WO2020037608A1 - 人工智能计算设备、控制方法及装置、工程师站及工业自动化系统 - Google Patents

人工智能计算设备、控制方法及装置、工程师站及工业自动化系统 Download PDF

Info

Publication number
WO2020037608A1
WO2020037608A1 PCT/CN2018/101973 CN2018101973W WO2020037608A1 WO 2020037608 A1 WO2020037608 A1 WO 2020037608A1 CN 2018101973 W CN2018101973 W CN 2018101973W WO 2020037608 A1 WO2020037608 A1 WO 2020037608A1
Authority
WO
WIPO (PCT)
Prior art keywords
computing
component
data
controller
computing device
Prior art date
Application number
PCT/CN2018/101973
Other languages
English (en)
French (fr)
Inventor
介鸣
许斌
冯尚科
Original Assignee
西门子股份公司
介鸣
徐云龙
许斌
冯尚科
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 西门子股份公司, 介鸣, 徐云龙, 许斌, 冯尚科 filed Critical 西门子股份公司
Priority to KR1020217008642A priority Critical patent/KR102566324B1/ko
Priority to JP2021510083A priority patent/JP2021536055A/ja
Priority to EP18930654.1A priority patent/EP3835898A4/en
Priority to CN201880095710.3A priority patent/CN112424713A/zh
Priority to PCT/CN2018/101973 priority patent/WO2020037608A1/zh
Priority to US17/270,098 priority patent/US20210181695A1/en
Publication of WO2020037608A1 publication Critical patent/WO2020037608A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G05CONTROLLING; REGULATING
    • G05BCONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B13/00Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
    • G05B13/02Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
    • G05B13/0265Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric the criterion being a learning criterion
    • GPHYSICS
    • G05CONTROLLING; REGULATING
    • G05BCONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B19/00Programme-control systems
    • G05B19/02Programme-control systems electric
    • G05B19/418Total factory control, i.e. centrally controlling a plurality of machines, e.g. direct or distributed numerical control [DNC], flexible manufacturing systems [FMS], integrated manufacturing systems [IMS], computer integrated manufacturing [CIM]
    • GPHYSICS
    • G05CONTROLLING; REGULATING
    • G05BCONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B19/00Programme-control systems
    • G05B19/02Programme-control systems electric
    • G05B19/418Total factory control, i.e. centrally controlling a plurality of machines, e.g. direct or distributed numerical control [DNC], flexible manufacturing systems [FMS], integrated manufacturing systems [IMS], computer integrated manufacturing [CIM]
    • G05B19/4185Total factory control, i.e. centrally controlling a plurality of machines, e.g. direct or distributed numerical control [DNC], flexible manufacturing systems [FMS], integrated manufacturing systems [IMS], computer integrated manufacturing [CIM] characterised by the network communication
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F15/00Digital computers in general; Data processing equipment in general
    • G06F15/16Combinations of two or more digital computers each having at least an arithmetic unit, a program unit and a register, e.g. for a simultaneous processing of several programs
    • G06F15/163Interprocessor communication
    • G06F15/173Interprocessor communication using an interconnection network, e.g. matrix, shuffle, pyramid, star, snowflake
    • G06F15/17306Intercommunication techniques
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F15/00Digital computers in general; Data processing equipment in general
    • G06F15/16Combinations of two or more digital computers each having at least an arithmetic unit, a program unit and a register, e.g. for a simultaneous processing of several programs
    • G06F15/163Interprocessor communication
    • G06F15/173Interprocessor communication using an interconnection network, e.g. matrix, shuffle, pyramid, star, snowflake
    • G06F15/1735Network adapters, e.g. SCI, Myrinet
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00Machine learning
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00Computing arrangements based on biological models
    • G06N3/02Neural networks
    • G06N3/08Learning methods
    • GPHYSICS
    • G05CONTROLLING; REGULATING
    • G05BCONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B2219/00Program-control systems
    • G05B2219/10Plc systems
    • G05B2219/15Plc structure of the system
    • G05B2219/15029I-O communicates with local bus at one end and with fieldbus at other end
    • GPHYSICS
    • G05CONTROLLING; REGULATING
    • G05BCONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B2219/00Program-control systems
    • G05B2219/10Plc systems
    • G05B2219/15Plc structure of the system
    • G05B2219/15119Backplane controller
    • GPHYSICS
    • G05CONTROLLING; REGULATING
    • G05BCONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B2219/00Program-control systems
    • G05B2219/20Pc systems
    • G05B2219/23Pc programming
    • G05B2219/23275Use of parser
    • GPHYSICS
    • G05CONTROLLING; REGULATING
    • G05BCONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B2219/00Program-control systems
    • G05B2219/20Pc systems
    • G05B2219/25Pc structure of the system
    • G05B2219/25145I-O communicates with local bus at one end and with fieldbus at other end
    • GPHYSICS
    • G05CONTROLLING; REGULATING
    • G05BCONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B2219/00Program-control systems
    • G05B2219/20Pc systems
    • G05B2219/25Pc structure of the system
    • G05B2219/25464MBO motherboard, backplane special layout
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L45/00Routing or path finding of packets in data switching networks
    • H04L45/02Topology update or discovery
    • H04L45/08Learning-based routing, e.g. using neural networks or artificial intelligence
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L9/00Cryptographic mechanisms or cryptographic arrangements for secret or secure communications; Network security protocols
    • H04L9/40Network security protocols
    • YGENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02PCLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
    • Y02P80/00Climate change mitigation technologies for sector-wide applications
    • Y02P80/10Efficient use of energy, e.g. using compressed air or pressurized fluid as energy carrier
    • YGENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02PCLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
    • Y02P90/00Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
    • Y02P90/02Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]

Definitions

  • the present application relates to the field of industrial automation, in particular to an artificial intelligence (AI) computing device, a control method and device, an Engineer Station (ES), and an industrial automation system.
  • AI artificial intelligence
  • ES Engineer Station
  • the industrial automatic control system uses the industrial control computer to summarize, analyze, and organize various parameters in industrial production collected by sensors to achieve information management and automatic control.
