Equipment control system for industrial site

Abstract

The invention provides an industrial field-oriented equipment control system, and belongs to the field of equipment control, the system comprises a controller module and an AI intelligent module, the controller module is used for collecting related data of industrial field equipment, and according to the related data, the AI intelligent module is used for controlling the AI intelligent module; constructing a digital twinborn model of the industrial field equipment through an edge computing function and a digital twinborn technology; the AI intelligent module is used for predicting the working state of the digital twin model through the intelligent prediction model and performing feedback control on the industrial field equipment based on a prediction result; a Yov5 network model is adopted as the intelligent prediction model, CSPDarkNet is introduced into the Yov5 network model to serve as a backbone network, a space-time attention module is introduced into a neck module of Yov5, a self-adaptive attention mechanism is introduced into a head module, and a target neural network model is obtained; and training the target neural network to obtain an intelligent prediction model. Efficient monitoring, intelligent prediction and optimal control of field equipment are realized, and support is provided for industrial intelligent transformation.

Description

Technical Field

[0001] The present invention belongs to the field of equipment control, and particularly relates to an equipment control system for industrial sites. Background Art

[0002] In an industrial environment, a control system is used to monitor and control equipment such as industrial and chemical processes. The control system utilizes field devices distributed at key positions in the industrial process to perform its functions, and these field devices are connected to the control circuit in the control system through a process control loop. The term "field device" refers to any device that performs functions in a distributed control or process monitoring system, including all devices known or unknown currently for measurement, control, and / or monitoring of industrial processes.

[0003] Typical field devices include an equipment circuit, which enables the field device to perform conventional field device tasks, such as monitoring and measuring process parameters using one or more sensors, and / or performing process control operations using one or more control devices. Exemplary sensors include pressure sensors, level sensors, temperature sensors, and other sensors for industrial processes. Exemplary control devices include actuators, solenoid valves, valves, and other control devices.

[0004] In the prior art, the management and control of control devices require manual real-time adjustment, so the demand for manpower is large, the efficiency is low, and the processing of equipment status depends on manual experience, resulting in poor management of equipment in industrial sites and being unable to effectively avoid the harm of potential harmful factors to field devices. Summary of the Invention

[0005] In order to solve the problems of poor management of industrial sites and poor control effect of potential harmful factors in the prior art, the present invention provides an equipment control system for industrial sites.

[0006] To achieve the above object, the present invention provides the following technical solutions: An equipment control system for industrial sites, comprising a controller module and an AI intelligent module; The controller module is used to collect relevant data of industrial site equipment, and based on the relevant data, construct a digital twin model of industrial site equipment through edge computing function and digital twin technology; The described AI intelligent module is used to predict the working state of the digital twin model through an intelligent prediction model, and perform feedback control on industrial field devices based on the prediction results; the intelligent prediction model adopts the Yolov5 network model, introduces CSPDarkNet as the backbone network in the Yolov5 network model, introduces a spatio-temporal attention module in the neck module of Yolov5, and introduces an adaptive attention mechanism in the head module to obtain the target neural network model; the target neural network is trained to obtain the intelligent prediction model.

[0007] Optionally, for the input part of the intelligent prediction model, each input channel uses a 1x1 convolution kernel to combine the output of the depth convolution.

[0008] Optionally, the controller module is used to perform digital production line modeling on entities such as transfer AGVs, inbound and outbound warehousing objects, raw material accessories, conveyor lines, transfer machines, accessories, and finished products in the field devices; By coupling the visualization model and three-dimensional simulation technology, real-time mapping and tracking are performed on the transfer process, process transfer connection process, component online process, and each sub-item process of automatic assembly, realizing real-time three-dimensional visualization presentation of the actions and states of the physical production line.

[0009] Optionally, the AI intelligent module is used to collect displacement detection, weighing data, screw tightening force values, and angle feedback values of the production line through the sensing device at the device end, and clean the relevant data; By sorting out the process routes of the physical production line, key process technologies are extracted, and based on the key process technologies and the target neural network model, an analysis and prediction model for the digital twin model is constructed to predict the future state of the digital production line.

[0010] Optionally, the system further includes a medium-sized intelligent controller and a terminal platform. The medium-sized intelligent controller is used to integrate the controller module and the AI intelligent module; the terminal platform is connected to the medium-sized intelligent controller and is used to display the relevant data uploaded by the medium-sized intelligent controller through digital twin technology, perform visual inspection, display big data prediction results, and perform remote monitoring and prediction alarm.

[0011] Optionally, the medium-sized intelligent controller further includes a backplane module, and the controller module and the AI intelligent module are connected through the bus cascading method of the backplane module.

