the target protocol is PROFINET, it uses a differential compression algorithm on JSON data to reduce transmission volume by 40%; when the target protocol is LoRaWAN, it optimizes the frame structure to shorten air transmission time (reducing 20% energy consumption); when the target protocol is MQTT, it adapts the QoS level according to data importance (critical alarm data is set to QoS 2, and ordinary monitoring data is set to QoS0); when the target protocol is OPC UA, it increases data signature and timestamp to ensure integrity.

[0030] Workflow: The physical connection layer collects and preprocesses multi-protocol data and transmits it to the AI protocol processing layer; the AI protocol processing layer completes protocol recognition, data fusion, and intermediate format conversion and transmits it to the intelligent conversion execution layer; the intelligent conversion execution layer dynamically adjusts the conversion strategy according to the target protocol and outputs target format data and sends it, realizing multi-protocol adaptive communication.

[0031] In a certain smart factory comprehensive scenario, the system needs to handle the conversion requirements of 8 types of protocols, which are specifically manifested as: The protocol classification module classifies protocols into real-time control type (EtherCAT, PROFINET), low-power wireless type (LoRaWAN, Zigbee), general monitoring type (Modbus, DNP3), and message interaction type (MQTT, OPC UA) through device library feature matching. The adaptive identification module uses a high-frame-rate inference model (identification time ≤0.5ms) for real-time control type protocols and a lightweight feature extractor (model size reduced by 60%) for low-power type protocols. The differentiated fusion module uses a timestamp-first fusion rule (fusion delay ≤1ms) for real-time control type data and a sparse sampling fusion strategy (reducing data volume by 30%) for low-power type data. The intelligent conversion execution layer has a conversion success rate of more than 99.5% for different protocols, among which the real-time error of EtherCAT to PROFINET is ≤50μs, and the energy optimization rate of LoRaWAN to MQTT is 25%.

Claims

1. A multi-protocol adaptive communication system for industrial internet of things, characterized in that, The application relates to an intelligent protocol conversion system and method for industrial Internet of Things. The physical connection layer is configured with a multi-type interface and a clustering node deployment module, is used for establishing physical connection with heterogeneous devices in the industrial Internet of Things through the multi-type interface, collecting original data based on the clustering node deployment model and pre-processing the original data; The AI protocol processing layer comprises a protocol identification module and a data fusion module, the protocol identification module adopts a lightweight deep learning model to identify protocol types, the data fusion module adopts an improved evidence theory weighted fusion algorithm, the AI protocol processing layer is used for receiving the data pre-processed by the physical connection layer, identifying the protocol types corresponding to the data by using the lightweight deep learning model, performing fusion processing on the data by using the improved evidence theory weighted fusion algorithm, and converting the fused data into a unified intermediate format; The intelligent conversion execution layer is used for receiving the intermediate format data converted by the AI protocol processing layer, converting the intermediate format data into a target protocol format according to target protocol requirements and sending the target protocol format to corresponding devices or systems in combination with a parameter optimization model. The multi-type interface of the physical connection layer comprises RS485, an Ethernet, a wireless module, RS232 and a CAN interface, can adapt to the access requirements of multiple protocols such as Modbus, EtherCAT, PROFINET, MQTT, LoRaWAN, Zigbee, OPC UA and DNP3, and the wireless module comprises oRa, a Bluetooth and a Wi-Fi module.

2. The industrial internet of things oriented multi-protocol adaptive communication system according to claim 1, characterized in that, The clustering node deployment model in the physical connection layer realizes node positioning optimization through the following formula:

3. The industrial internet of things oriented multi-protocol adaptive communication system according to claim 1, characterized in that, The parameter optimization model is: ； wherein is the networked prediction value, is the data distribution dimension, r is the node distance, is the random error.

4. The industrial internet of things oriented multi-protocol adaptive communication system according to claim 1, characterized in that, The improved evidence theory weighted fusion algorithm calculates an evidence normalization composite weight The data is fused and converted into the same intermediate format, and the evidence normalization composite weight is calculated The calculation formula is: ； ； wherein, is the composite weight of the evidence, n is the number of evidences, is the posterior weight, is the prior weight.

5. The industrial internet of things oriented multi-protocol adaptive communication system according to claim 4, characterized in that, The parameter optimization model is realized by a fuzzy iteration algorithm and comprises a node balanced scheduling fusion rule. ； wherein, outputting a feature quantity for node positioning, an information clustering interval, a maximum sampling threshold.

6. The industrial internet of things oriented multi-protocol adaptive communication system according to claim 5, characterized in that, The application further relates to an intelligent protocol conversion method for industrial Internet of Things. ； wherein, is a transmission communication graph parameter for the Internet of Things gateway multi-protocol conversion, and together constitute a transmission communication graph set, represents a distance parameter between the Source and Sink nodes of the Internet of Things gateway multi-protocol conversion, used to measure the spatial distance between the source node and the sink node.

7. A multi-protocol adaptive communication method for industrial internet of things, characterized in that, In step S1, the pre-processing comprises removing electromagnetic interference noise in the original data by using a filtering process and unifying different formats of the original data into a specific data frame structure. In step S2, when identifying the protocol types, the frame header identifier, the check bit format, the data length rule, the protocol specific field and other characteristics of the data frame are extracted to judge. In step S3, the dynamic adjustment of the conversion strategy comprises adopting a compression algorithm to reduce the data amount for a protocol with high real-time requirement, optimizing the transmission frame structure to reduce energy consumption for a low-power wide-area network protocol, preferentially guaranteeing data integrity for a general protocol and adapting QoS level parameters for a message queue type protocol. ​ 8. The industrial internet of things oriented multi-protocol adaptive communication method according to claim 7, characterized in that, ​ 9. The industrial internet of things oriented multi-protocol adaptive communication method according to claim 8, characterized in that, ​ 10. The industrial internet of things oriented multi-protocol adaptive communication method according to claim 9, characterized in that, ​

Images