  • the realization of intelligent control in industrial automation systems has become a trend in the field of industrial automation.
  • the challenge is that traditional industrial controllers cannot provide sufficient computing power and there is no flexible solution to add artificial intelligence to the control system.
  • the embodiments of the present application propose an artificial intelligence computing device, a control method and device, an engineer station, and an industrial automation system for implementing closed-loop control with artificial intelligence in an industrial automation system to enhance control of the automation system Ability to improve the control efficiency of the automation system.
  • the AI computing device in the embodiment of the present application which is applied to an industrial automation system, may include: a backplane, a communication component, and a computing component;
  • the backplane includes a backplane bus and a fieldbus interface, the backplane bus connects the communication component and the computing component, and the fieldbus interface can be connected to a fieldbus of an industrial automation system through the fieldbus interface. Communication; wherein the industrial automation system includes at least one controller;
  • the communication component realizes data interaction between the controller and the computing component
  • the computing component receives data sent by the controller through the communication component, analyzes the data using an embedded AI computing architecture, and sends an analysis result to the controller through the communication component.
  • the AI computing device of each embodiment has a field bus interface, which can be directly connected to the field bus of an industrial automation system, provides a plug-and-play intelligent control function, and enhances the processing capacity of the control system.
  • the AI computing equipment is directly connected to the field bus, which can help the system achieve real-time intelligent closed-loop control and improve the control efficiency of the system.
  • the control method in the embodiment of the present application which is applied to a controller in an industrial automation system, may include:
  • AI artificial intelligence
  • control method of each embodiment analyzes the data by using the AI computing device connected to the fieldbus, and uses the analysis result to generate an automatic control instruction, so that the In the case of limitation, the processing capacity of the controller is enhanced, and real-time intelligent closed-loop control can be realized.
  • the control device in the embodiment of the present application which is applied to a controller in an industrial automation system, may include:
  • Production data acquisition unit for acquiring data in an industrial automation system
  • a task sending unit sending the data to a first computing component in an artificial intelligence (AI) computing device connected to the field bus;
  • AI artificial intelligence
  • a result collection unit configured to receive an analysis result sent by the AI computing device through the field bus, where the analysis result is obtained by analyzing the data by the first computing component; and providing the analysis result to a controller
  • a decision device in which the decision device generates control instructions for automated control.
  • control device of each embodiment analyzes the data by using the AI computing device connected to the field bus, and uses the analysis result to generate an automatic control instruction, so that the In the case of limitation, the processing capacity of the controller is enhanced, and real-time intelligent closed-loop control can be realized.
  • the engineer station of the embodiment of the present application connected to a field bus of an industrial automation system may include a processor and a memory; the memory includes machine-readable instructions, and the instructions may be executed by the processor to:
  • the device configuration information includes an identification of an artificial intelligence (AI) computing device and a computing component identification of at least one computing component of the AI computing device, and the AI computing device passes a fieldbus An interface is connected to the field bus; sending the device configuration information to a controller of the industrial automation system, so that the controller uses the device configuration information to communicate with the AI computing device;
  • AI artificial intelligence
  • Acquiring control logic corresponding to the AI computing device and receiving control configuration information for the AI computing device from the device configuration interface, where the control configuration information includes calculation parameters of a first computing component of the at least one computing component, Loading the control configuration information into the control logic, and loading the control logic to the controller, the control logic being configured to cause the controller to configure the first computing component,
  • the data in the industrial automation network is sent to the first computing component for analysis, and an analysis result fed back by the first computing component is obtained.
  • the engineer station of each embodiment can configure the controller through the field bus, so as to realize the communication between the controller and the AI computing device, thereby real-time intelligent closed-loop control in the industrial automation system.
  • the industrial automation system in the embodiment of the present application may include: an Engineer Station (ES), a controller, a production device, and an artificial intelligence (AI) computing device, and a field bus connecting each device;
  • ES Engineer Station
  • AI artificial intelligence
  • Providing a device configuration interface receiving configuration information from the device configuration interface, the configuration information including information of the AI computing device, and loading the configuration information to the controller;
  • Executing the control logic obtaining values of a plurality of production parameters in the industrial automation system, and sending the values of the plurality of production parameters to the AI computing device; receiving an analysis result sent by the AI computing device;
  • the AI computing device includes
  • the industrial automation system of each embodiment improves the control capability of the system by using AI computing equipment connected to the field bus, and at the same time, it can also realize real-time intelligent closed-loop control.
  • An embodiment of the present application further provides a computer-readable storage medium, in which a machine-readable instruction is stored, and the instruction can cause a processor to execute the control method of the embodiments.
  • FIG. 1 is a schematic diagram of an industrial automation system according to an embodiment of the present application.
  • FIG. 2 is a schematic diagram of an AI computing device according to an embodiment of the present application.
  • FIG. 3 is a logic diagram of data communication within the AI computing device according to an embodiment of the present application.
  • FIG. 4 is a logic diagram of internal data processing of the AI computing device in the embodiment of the present application.
  • FIG. 5 is a schematic diagram of a state machine of an AI computing architecture according to an embodiment of the present application.
  • FIG. 6 is a flowchart of a control method according to an embodiment of the present application.
  • FIG. 7 is a schematic diagram of a control device according to an embodiment of the present application.
  • FIG. 8 is a schematic diagram of an engineer station according to an embodiment of the present application.
  • FIG. 9 is a schematic diagram of a production process according to an embodiment of the present application.
  • FIG. 1 is a schematic diagram of an industrial automation system according to an embodiment of the present application.
  • the industrial automation system 10 is an industrial production system that uses automatic control and automatic adjustment devices to replace manual operation of machines and machine systems for processing and production.
  • the system 10 may include an AI computing device 20, an Engineering Station (ES) 30, a controller 40, and a field bus 60 connected to each device.
  • ES Engineering Station
  • the field bus 60 also known as the industrial data bus, is used to implement digital communications among field devices such as the controller 40, data acquisition equipment (not shown), and actuators (not shown) in the industrial production field, and these field devices And advanced control systems (such as the engineering station ES 30).