[0012] Optionally, the medium-sized intelligent controller further includes an I / O expansion module and a communication module. The I / O expansion module includes digital / analog input / output modules, and the communication module includes serial communication modules of RS485 and RS232 and a communication module that supports multiple industrial bus protocols such as EtherNet / IP, EtherCAT, or CANopen.

[0013] Optionally, the terminal platform is used to perform full - life - cycle health management on the engineering projects of on - site devices through digital twin technology, display the status and statistical analysis data of the on - site device parameters uploaded by the medium - sized intelligent controller, complete the issuance of superior instructions to the controller module and the non - intrusive downloading of programs, and perform visual viewing and operation on the engineering projects and on - site devices.

[0014] Optionally, the controller module adopts a blade - type structure design. The modules in the medium - sized intelligent controller are extended and connected using a private backplane protocol. The program is downloaded between the controller module and the programming software that performs logical control on it using an Ethernet / programming port protocol. The data acquisition and instruction issuance between the controller module and the on - site devices are carried out using ModBUS TCP / EtherNet / IP / EtherCAT protocols.

[0015] The device control system for industrial sites provided by the present invention has the following beneficial effects: Firstly, the edge - computing function can be used to realize the real - time processing of relevant data of industrial - site devices, ensuring the timeliness and accuracy of the relevant data, and improving the reliability of the digital twin model constructed based on the relevant data. By using digital twin technology, the data of physical devices is converted into virtual models, and the real - time monitoring and management of these models improve the transparency of device operation. The AI intelligent module constructs an intelligent prediction model, uses the improved Yolov5 network model to predict the working state of the digital twin model, introduces a spatio - temporal attention module in the intelligent prediction model to improve the extraction efficiency of key features in the on - site device data, and then introduces an adaptive attention mechanism to process various complex data features of the device, enhancing the control ability of the device state and facilitating the management and control of potential harmful factors. Based on the prediction results, feedback control is carried out. This control method significantly improves the management speed of industrial - site devices. Secondly, through the prediction results, potential harmful factors interfering with device operation can also be excluded, enhancing the control intensity of industrial - site devices. In this way, through the combination of digital twin technology and intelligent prediction models, the production process can be optimized, resource waste can be reduced, production efficiency can be improved, and efficient monitoring, intelligent prediction, and optimized control of on - site devices are realized, providing strong support for the industrial intelligent and digital transformation. Description of the Drawings

[0016] In order to more clearly illustrate the embodiments of the present invention and their design schemes, the accompanying drawings required for this embodiment will be briefly introduced below. The drawings described below are only partial embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0017] Figure 1 Schematic diagram of a device control system for industrial field provided by the present invention according to an exemplary embodiment.

[0018] Figure 2 Another schematic diagram of a device control system for industrial field provided by the present invention according to an exemplary embodiment.

[0019] Figure 3 Schematic diagram of the operation architecture of the device control system provided by the present invention according to an exemplary embodiment.

[0020] Figure 4 Schematic diagram of a medium-sized intelligent controller provided by the present invention according to an exemplary embodiment.

[0021] Figure 5 Hardware design block diagram of the controller CPU module provided by the present invention according to an exemplary embodiment.

[0022] Figure 6 Schematic diagram of the production line architecture provided by the present invention according to an exemplary embodiment.

[0023] Figure 7 Flowchart of a device control method for industrial field provided by the present invention according to an exemplary embodiment. Detailed implementation manners

[0024] In order to enable those skilled in the art to better understand the technical solutions of the present invention and implement them, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. The following embodiments are only used to illustrate the technical solutions of the present invention more clearly, and cannot be used to limit the protection scope of the present invention.

[0025] In order to strengthen the information collection at the project site and optimize the construction of the digital twin model, and realize visual interaction functions such as remote real-time monitoring, analysis and control, the device control system of the present invention adopts the technical route of "industrial control + Internet + intelligence". At the site, a connection is established between the "business layer" terminal platform and the "perception layer" field devices and sensors through an intelligent controller to complete the information transmission from the field information to the terminal platform and the cloud platform, and meet the relevant operations of remote terminal visual information viewing and device control. The intelligent controller is the core device of the control system, responsible for collecting on-site data and processing the instructions issued by the superior; the terminal platform realizes the visual viewing, management and control of on-site information and big data prediction to ensure the healthy operation and maintenance of the project industry; the programming software provides a program writing platform for the logic control of the medium-sized intelligent controller, supports programming functions compliant with the IEC61131-3 standard, and can customize appropriate control programs for different industrial sites through the programming software.

[0026] The following will describe in detail the technical solutions provided by each embodiment of the present invention with reference to the accompanying drawings.

[0027] First, the present invention provides a device control system for industrial sites, such as Figure 1 shown Figure 1 is a schematic diagram of a device control system for industrial sites in the present invention. It includes: a controller module 1011 and an AI intelligent module 1012.

[0028] The controller module 1011 is used to collect relevant data of industrial site devices, and based on the relevant data, construct a digital twin model of industrial site devices through edge computing functions and digital twin technologies.

[0029] The AI intelligent module 1012 is used to predict the working state of the digital twin model through an intelligent prediction model, and perform feedback control on industrial site devices based on the prediction results; the intelligent prediction model adopts the Yolov5 network model, introduces CSPDarkNet as the backbone network in the Yolov5 network model, introduces a spatio-temporal attention module in the neck module of Yolov5, and introduces an adaptive attention mechanism in the head module to obtain a target neural network model; the target neural network is trained to obtain an intelligent prediction model. Among them, for the input part of the intelligent prediction model, each input channel uses a 1x1 convolutional kernel to combine the output of depth convolution.

[0030] Furthermore, in an embodiment, the device control system for industrial sites may further include a medium-sized intelligent controller 101 and a terminal platform 102; the medium-sized intelligent controller 101 is used to integrate the controller module 1011 and the AI intelligent module 1012; the terminal platform 102 is connected to the medium-sized intelligent controller 101. Specifically, as Figure 2 shown Figure 2 is another schematic diagram of a device control system for industrial sites in the present invention.

[0031] Among them, the medium-sized intelligent controller 101 includes a controller module 1011, an AI intelligent module 1012, a backplane module 1013, etc. Among them, the controller module 1011 and the AI intelligent module 1012 are connected by a bus cascading method of the backplane module 1013; the terminal platform 102, connected to the medium-sized intelligent controller 101, is used to display relevant data uploaded by the medium-sized intelligent controller 101 through digital twin technology, for visual viewing, display of big data prediction results, remote monitoring and prediction alarm. In addition, the terminal platform 102 is also used to perform full-life cycle health management on the engineering projects of on-site devices through digital twin technology, display the status and statistical analysis data of on-site device parameters uploaded by the medium-sized intelligent controller 101, complete the upper-level command issuance and program non-intrusive downloading to the controller module 1011, and perform visual viewing and operation on the engineering projects and on-site devices.

[0032] The controller module 1011 is used to collect relevant data of industrial field devices and process the issued superior instructions. Based on the relevant data, a digital twin model of the industrial field devices is constructed through edge computing functions and digital twin technology, and the digital twin model is controlled based on the processing results of the issued processing.

[0033] In one embodiment, the medium-sized intelligent controller 101 of the present invention serves as the core center of the control system, mainly facing medium and large industrial sites, and has characteristics such as good stability, reliability, and security.

[0034] Exemplarily, the controller module 1011 is designed using the Feiling embedded FETMX8MP-C core board. This series of processors focuses on machine learning and vision, advanced multimedia, and industrial automation with high reliability. It aims to meet the requirements of applications such as smart cities, industrial Internet, intelligent healthcare, and intelligent transportation. The medium-sized intelligent controller can provide accurate real-time data for the model construction of the digital twin system, establish efficient real-time interaction and function control between the platform interface and on-site applications, and promote the full life cycle management of engineering construction and intelligent manufacturing production lines.

[0035] In addition, the controller module 1011 adopts a blade-type structure design. Private backplane protocols are used for extended connection between modules in the medium-sized intelligent controller 101. The Ethernet / Programming Port protocol is used for program downloading between the controller module 1011 and the programming software that logically controls it. The ModBUS TCP / EtherNet / IP / EtherCAT protocol is used for data collection and instruction issuance between the controller module 1011 and field devices.

[0036] For the construction of the digital twin model, the controller module 1011 is used to perform digital production line modeling on the entities of transfer AGVs, inbound and outbound warehousing objects, raw materials and accessories, conveyor lines, transfer machines, accessories, and finished products in field devices; through coupling visualization models and 3D simulation technologies, real-time mapping and tracking are carried out on the transfer process, process flow connection process, component online process, and each sub-item process of automatic assembly, realizing real-time 3D visualization presentation of the actions and states of the physical production line.

[0037] In another embodiment, the AI intelligent module 1012 is used to predict the working state of the digital twin model through an intelligent prediction model and perform feedback control based on the prediction results.

[0038] Specifically, the AI intelligent module 1012 is used to collect displacement detection, weighing data, screw tightening force values, and angle feedback values of the production line through the sensing devices on the device side, and clean the relevant data; extract key process technologies by sorting out the process route of the physical production line, and build an analysis and prediction model for the digital twin model based on the key process technologies and the target neural network model to predict the working state of the digital production line.

[0039] Exemplarily, to meet the functions of the control system such as matching application, autonomous learning, and model construction in different scenarios, an AI intelligent module 1012 is added to the medium-sized intelligent controller, which customizes an adaptive AI algorithm to meet the logical analysis and calculation in simple scenarios. The AI intelligent module 1012 can be used to detect and identify industrial on-site target objects including personnel, flame / smoke, quality defects, belt tearing, idler abnormalities, etc., and assist in intelligent control.