  • the field bus 60 may be implemented by using a certain field bus technology, such as ProfiBus, InterBus, a controller area network (CAN) bus, an addressable remote sensor high-speed channel (HART) bus, and the like.
  • the controller 40 may be one or more industrial controllers, such as a programmable logic (PLC) controller, a PC bus industrial computer (IPC) controller, a decentralized control system (DCS) controller, a fieldbus control system (FCS) Controller, CNC system controller, etc.
  • the controller 40 can communicate with various data acquisition equipment (such as intelligent instruments, sensors, etc.) and actuators (e.g., current regulating valves, voltage regulating valves, feeding valves, etc.) at the production site.
  • actuators e.g., current regulating valves, voltage regulating valves, feeding valves, etc.
  • Production equipment (not shown) refers to a collection of one or more equipment used for industrial processing and manufacturing, such as machine tools, lathes, assembly line equipment, and the like.
  • the controller 40 can also communicate with the production equipment through the field bus 60, collect process information of the production equipment, and monitor and control the production process of the production equipment.
  • the controller 40 can acquire various parameter values (hereinafter also referred to as production parameter values and production process data) of the production equipment, and exert control on the operation of the production equipment according to these production parameter values.
  • the production parameters can be any parameters related to the production process of the production equipment, such as the voltage, current, motor speed of the production equipment, the feed rate of the raw materials, and so on.
  • the values of the production parameters can be obtained from data acquisition equipment or from program-controlled production equipment.
  • the controller 40 may also adjust the value of the production parameter by sending a control instruction to an actuator.
  • the actuator can receive the control signal of the controller 40 and perform corresponding adjustment actions to change the value of the production parameter.
  • the AI computing device 20 is a plug-and-play device with AI computing capabilities and can be connected to the field bus through a field bus interface.
  • the AI computing device 20 may receive the values of the plurality of production parameters sent by the control device 41 in the controller 40 through the field bus 60, analyze the values of the plurality of production parameters to obtain an analysis result, and send the analysis result to the control through the field bus 60
  • the controller 40 causes the decision-making device 42 in the controller 40 to generate a control instruction for automatic control based on the analysis result.
  • the system 10 may be connected to a plurality of AI computing devices 20.
  • the plurality of AI computing devices 20 may include devices with different AI functions, or may include several devices that are identical to provide redundant backup or load sharing.
  • the controller 40 may be configured for the AI computing device 20 at the engineering station ES30.
  • the engineering station ES30 refers to industrial process monitoring and management equipment for industrial process control engineers.
  • the controller 40 in the industrial control system can be configured so that the controller 40 can communicate with other equipment on the production site, perform data processing, Make control decisions to monitor and control the production process.
  • the engineering station ES30 can provide a device configuration interface, receive configuration information for the AI computing device 20 from the device configuration interface, and load the configuration information to the controller 40.
  • the controller 40 communicates with the AI computing device 20 using the configuration information.
  • the device configuration interface of the ES 30 may include a human-machine interaction interface.
  • the ES30 can display the configuration and operation status of each device in the system 10 through a human-machine interactive interface, and display the configuration interface through a human-machine interactive interface.
  • the ES30 can complete various device configuration operations, such as adding devices, deleting devices, adding device configuration information, modifying device configuration information, deleting device configuration information, and so on.
  • the configuration information received from the device configuration interface may include text information entered by the operator into the ES30 (such as the ID, address, etc. of the AI computing device), and configuration files obtained from the operator through the device configuration interface or selected paths , Control logic, etc.
  • the path entered or selected by the operator can be the storage path in the ES30's built-in storage device, the path in the external extended storage device of the ES30 device, or the location on the network (such as URL, etc.).
  • the configuration information may include an identification of the AI computing device 20 (eg, a media access control (MAC) address, a device name, etc.), so that the controller 40 may communicate with the AI computing device 20.
  • the configuration information may further include information about a plurality of production parameters required by the AI computing device 20 to execute the AI calculation process, so that the controller 40 may provide the values of these production parameters collected by the sensor to the AI computing device 20 .
  • the ES 30 may also obtain the control logic 31 corresponding to the AI computing device 20, and load the control logic 31 to the controller 40.
  • the ES 30 may obtain the control logic 31 from an operator through a device configuration interface input or a selected path.
  • the path entered or selected by the operator can be a storage path in the ES30's built-in storage device, a path in the ES30's external extended storage device, or a location on the network (such as a URL, etc.).
  • the ES 30 can also configure and edit the functions of the control logic 31 through the interface provided by the control logic 31.
  • the controller 40 controls the AI computing device 20 by executing the control logic 31.
  • the above-mentioned information of the plurality of production parameters can also be loaded into the control logic 31 through the interface provided by the control logic 31.
  • the controller 40 executes the control logic 31 to obtain the values of these production parameters and provides them to the AI computing device 20.
  • the AI computing device 20 may include a communication component 22 and one or more computing components 23.
  • the AI computing device 20 may include a plurality of computing components 23 having different AI functions, and may also have a plurality of computing components having the same functions.
  • the computing components 23 having different AI functions may correspond to different control logics 31, and the computing components 23 having different AI functions may be controlled by the same control logic 31.
  • the industrial automation system of each embodiment adopts an AI computing device having a field bus interface, so that plug-and-play can be realized without replacing existing equipment and without being limited by the processing capabilities of the existing equipment.
  • Intelligent control function enhances the processing capacity of the control system.
  • the AI computing equipment is directly connected to the field bus, and the production process data is directly analyzed and processed in the field, which can realize intelligent closed-loop control and improve control efficiency.
  • the AI computing device 20 has a physical interface connectable to a field bus and supports communication using an industrial communication protocol used by an industrial automation system.
  • the following uses an implementation manner of an AI computing device 20 as an example for description.
  • FIG. 2 is a schematic diagram of an AI computing device according to an embodiment of the present application. As shown in FIG. 2, the AI computing device 20 may include a backplane 21, a communication component 22, and a computing component 23.
  • the backplane 21 includes a backplane bus 211 and a fieldbus interface 212.
  • the backplane bus 211 is used to connect the communication unit 22 and the computing unit 23.