[0040] Among them, the intelligent prediction model adopts the Yolov5 network model. In the Yolov5 network model, CSPDarkNet is introduced as the backbone network, convolutional kernels are applied to each input channel respectively, and 1x1 convolutional kernels are used to combine the outputs of depth convolutions; a spatio-temporal attention module is introduced into the neck module of Yolov5, and an adaptive attention mechanism is introduced into the head module to obtain the target neural network model; the target neural network is trained to obtain the intelligent prediction model. For the training of the target neural network, training samples can be obtained in advance. The training samples include relevant data samples of industrial on-site equipment and corresponding real states. The relevant data samples are input into the target neural network to obtain the predicted state of the industrial on-site equipment, and the target neural network is trained with the goal of minimizing the difference between the predicted state and the real state to obtain the intelligent prediction model.

[0041] In industrial on-site operations, in order to realize the real-time monitoring of operation scenario environment parameters and equipment working states through the control system, and achieve the purpose of safety control and effective prevention of harmful factors. The control system platform is designed by combining the edge computing function of the controller module 1011 with digital twin technology. The terminal platform 102 can display the relevant data uploaded by the intelligent controller, including the state acquisition and statistical analysis of on-site environmental equipment parameters, can complete the logical instruction issuance to the controller and the non-intrusive downloading of programs, can visually view and operate on the project site and equipment, realize the remote monitoring and prediction alarm functions of the project site, promote the healthy management of the entire life cycle of intelligent production lines and intelligent manufacturing, and realize the intelligent cloud control of the project site. Through the access of cloud platform data, the progress status of multiple projects can be integrated and displayed to implement the discrete management of industrial manufacturing.

[0042] The terminal platform 102 is connected to the medium-sized intelligent controller 101, and is used to display the relevant data uploaded by the medium-sized intelligent controller through the digital twin technology, realize the visual viewing of the relevant data, input of superior instructions, display of big data prediction results, and remote monitoring and prediction alarm functions.

[0043] The terminal platform 102 uses digital twin technology to manage the health of field equipment throughout the entire life cycle of engineering projects. It is used to display the status and statistical analysis data of field equipment parameters uploaded by the medium-sized intelligent controller 101, complete the issuance of logical instructions to the controller and the non-intrusive downloading of programs, and visualize and operate engineering projects and field equipment.

[0044] In addition, the medium-sized intelligent controller 101 also includes an I / O expansion module 1014 and a communication module 1015. The I / O expansion module 1014 includes a digital / analog input / output module, and the communication module 1015 includes RS485 and RS232 serial port communication modules and a communication module that can support multiple industrial bus protocols such as EtherNet / IP, EtherCAT or CANopen.

[0045] The communication module 1015 determines the system networking and communication capabilities of the product. The use of different industrial buses and industrial Ethernet protocols determines the product's ability to connect to other devices and systems in system applications, and to a certain extent determines the system's expansion and compatibility capabilities. Based on the current status and trends of industrial communication network development, the industrial buses and industrial Ethernet protocols adopted by mainstream manufacturers, the industrial buses and industrial Ethernet protocols supported by the planned products are: ModBUS TCP master / slave, ModBUS RTU master / slave, EtherCAT master station protocol, EtherNet / IP protocol, ProfiBUS-DP master / slave, and DeviceNet master stack protocol.

[0046] In the equipment control system of the present invention, the modules of the medium-sized intelligent controller 101 are connected with each other using a private backplane protocol (CANopen). The medium-sized intelligent controller 101 uses Ethernet / programming port protocol to download programs with programming software, and uses ModBUS TCP / EtherNet / IP / EtherCAT protocol to collect data and issue instructions with field equipment.

[0047] In order to facilitate real-time control, programming software needs to be deployed on the device control system. Therefore, the implementation of the device control system of the present invention mainly relies on the coordinated use of medium-sized intelligent controllers, terminal platforms and programming software. Figure 3As shown, the medium-sized intelligent controller, as the core device, establishes a connection between the terminal platform and field devices, completes the information transmission from the field information to the terminal platform and cloud platform, and meets the relevant operations of remote terminal visual information viewing and device control. The medium-sized intelligent controller is connected to field devices through communication interfaces such as RS485 / RS232 / CAN, and data is transmitted using protocols such as Modbus TCP / EtherNet-IP / EtherCAT; the medium-sized intelligent controller communicates with the terminal platform through methods such as 4G / 5G / Ethernet; the terminal platform can access the cloud server to meet the data cloud function and realize the integrated management of information between the terminal platform and the cloud platform.