  • the fieldbus interface 212 can communicate with the fieldbus 60 through the fieldbus interface 212.
  • the fieldbus interface 212 is an interface that conforms to the fieldbus technology adopted by the fieldbus 60, such as ProfiBus, InterBus, CAN, HART, and other interfaces.
  • the backplane bus 211 may use any bus technology, such as a bus technology designed according to requirements, or an existing data transmission bus.
  • the backplane bus interface provided by the backplane bus 211 may be a Low-Voltage Differential Signaling interface, an S422 interface, an RS485 interface, and the like.
  • the communication component 22 can implement data interaction between the controller 40 and the computing component 23.
  • the computing component 23 includes an AI computing architecture.
  • the AI computing architecture may include AI computing logic, such as machine learning logic and neural network algorithms.
  • the AI computing architecture may further include dedicated hardware (such as GPU, FPGA, ASIC, neural network processor, etc.) with high computing performance customized for AI computing.
  • the computing component 23 can receive the data sent by the controller 40 through the communication component 22, analyze the data using the embedded AI computing architecture, and send the analysis result to the controller 40 through the communication component 22.
  • AI computing device 20 of each embodiment has a field bus interface, plug-and-play intelligent control functions can be implemented in the industrial automation system, and the control capability of the industrial automation system is enhanced.
  • AI computing equipment analyzes and processes the production process data at the production site, which can realize intelligent closed-loop control and improve control efficiency.
  • the communication component 22 may communicate with the field bus 60 using a message format defined by an industrial communication protocol (such as PROFINET, EtherCat, etc.).
  • the communication unit 22 may use a message format defined by the industrial communication protocol to parse a message sent by the controller 40, and send the parsed message content to the computing unit 23.
  • the communication unit 22 may also use the message format to encapsulate the feedback data sent by the computing unit 23, and send the encapsulated message to the controller 40.
  • the AI computing device 20 can be directly used in the existing industrial communication network without modifying the network, which is convenient to use.
  • the communication component 22 may be implemented by an FPGA, an ASIC, an integrated circuit, an industrial communication chip, or the like.
  • the AI computing device 20 may include one or a plurality of computing components 23.
  • the communication component 22 may forward data between the computing component 23 and the controller 40.
  • the communication component 22 may use the computing component identification of each computing component 23 to implement communication between each computing component 23 and the controller 40.
  • the communication component 22 may receive, through the fieldbus interface 212, a message conforming to the industrial communication protocol sent by the controller 40 (in order to distinguish it from the message sent by the AI computing device 20, it is hereinafter referred to as the first message ).
  • the first message includes a header and a payload.
  • the payload includes an identifier of one or more computing components 23 (hereinafter referred to as a computing component identification) and message data or message content corresponding to each computing component identification.
  • the computing component identifier of the computing component 23 is used to distinguish each computing component 23 in the AI computing device 20, and may be a name, a serial number, etc., may be a device identification code configured in the computing component 23 during production, or may be a controller 40 other identification assigned.
  • the communication component 22 may parse the identifier of the computing component 23 and the message data corresponding to the identifier from the first message according to the message format defined by the industrial communication protocol, and send the message data to the The identification corresponds to the computing component 23.
  • the communication unit 22 may also receive feedback data from the first calculation unit 23 through the backplane bus 211 (the first calculation unit 23 is a calculation unit of the plurality of calculation units 23), and the feedback data and the calculation unit of the first calculation unit 23
  • the identification is used as message data to generate a message with a message format defined by the industrial communication protocol (to distinguish it from the message sent by the controller 40, which is referred to herein as the second message), and pass the second message through the scene
  • the bus interface 212 is transmitted to the controller 40.
  • the feedback data is the processing result of the message data by the computing component 23, and is used to enable the controller 40 to grasp the status of the computing component 23, obtain the analysis results, and the like, thereby generating AI computing equipment 20 or other equipment (such as production equipment, Actuator, etc.).
  • the feedback data can be a confirmation message that the configuration is complete; when the message data is a training instruction, the feedback data can be training result information; when the message data is production process data, the feedback Data can be the result of analysis of production process data.
  • the controller 40 can communicate with a plurality of computing components 23 in the AI computing device 20, and a single AI computing device 20 can include a plurality of computing components 23, which greatly improves The computing capacity and computing power of a single AI computing device 20.
  • the data interacted between the computing component 23 and the controller 40 may be divided into a plurality of types.
  • the data may be divided into a plurality of types according to different priorities and different transmission time requirements.
  • the communication component 22 may process the received data according to a preset processing strategy corresponding to the type of the received data.
  • the preset processing strategy may include a strategy for a data processing sequence, a strategy for a data transmission sequence, a strategy for a data transmission mode, and the like.
  • different types of data for processing can be preset for different types of data.
  • the communication component 22 may determine the type of data according to the data channel identifier corresponding to the data in the first message sent by the controller 40, and process the data according to the processing strategy corresponding to the type; the backplane bus 211 is used to obtain the first computing component 23 to provide The type of the feedback data is transmitted to the controller 40 through the fieldbus interface 212 through a field bus interface 212 through a transmission channel having a preset correspondence relationship with the type of the feedback data.
  • FIG. 3 is a schematic diagram of internal data communication between the computing component 23 and the communication component 22 in the AI computing device 20 in the embodiment of the present application.
  • one or more computing components 23 in the AI computing device 20 may send feedback data and types to the communication component 22.
  • the communication component 22 maps the feedback data to different data channels according to the type of the feedback data, and transmits the feedback data to the field bus 60 through the field bus interface 212.
  • the types of feedback data may include periodic data 221, aperiodic data 222, diagnostic service data 223, and the like.
  • FIG. 4 is a schematic diagram of data communication between the computing component 23 in the AI computing device 20 and the controller 40 using different logical data channels in the embodiment of the present application.
  • a logical channel (hereinafter also referred to as an IO channel 27) between the AI computing device 20 and the controller 40 is divided into a first transmission channel 271 and a second transmission channel 272.
  • the first transmission channel 271 is an aperiodic transmission channel, that is, its transmission timing is not periodic, and it is transmitted only when needed.
  • the second transmission channel 272 is a periodic transmission channel and transmits at a fixed time interval.
  • the first transmission channel 271 and the second transmission channel 272 can be distinguished by a channel identifier.
  • the computing component 23 may perform record data interaction with the controller 40 through the first transmission channel 271.
  • the recorded data may be data of low importance or low real-time communication requirements, such as configuration information (such as configuration information of the structure of the AI computing architecture, configuration information of input parameters and output parameters of the AI computing architecture 231, etc.), Training data, etc.
  • the computing component 23 may store the configuration information in the built-in configuration data storage module 232 and the training data in the training data storage module 233.
  • the computing component 23 may send the configuration information in the built-in configuration data storage module 232 to the controller 40 through the first transmission channel 271, and the controller 40 may adjust the computing architecture of other modules according to these configuration information.
  • the computing component 23 may also perform control data interaction with the controller 40 through the second transmission channel 272.
  • the control data may include a state switching instruction to the AI computing architecture 231 sent by the controller 40, production process data to be analyzed sent by the controller 40, a status report of the AI computing architecture 231 sent by the computing component 23, and the like.
  • the computing component 23 may include a state switching module 235 for changing the running state of the AI computing architecture 231 according to the state switching instruction sent by the controller 40.
  • the state switching module 235 may use the second transmission channel 272 to send the current operating state of the AI computing architecture 231 to the controller 40 through the communication component 22, and receive the state switching instruction sent by the controller 40 through the communication component 22 to send the AI
  • the computing architecture 231 switches from a first operating state to a second operating state.
  • the state switching module 235 may send the status of the AI computing architecture 231 to the controller 40 through the second transmission channel 272 by inputting the status word (ISW), and switch the AI computing architecture according to the output control word (OCW) sent by the controller 40. 231 status.
  • ISW status word
  • OCW output control word
  • the state switching module 235 may also input data corresponding to the current operating state provided by the controller (40) into the AI computing framework (231) according to the current operating state of the AI computing framework (231). For example, when the AI computing architecture 231 is in a training state, the state switching module 235 may input the training data sent by the controller 40 through the first transmission channel 271 and stored in the training data storage module 233 into the AI computing architecture 231 for training AI Computing Architecture 231. When the AI computing architecture 231 is in an operating state, the state switching module 235 may input the data to be analyzed periodically sent by the controller 40 through the second transmission channel 272 into the AI computing architecture 231, so that the AI computing architecture 231 outputs an analysis result. The analysis result generated by the AI computing architecture 231 may also be fed back to the controller 40 through the second transmission channel 272.
  • the system 10 may be provided with one or a plurality of AI computing devices 20, each AI computing device 20 may include one or a plurality of computing components 23, and different computing components 23 may have AI computing architectures 231 with different functions.
  • Functions that the AI computing architecture 231 can implement may include, but are not limited to, online parameter optimization, process monitoring, fault diagnosis, and the like.
  • the AI calculation architecture 231 may execute a preset optimization calculation process to optimize a plurality of production parameters in the production process data, and output recommendations for the plurality of production parameters. value.
  • the calculation module 23 may also convert the recommended value into a production adjustment recommendation that can be identified by the controller 40 as an analysis result, and the production adjustment recommendation includes the recommended value of at least one of the plurality of production parameters.
  • processing strategies for suggested values can be preset. For example, a threshold may be preset. When the difference between the recommended value of the parameter and the actual value is not greater than the threshold, the adjustment result of the parameter may not be included in the analysis result. This prevents frequent and unnecessary changes in production parameters.
  • the AI computing architecture 231 may execute a preset parameter detection process to detect values of a plurality of production parameters in the production process data, and output a parameter detection result.
  • the parameter test results indicate whether the values of the plurality of production parameters are normal.
  • the calculation module 23 may also convert the parameter detection result into a status monitoring report recognizable by the controller 40 as the analysis result, and the status monitoring report includes information for indicating whether the status of the production equipment is normal.
  • the AI computing architecture 231 may execute a preset fault diagnosis process to use a plurality of production parameter values in the production process data to perform fault diagnosis, and output a fault diagnosis result and fault diagnosis.
  • the results include information for one component (eg, collector, actuator, production equipment, etc.) in the system 10.
  • the calculation module 23 may also convert the fault diagnosis result into a fault report that can be identified by the controller 40 as an analysis result, and the fault report includes information of the component.
  • the control capability of the industrial automation system can be enhanced.
  • the computing component 23 may further include a configuration unit (not shown) for configuring the AI computing architecture 231.
  • the configuration unit may receive the configuration parameters sent by the controller 40 through the communication component 22, and the configuration parameters include a plurality of attribute values of the AI computing architecture 231; and set the values of the plurality of attributes corresponding to the AI computing architecture 231 to the plurality of attributes in the configuration parameters value.
  • the plurality of attribute values may include structural attribute values
  • the configuration unit may use the structural attribute values in the configuration parameters to set components and connection methods of the AI computing architecture 231, such as the hierarchical structure of the neural network in the AI computing architecture 231. Different AI functions and different production processes have different computing requirements.
  • the required structural information of the AI computing architecture can be configured to the controller 40 through the ES 30, and the controller 40 configures the corresponding computing component 23.
  • the plurality of attribute values may include a first parameter and a second parameter, and the first parameter is one or a plurality of production parameters of the production equipment.
  • the configuration unit may set the first parameter as an input parameter of the AI computing architecture 231; and set the second parameter as an output parameter of the AI computing architecture 231. Different production processes have different production parameters.
  • the information of the production parameters that need to be used can be configured to the controller 40 through the ES 30, and the controller 40 configures the corresponding computing component 23.
  • the configuration unit may also save the configuration parameters in the configuration data storage module 232.
  • the state switching module 235 may also use a state machine mechanism to manage the running state of the AI computing architecture 231.
  • FIG. 5 is a schematic diagram of a state machine of an AI computing architecture according to an embodiment of the present application.
  • the AI computing architecture 231 can have seven states, namely initialization complete (S1), preparation training (S2), training (S3), training error (S4), preparation operation (S5), and operation (S6) ), And operation error (S7).