[0048] The program download of the controller can be carried out in two ways: on-site and remote. The on-site download method mainly connects the computer to the controller through the Ethernet or programming port supported by the controller to complete the on-site program download; the remote download method relies on the cloud platform function to first upload the completed program in the computer to the cloud, and then through the Ethernet / 4G / 5G communication function, remotely and non-disruptively download the program to the controller connected to the network.

[0049] With the above system, first, the edge computing function is used to process the relevant data of the industrial site in real time, ensuring the timeliness and accuracy of the relevant data and improving the reliability of the digital twin model constructed based on the relevant data; using digital twin technology, the data of physical devices is transformed into virtual models, and the real-time monitoring and management of these models not only improve the transparency of device operation but also enhance the control ability of device status, which is beneficial to the management of potential harmful factors; the AI intelligent module constructs an intelligent prediction model, which uses an improved Yolov5 network model, not only improving the detection speed and accuracy of industrial site devices, being able to better handle target detection tasks in complex scenarios and being more suitable for real-time prediction tasks in the industrial site; feedback control is carried out based on the prediction results, and this control method significantly improves the management speed of industrial site devices. Secondly, potential harmful factors can be excluded from interfering with device operation through the prediction results, enhancing the control intensity of industrial site devices. In this way, through the combination of digital twin technology and intelligent prediction models, the medium-sized intelligent controller can optimize the production process, reduce resource waste, improve production efficiency, realize the efficient monitoring, intelligent prediction, and optimized control of field devices, and provide strong support for the industrial intelligent and digital transformation.

[0050] Based on the above device control system, the present invention also provides a specific implementation structure of a medium-sized intelligent controller, as Figure 4As shown in the figure, the medium-sized intelligent controller system adopts a modular design scheme and mainly consists of a power module, a backplane module, a CPU module (i.e., the controller module), an IO expansion module, a communication module, a 5G module, an AI intelligent module, etc. Its module combination is as shown in Figure 4 the figure. The system adopts an active backplane cascading method, and the power module, CPU module, communication module, IO expansion module, etc. are inserted into the backplane in sequence from left to right. Among them, the IO expansion module includes digital / analog input / output modules, and the communication module includes RS485 or RS232 serial communication modules developed in the next stage and communication modules that support industrial bus protocols such as EtherNet / IP, EtherCAT, and CANopen. The backplane communication uses a CAN FD interface and communicates using the CANopen protocol. Each backplane can support at least 10 expansion modules (the specific quantity is calculated according to the power consumption of the power module). The specific details of each module are as follows: 1) Power module: The power module selects the on-board power supply of the Jin Shengyang LO65-20B24MU model, with a power of 65W, supporting input voltages of 85-264VAC and 100-370VDC, an output voltage of 24VDC, and an output current of 2.71A. The power module mainly provides 5VDC and 24VDC voltages for the CPU module and expansion modules through the backplane to supply power to different modules.

[0051] 2) Backplane module: The backplane module design adopts a 40 / 50Pin Small Computer System Interface (SCSI) to achieve the transmission of power and multiple paths of data between modules. The backplane module adopts an active backplane cascading method and can be flexibly assembled according to the number of modules required by the project. Its circuits mainly include power supply, CAN-FD, RS485 signals, etc.

[0052] 3) Digital input module: The digital input module uses the GD32F303CCT6 of GigaDevice as the main control chip and realizes a 16-channel digital optocoupler input circuit through optocoupler input.

[0053] 4) Digital output module: The digital output module uses the GD32F303CCT6 of GigaDevice as the main control chip and realizes a 16-channel transistor output control circuit through the method of optocoupler isolation + PMOS tube.

[0054] 5) Analog input module: The analog input module uses the GD32F303CCT6 of GigaDevice as the main control chip and realizes a 4-channel analog input control circuit through a technical solution of isolated power supply + capacitive coupling isolation + ADC acquisition, supporting voltage signal types of 0-10V or current signals of 0-20mA / 4-20mA.

[0055] 6) Analog output module: The analog output module uses the GD32F303CCT6 of GigaDevice as the main control chip, and adopts a technical solution of isolated power supply + capacitive coupling isolation + DAC digital-to-analog conversion + operational amplifier to realize a 4-channel analog output control circuit, supporting 0-10V voltage or 0-20mA / 4-20mA current signal types.

[0056] 7) 5G expansion module: The 5G module is implemented by the design method of main control chip + communication module. The main control chip uses MT7621AT, and the communication module uses Quectel RM500Q, supporting wireless communication data reporting, firmware online upgrade, supporting local storage of collected data and log functions, and supporting Modbus TCP protocol communication with the CPU module.

[0057] 8) AI expansion module: The main control chip of the AI intelligent module uses RK3588, and the NPU with 6 TOPS computing power empowers various AI scenarios, providing possibilities for applications such as AI computing in complex scenarios and complex video stream analysis. The module supports an external power input interface, HDMI input / output interfaces, USB3.0 interfaces, and 2-way 10 / 100 / 1000Mbps adaptive Ethernet ports. Corresponding function commands are issued based on the detected target objects, and multiple AI algorithms such as face recognition, target detection, and construction site safety monitoring are built-in, supporting target object recognition and detection, and supporting Modbus TCP protocol communication with the CPU module.

[0058] 9) CPU module: The CPU module of this invention's controller is designed based on the FETMX8MP-C core board produced by Feilin Embedded. The hardware design block diagram of the controller CPU module is as Figure 5 shown. It includes a main control board and a lamp board. Among them, the main control board inputs 24V power from the backplane through a backplane connector. The main control board of the controller CPU module is designed using the FETMX8MP-C core board produced by Feilin Embedded. The memory units supported by the core board include 4GB of RAM, 16MB of NorFlash, and 16GB of eMMC. The main control board of the module also externally expands a 16MB SPI Flash through the ECSPI interface. In addition, the CPU module supports RTC function, power-down detection and power-down retention function, supports indicator light indication and screen display functions through the lamp board, and the screen supports button input function. In addition, the application circuit has been designed for wide temperature and EMC, and can be applied to industrial environments with relatively harsh ambient temperature and electromagnetic compatibility, further providing guarantee for the normal operation of the product.

[0059] Based on the above device control system, the present invention also provides an actual engineering project based on this device control system. The engineering project includes a digital twin part and an AI intelligent module part, specifically as follows: I. Digital twin part The research content of the application research on digital twin technology for the intelligent assembly production line of military-civilian integrated pyrotechnic products in a certain factory mainly includes: constructing a 3D model of the production line facilities and equipment, realizing real-time 3D visualization of the physical production line, integrating key data, managing and opening interfaces, and the technology of using the digital production line feedback to drive the operation and maintenance of the physical production line.

[0060] (I) Constructing a 3D model of the production line facilities and equipment The architecture of this production line is as Figure 6 shown, including the physical entity layer, the twin data layer, and the application function layer. According to the requirements analysis of the application research on digital twin technology for the production line, it is necessary to realize the construction of a 3D model of the production line facilities and equipment. Divided by function, the models are divided into three categories: materials, production line equipment, and finished products: 1. Materials Materials refer to three raw material accessories: nozzle, combustion chamber, and igniter.

[0061] 2. Production line equipment Production line equipment refers to equipment such as transfer AGV, conveyor line, transfer machine, and pallet.

[0062] 3. Finished products Finished products refer to the finished products after the materials are assembled.

[0063] The purpose of building the 3D model is to restore its physical characteristics such as shape, material, specification ratio, and color, as well as its interactive characteristics such as geometric structure and motion state, to form a digital twin that accurately maps the physical entity, and to provide a component basis for the animation presentation of the production process.

[0064] (II) Real-time 3D visualization of the physical production line Perform 3D visualization on the complete process of the finished product from materials to finished product on the production line, transforming the complex and difficult-to-directly-observe production process into a vivid and intuitive 3D visualization effect. Specifically achieve the following goals: 1. Data collection of the finished product production process Collect detailed data on equipment such as materials, production line equipment, and finished products, including the exact dimensions of the equipment, working principles, material flow paths, and morphological changes of the product at different stages.

[0065] 2. Construction of a virtual production line environment Use 3D modeling software to combine and list the 3D models to construct a highly realistic virtual production line environment.

[0066] 3. Material optimization Through the application of material mapping, light and shadow simulation, and physical engines, the scenes of each station and the three types of models in the animation present a sense of reality. The metallic luster, plastic texture, etc. are perfectly reproduced, approaching the real production site.

[0067] 4. Dynamic process construction Through the analysis of the production line process route, corresponding visual algorithms for different processes are constructed according to different process principles. By coupling the process visual algorithms and 3D simulation technology, based on the actual operation of the production line, dynamic processes such as the kit preparation process, process transfer connection process, impulse component online preparation process, and the start, operation, and stop of equipment, the input, transmission, and processing of materials, and the gradual shaping of products in the automatic assembly process are simulated. A mechanical simulation algorithm for key processes is constructed. Through the constructed process visualization algorithm, precise time control and smooth trajectory planning are achieved. Through the mechanical simulation algorithm of key processes, reasonable mechanical simulation of the digital production line is realized, and real-time 3D visualization of the physical production line is presented. By integrating interactive operations, by clicking on specific areas of the screen or dragging the slider, viewers can freely switch perspectives, zoom in on details, view the production status at different time periods, clearly see every detail, and understand the meaning of each operation.