  • Operation means that the AI computing architecture 231 uses the trained model to analyze the production process data.
  • the controller 40 can use six control words to control the state transition of the AI computing architecture 231. The six control words are implemented with six signals.
  • 6 types of control words can be represented by information having 6 bits, and each bit represents an instruction.
  • bit1 means preparing for training (migrating from S1 to S2)
  • bit2 means initiating operation (migrating from S1 to S5)
  • bit3 means starting training (migrating from S2 to S3)
  • bit4 means starting operation (migrating from S5 to S6)
  • Bit5 indicates the end of training (migrating from S2 / S3 to S1)
  • bit6 indicates the end of operation (migrating from S6 / S5 to S1). Setting the bit corresponding to an instruction to a bit indicates that the AI computing architecture 231 is required to perform a state transition operation corresponding to the instruction.
  • the AI computing architecture 231 When the AI computing architecture 231 enters an error state of S4 or S7, after reporting the state S4 or S7 to the controller 40, it can directly enter S1 without waiting for an instruction from the controller 40.
  • the operation control of the AI computing architecture 231 is more standardized and easy to manage.
  • the controller 40 may also use the state machine of the computing unit 23 to synchronize the parallel tasks of the plurality of computing units 23. For example, the controller 40 assigns a calculation task to a plurality of calculation units 23, and each calculation unit 23 sets the status bit to S1 after the calculation is completed. Then the controller 40 only needs to wait for these calculation units 23 to return to the state S1, and then determine These calculation units 23 are currently synchronized and can begin the operation of the next cycle.
  • the computing component 23 may further include an energy-saving unit (not shown) for changing the working mode of the computing component 23 according to an instruction of the controller 40.
  • the energy saving unit may receive the first instruction sent by the controller 40 through the communication component 22 and enter the low power consumption mode according to the first instruction; and receive the second instruction sent by the controller 40 through the communication component 22 and exit the low power consumption mode according to the second instruction .
  • the controller 40 will send a PROFIenergy command to sleep or wake up to all devices when the automation system enters or exits the low-power mode.
  • the computing unit 23 may enter or exit the low power consumption mode according to a PROFIenergy command sent by the controller 40. This can reduce the energy consumption of the AI computing device 20 when the automation system is running in a low power consumption mode.
  • the backplane 21 may include a plurality of slots, and the computing component 23 is connected to the backplane bus 211 as a pluggable expansion card through the slots.
  • the computing component 23 is connected to the backplane bus 211 as a pluggable expansion card through the slots.
  • the computing component 23 needs to be replaced (for example, the existing calculation component 23 for parameter optimization is replaced with the calculation component 23 for fault diagnosis), the old calculation component 23 can be removed from the slot and a new calculation can be inserted.
  • This pluggable design can conveniently expand the computing power of the AI computing device 20 or change the AI functions of the AI computing device 20.
  • FIG. 6 is a flowchart of a control method according to an embodiment of the present application. This method can be executed by the control device 41 in the controller 40. The method may include the following steps.
  • Step S61 Acquire data in the industrial automation system.
  • step S62 the data is sent to the AI computing device connected to the field bus 60 through the field bus 60 of the industrial automation system.
  • Step S63 Receive the analysis result obtained by analyzing the data by the AI computing device 20 through the field bus 60, and provide the analysis result to the decision device 42 in the controller 40 for the decision device to generate a control instruction for the production equipment.
  • intelligent closed-loop control can be implemented in an industrial automation system to improve control efficiency.
  • the control device 41 may determine the plurality of input parameters in the preset first configuration information as the plurality of production parameters from data in the industrial automation system.
  • the data provided by the acquisition equipment is used to obtain values of a plurality of production parameters as production process data.
  • the control device 41 may obtain values of a plurality of production parameters from data provided by a data acquisition device in a previous time period as a production process data every preset period of time.
  • the control device 41 may obtain the information of the first computing component 23 in the AI computing device 20 from the preset second configuration information; sending the production process data To the first computing unit 23 in the first AI computing device 20.
  • the adaptive adjustment of the control device 41 can be realized only by updating the configuration information, which is convenient and flexible.
  • the control device 41 needs to acquire information of these computing components so as to achieve communication with each computing component.
  • the control device 41 may obtain the second configuration information, in which information of a plurality of computing components is recorded.
  • the control device 41 may obtain the information of the plurality of computing components 23 from the second configuration information, and use the information to allocate computing tasks to the computing modules 23.
  • the control device 41 may obtain load information of a plurality of computing components 23 from at least one AI computing device 20; according to the load information, a computing component 23 is selected from the plurality of computing components 23 as the first computing component 23, and the first The AI computing device 20 to which a computing component 23 belongs is determined as the AI computing device 20.
  • the computing component 23 may upload its load information (such as computing load, processor temperature, computing bandwidth (number of operations completed per second), etc.) to the controller 40 through the identification and maintenance (I & M) information, and the controller 40
  • a computing task may be assigned to the computing component 23 according to a load balancing algorithm (for example, a boxing algorithm, etc.), and production process data corresponding to the computing task is sent.
  • a load balancing algorithm for example, a boxing algorithm, etc.
  • the computing task assignment method depends on the AI algorithm used, for example, the plurality of computing components 23 can process different parameters, or process data of different periods, or process the same data set in parallel, and so on.
  • control device 41 may use the periodic data to input the normalized production process data to the calculation component 23 according to a fixed period, and obtain the normalized output data of the calculation component 23.
  • the computing components 23 of different structures may have different numbers of input data and output data.
  • the configuration information of the control device 41 may indicate the message length (such as 16 bytes, 64 bytes, etc.) adopted by each computing unit 23.
  • the configuration information may be provided to the control device 41 by the ES 30 through, for example, a configuration file, a device description file, and the like.
  • the configuration file can be a GSDML file, which can include manufacturer information, communication port information, module and submodule information, and alarm diagnostic information.
  • different computing components may have AI computing architectures 231 with different functions, the output data of the AI computing architectures 231 with different functions are also different, and the analysis results provided to the decision device are also different.
  • the AI computing architecture 231 when the AI computing architecture 231 includes an optimization computing process for online optimization, the AI computing architecture 231 optimizes a plurality of production parameters.
  • the control device 41 can obtain the suggested values of a plurality of production parameters output by the AI calculation framework 231, and convert the suggested values into production adjustment recommendations that can be identified by the decision device 42 as the analysis results provided to the decision device 42. How many plural production adjustment recommendations include Recommended value for at least one of the production parameters.
  • the AI computing architecture 231 detects values of a plurality of production parameters.
  • the control device 41 can obtain the parameter detection result output by the AI computing architecture 231, and the parameter detection result indicates whether the values of the plurality of production parameters are normal; the parameter detection result is converted into a status monitoring report recognizable by the decision device 42 as provided to the decision device 42 Analyzing the results, the status monitoring report includes information used to indicate whether the status of the production equipment is normal.
  • the AI computing architecture 231 uses the values of a plurality of production parameters for fault diagnosis.
  • the control device 41 can obtain the fault diagnosis result output by the AI computing architecture 231, and the fault diagnosis result includes information of a component of the production equipment; the fault diagnosis result is converted into a fault report identifiable by the decision device 42 as the analysis result provided to the decision device 42
  • the fault report includes information about the component.
  • the conversion of the AI calculation result to the analysis result by the control device 41 can simplify the implementation of the calculation unit 23.
  • the control process of the control device 41 can be adjusted through configuration and programming, it is easier to adjust the output data of the AI computing architecture at the control device 41.
  • the control device 41 may obtain the configuration parameters of the first computing component 23 from the preset third configuration information, and the configuration parameters include the AI calculation used by the first computing component 23 Multiple attribute values of the architecture; sending configuration parameters to the first computing component 23, so that the first computing component 23 sets the values of the multiple attributes of the AI computing architecture to the multiple attribute values.
  • the plurality of attribute values of the AI computing architecture may include structural attribute values, or input parameters, output parameters, and the like.
  • the training data of the AI computing architecture may also be obtained from the configuration information.
  • the control device 41 may obtain training data from preset fourth configuration information, and send the training data to the first computing component 23, so that the first computing component 23 uses the training data to calculate the AI computing architecture in the first computing component 23. Training.
  • FIG. 7 is a schematic diagram of a control device according to an embodiment of the present application.
  • the control device 41 may include: a production data acquisition unit 411, a task transmission unit 412, and a result collection unit 413.
  • the production data acquisition unit 411 may acquire production process data of production equipment in the industrial automation system, and the production process data includes values of a plurality of production parameters of the production equipment.
  • the task sending unit 412 may send the production process data to the first computing component 23 in the AI computing device connected to the field bus 60 through the field bus 60 of the industrial automation system.
  • the result collection unit 413 may receive the analysis result sent by the AI computing device 20 through the field bus 60, and the analysis result is obtained by analyzing the production process of the production equipment by using the production process data by the first calculation component 23; and providing the analysis result to the controller 40 Decision-making device for the decision-making device to generate control instructions for the production equipment.
  • control device may further include a configuration unit 414.
  • the configuration unit 414 obtains configuration parameters of the first computing component 23 from the preset third configuration information, and the configuration parameters include a plurality of attribute values of the AI computing architecture of the first computing component 23; and sends the configuration parameters to the first computing component 23 , Thereby setting the values of the plurality of attributes of the AI computing architecture to the plurality of attribute values.
  • control device may further include a training unit 415.
  • the training unit 415 may obtain training data from the preset fourth configuration information, and send the training data to the first computing component 23, so that the first computing component 23 uses the training data to train the AI computing architecture.
  • the control method in the embodiment of the present application may also be implemented by software code.
  • the software code may be stored in a computer-readable storage medium.
  • the software code can be machine-readable instructions that conform to industrial control programming language standards (such as IEC61131-3).
  • the software code may be read by the ES 30 from a remote storage device, a local storage device, or a removable storage device (such as an optical disk, a flash memory, etc.), configured, and loaded into the controller 40. 40 is executed to implement the above control method.
  • the engineering station ES 30 may provide the configuration information and control logic required to control the AI computing device 20 to the controller 40, so that the controller 40 has the control capability of the AI computing device 20.
  • FIG. 8 is a schematic diagram of an engineer station according to an embodiment of the present application. As shown in FIG. 5, the ES 30 may include a processor 32, a memory 33, and a communication device 34. The communication device 34 is used for the ES 30 to communicate with other devices in the network.
  • the memory 33 may include a management module 37.
  • the management module 37 includes an interface module 371, a device configuration module 372, and a control configuration module 373.
  • the interface module 371 may provide a device configuration interface.
  • the device configuration module 372 may receive device configuration information from the device configuration interface.
  • the device configuration information includes an identification of the AI computing device and identification information of at least one computing component 23 in the AI computing device 20; and sends the device configuration information to the controller of the industrial automation system. 40, so that the controller 40 uses the device configuration information to communicate with the AI computing device 20.
  • the control configuration module 373 can obtain the control logic 31 corresponding to the AI computing device 20, and receive control configuration information for the AI computing device 20 from the device configuration interface.
  • the control configuration information includes the calculation parameters of the first calculation component 23 of at least one calculation component 23.
  • the control configuration information is loaded into the control logic 31, and the control logic 31 is loaded into the controller 40.
  • the control logic 31 can cause the controller 40 to configure the first computing component 23, send the production process data of the production equipment in the industrial automation network to the first computing component 23 for analysis, and obtain the analysis result fed back by the first computing component 23 .
  • the interface module 371, the device configuration module 372, and the control configuration module 373 may be implemented by machine-readable instructions.
  • the memory 33 may further include an operating system 35 and a network communication component 36.
  • the device configuration module 372 may receive channel information of the first computing component 23 from the device configuration interface, and the channel information includes identifiers of one or more channels; the channel information is sent to the controller 40, and the controller 40 causes the controller 40 to send data The channel corresponding to the type of data is sent to the first computing component 23.