[0068] (III) Key data integration, interface management, and opening Realize the integrated acquisition of data, including static data such as equipment parameters and accessory parameters, and use workstations to collect dynamic data such as displacement detection, weighing data, screw tightening force values, and angle feedback. Use data processing technology to clean the collected data, and finally integrate the cleaned data.

[0069] Realize system interface management, mainly starting from aspects such as standardized interface design, interface version management, interface security management, performance optimization, feedback and iteration, and overall architecture optimization, in order to achieve orderly management of static, dynamic, and other data required for digital twins. By formulating a unified interface document template, including interface descriptions, request / response formats, error code descriptions, etc., to develop standardized interface documents, and adopting widely recognized standards in the industry such as RESTful API and GraphQL to improve compatibility and maintainability.

[0070] (IV) Driving the operation and maintenance of the physical production line through feedback from the digital production line By sorting out the process route of the physical production line, analyzing the key process technologies in the process route, and developing a key process analysis and prediction model based on the working principles of key processes. Using the data acquisition and cleaning algorithms constructed by the system, data such as production line displacement detection, weighing data, screw tightening force values, and angle feedback are obtained. Based on the constructed process analysis and prediction model, the real-time operation status of the digital production line is analyzed and the future status of the production line is predicted, and it is presented on the digital production line, realizing real-time perception of the operation status of the physical production line, timely identifying the operation risks of the physical production line, so as to achieve the purpose of guiding operations such as production line production, maintenance, and maintenance.

[0071] II. AI intelligent module part The research on vision detection technology based on the improved YOLO framework algorithm is the core of the AI intelligent module. This project focuses on improving the performance of vision detection technology in detecting flames and smoke in complex environments, especially considering its deployment on intelligent vision controllers.

[0072] First, design a flame and smoke detection algorithm based on CSPDarkNet as the backbone network, that is, introduce a lightweight spatio-temporal attention module on the basis of Yolov5. Introduce a lightweight depthwise separable convolution module to reduce the computational amount and the number of parameters, ensuring the lightweight and efficiency of the model; at the same time, use an adaptive attention mechanism, especially the important spatial attention in flame and smoke detection, to highlight important features and suppress irrelevant features, thereby improving the detection accuracy and reliability. By integrating lightweight attention modules, efficiently extract and fuse multi-scale local features and global features of flames, improve the feature expression ability of the backbone network, enhance the model's understanding of high-dimensional features, and at the same time, the lightweight processing ensures its detection speed and improves the detection ability of early fires, especially accurately identifying subtle flame and smoke features.

[0073] In addition, construct a flame and smoke dataset with rich and real scenes, adopt an expansion strategy to balance the number of various situations in the dataset, formulate a data cleaning strategy to optimize the dataset, and enhance the images in the dataset based on the generative adversarial network to ensure data quality and diversity, so as to train and verify the proposed detection model, improve the generalization ability of the detection algorithm, and reduce problems such as false alarms and missed alarms. Conduct comprehensive tests on the self-made flame and smoke image dataset, compare and analyze the performance of the model, and continuously iterate and optimize the model structure and parameters according to the experimental feedback.

[0074] Secondly, the present invention also provides an intelligent control method for industrial sites, which is applied to an equipment control system for industrial sites, such as Figure 7 shown, including: S701. Collect equipment data from the industrial site through a medium-sized intelligent controller and transmit the data to the terminal platform through a communication module.

[0075] S702. Display the on-site data on the terminal platform for visual viewing, management and control, and big data prediction.

[0076] S703. Apply digital twin technology and adaptive AI algorithms to digitally manage the equipment, and realize the full life cycle management and control of engineering projects on the equipment.

[0077] Using the above method, first, the edge computing function is used to process the relevant data in the industrial field in real time, ensuring the timeliness and accuracy of the relevant data and improving the reliability of the digital twin model constructed based on the relevant data. By using digital twin technology, the data of physical devices is converted into virtual models. The real-time monitoring and management of these models not only improve the transparency of device operation but also enhance the control ability of device status, which is conducive to the management and control of potential harmful factors. The AI intelligent module constructs an intelligent prediction model by using an improved Yolov5 network model. This model not only improves the detection speed and accuracy of industrial field devices, can better handle target detection tasks in complex scenarios, but is also more suitable for real-time prediction tasks in the industrial field. Feedback control is performed based on the prediction results. This control method significantly improves the management speed of industrial field devices. Secondly, potential harmful factors interfering with device operation can be excluded through the prediction results, enhancing the control intensity of industrial field devices. In this way, through the combination of digital twin technology and intelligent prediction models, the medium-sized intelligent controller can optimize the production process, reduce resource waste, improve production efficiency, and achieve efficient monitoring, intelligent prediction, and optimized control of on-site devices, providing strong support for the industrial intelligent and digital transformation.

[0078] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0079] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to the embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram can be implemented by computer program instructions, and the combination of processes and / or blocks in the flowchart and / or block diagram can also be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate a device for realizing the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0080] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to work in a particular manner, such that the instructions stored in the computer-readable memory produce a manufacture including an instruction device that implements the functions specified in one or more of the procedures Figure 1 one or more procedures and / or blocks Figure 1 specified in the block or blocks.

[0081] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus, so that a series of operation steps are performed on the computer or other programmable apparatus to produce a computer-implemented process, and thus the instructions executed on the computer or other programmable apparatus provide steps for implementing the functions specified in one or more of the procedures Figure 1 one or more procedures and / or blocks Figure 1 specified in the block or blocks.

[0082] It should be noted that the specific embodiments described above can enable those skilled in the art to more comprehensively understand the present invention, but do not limit the present invention in any way. Therefore, although this specification has described the present invention in detail, those skilled in the art should understand that the present invention can still be modified or equivalently replaced; and all technical solutions and improvements that do not depart from the spirit and scope of the present invention are covered by the protection scope of the patent of the present invention. Any reference signs in the claims should not be construed as limiting the claimed claims.

Claims

1. An industrial site-oriented equipment control system, characterized in that: Including controller module and AI intelligent module; The controller module is used to collect relevant data of industrial field equipment, and build a digital twin model of the industrial field equipment through edge computing function and digital twin technology based on the relevant data; The AI ​​intelligent module is used to predict the working status of the digital twin model through an intelligent prediction model, and to perform feedback control on the industrial field equipment based on the prediction results; the intelligent prediction model adopts the Yolov5 network model, introduces CSPDarkNet as the backbone network in the Yolov5 network model, introduces a spatiotemporal attention module in the neck module of Yolov5, and introduces an adaptive attention mechanism in the head module to obtain a target neural network model; the target neural network is trained to obtain the intelligent prediction model.

2. The industrial field-oriented equipment control system according to claim 1, characterized in that: For the input part of the intelligent prediction model, each input channel uses a 1x1 convolution kernel to combine the output of the depth convolution.

3. The industrial field-oriented equipment control system according to claim 1, characterized in that: The controller module is used to perform digital production line modeling on the entities of the transfer AGV, in-and-out storage objects, raw material accessories, conveyor lines, transfer machines, accessories and finished products in the on-site equipment; By coupling visualization models and 3D simulation technology, the transfer process, process flow connection process, component online process, and automatic assembly sub-processes are mapped and tracked in real time, achieving real-time 3D visualization of the actions and status of the physical production line.

4. The industrial field-oriented equipment control system according to claim 1, characterized in that: The AI ​​intelligent module is used to collect displacement detection, weighing data, screw tightening force value and angle feedback value of the production line through the sensing device on the equipment side, and clean the relevant data; By combing through the process routes of the physical production line, key process technologies are extracted. Based on the key process technologies and the target neural network model, an analysis and prediction model for the digital twin model is constructed to predict the future state of the digital production line.

5. The industrial field-oriented equipment control system according to claim 1, characterized in that: The system also includes a medium-sized intelligent controller and a terminal platform. The medium-sized intelligent controller is used to integrate the controller module and the AI ​​intelligent module. The terminal platform is connected to the medium-sized intelligent controller and is used to display the relevant data uploaded by the medium-sized intelligent controller through digital twin technology, and perform visual viewing, big data prediction result display, remote monitoring and prediction alarm.

6. The industrial field-oriented equipment control system according to claim 5, characterized in that: The medium-sized intelligent controller also includes a backplane module, and the controller module and the AI ​​intelligent module are connected via a bus cascade of the backplane module.

7. The industrial field-oriented equipment control system according to claim 5, characterized in that: The medium-sized intelligent controller also includes an I / O expansion module and a communication module. The I / O expansion module includes a digital / analog input / output module. The communication module includes RS485 and RS232 serial port communication modules and a communication module that can support EtherNet / IP, EtherCAT or CANopen and other industrial bus protocols.

8. The industrial field-oriented equipment control system according to claim 5, characterized in that: The terminal platform is used to perform full life cycle health management of engineering projects of field equipment through digital twin technology, display the status and statistical analysis data of field equipment parameters uploaded by medium-sized intelligent controllers, complete the issuance of superior instructions to controller modules and non-intrusive downloading of programs, and perform visual viewing and operation of engineering projects and field equipment.

9. The industrial field-oriented equipment control system according to claim 5, characterized in that: The controller module adopts a blade-type structure design, and the modules in the medium-sized intelligent controller are extended and connected using a private backplane protocol. The controller module and the programming software that logically controls it use Ethernet / programming port protocol to download programs, and the controller module and field equipment use ModBUS TCP / EtherNet / IP / EtherCAT protocol for data collection and command issuance.

Images