  • control configuration module 373 may receive the architecture information of the first computing component 23 from the device configuration interface.
  • the architecture information includes components and connection methods of the AI computing architecture in the first computing component 23, and the architecture information is loaded as the control configuration information.
  • To the control logic receiving input parameters and output parameters of the AI computing architecture of the first computing component 23 from the device configuration interface, and loading the input parameters and output parameters into the control logic as control configuration information; receiving the first calculation from the device configuration interface
  • the training data of the component 23 loads the training data as control configuration information into the control logic.
  • FIG. 9 is a schematic diagram of a production process according to an embodiment of the present application.
  • the controller 40 controls the flow rate of the raw materials through three servo valves K1, K2, and K3, the stirring speed is controlled by the rotation speed V of the motor, and the temperature T is controlled by the thermostat set point.
  • the output parameters of the production process are quality Q1 and output Q2 that can be measured by sensors.
  • the controller 40 may send multiple sets of sample values of the production parameters K1, K2, K3, V, T, Q1, and Q2 to the calculation unit 23 having a parameter optimization function, and configure K1, K2, K3, V, T is the input parameter of the AI computing architecture. Configure Q1 and Q2 as the output parameters of the AI computing architecture.
  • the computing component 23 may train an AI computing architecture through a neural network to obtain a neural network model corresponding to the production process. After the training, the controller 40 sends the values of the production parameters K1, K2, K3, V, T, Q1, and Q2 that are actually collected to the computing unit 23.
  • the AI calculation architecture of the calculation unit 23 is based on the trained model, and a set of recommended values of the production parameters K1, K2, K3, V, and T can be obtained, and the recommended values can be used to optimize the quality Q1 and the output Q2.
  • the controller 40 may send multiple sets of sample values of the production parameters K1, K2, K3, V, T, Q1, and Q2 to the calculation unit 23 having a parameter detection function, and configure K1, K2, K3, V, T is the input parameter of the AI computing architecture. Configure Q1 and Q2 as the output parameters of the AI computing architecture.
  • the computing component 23 may train an AI computing architecture through a neural network to obtain a neural network model corresponding to the production process. After the training is over, the controller 40 sends the values of the production parameters K1, K2, K3, V, T, Q1, and Q2 that are actually collected to the computing unit 23.
  • the AI calculation architecture of the calculation unit 23 is based on the trained model, and the estimated values of Q1 and Q2 corresponding to the actual values of K1, K2, K3, V, and T can be obtained.
  • output Represents detection results with abnormal parameter values.
  • the controller 40 may also send the values of the production parameters K1, K2, K3, V, T, Q1, and Q2 collected during the fault to the computing component 23 having a fault diagnosis function.
  • the computing component 23 can obtain the probability of failure of each actuator through the AI computing architecture obtained through training. If the probability of K2 failure is the largest, a fault diagnosis result indicating the failure of K2 servo valve can be output.

Abstract

提供了应用于工业自动化系统的人工智能AI计算设备(20)。该AI计算设备(20)可以通过现场总线接口连接到现场总线,与控制器(40)通信。该AI计算设备(20)通过利用内置的AI计算架构处理控制器(40)发送的数据,对数据进行分析,并将分析结果发送给控制器(40)。还提供了相应的控制方法及装置、工程师站及工业自动化系统。

Description

人工智能计算设备、控制方法及装置、工程师站及工业自动化系统 技术领域
本申请涉及工业自动化领域,特别是一种人工智能(AI)计算设备、控制方法及装置、工程师站(Engineer Station,ES)及工业自动化系统。
背景技术
工业生产过程自主化技术经过长期不断的发展,已经能够利用计算机技术进行生产过程控制。工业自动控制系统通过工业控制计算机对传感器所采集的工业生产中的各种参数进行归纳、分析、整理,实现信息管理与自动控制。目前,在工业自动化系统中实现智能化控制成为了工业自动化领域的趋势。为了实现在控制链中的智能化,面临的挑战是传统的工业控制器不能提供足够的算力以及没有灵活的解决方案以在控制系统中添加人工智能。
技术内容
有鉴于此,本申请实施例提出了一种人工智能计算设备、控制方法及装置、工程师站及工业自动化系统,用以在工业自动化系统中实现带有人工智能的闭环控制,增强自动化系统的控制能力,提高自动化系统的控制效率。
本申请实施例的AI计算设备,应用于工业自动化系统,可以包括:背板、通信部件,以及计算部件;
所述背板,包括背板总线和现场总线接口,所述背板总线连接所述通信部件和所述计算部件,所述现场总线接口能够通过所述现场总线接口与工业自动化系统的现场总线连接通讯;其中,所述工业自动化系统包括至少一个控制器;
所述通信部件,实现所述控制器和所述计算部件之间的数据交互;及
所述计算部件,通过所述通信部件接收所述控制器发送的数据,使用内嵌的AI计算架构对所述数据进行分析,以及将分析结果通过所述通信部件发送给所述控制器。
可见,各实施例的AI计算设备具有现场总线接口,可以直接连接到工业自动化系统的现场总线,提供即插即用的智能控制功能,增强了控制系统的处理能力。同时,AI计算设备直接连接到现场总线,可以帮助系统实现实时的智能化闭环控制,提高系统的控 制效率。
本申请实施例的控制方法,应用于工业自动化系统中的控制器,可以包括:
获取工业自动化系统中的数据;
将所述数据发送给连接到现场总线的人工智能(AI)计算设备;
通过所述现场总线接收所述AI计算设备对所述数据进行分析得到的分析结果,将所述分析结果提供给控制器中的决策装置,供所述决策装置生成用于自动化控制的控制指令。
可见,各实施例的控制方法通过使用连接到现场总线的AI计算设备对数据进行分析,利用分析结果生成自动化控制指令,使得在不需要更换已有设备、且不受已有设备的处理能力的限制的情况下,增强了控制器的处理能力,还可以实现实时的智能化闭环控制。
本申请实施例的控制装置,应用于工业自动化系统中的控制器,可以包括:
生产数据获取单元,用于获取工业自动化系统中的数据;
任务发送单元,将所述数据发送给连接到所述现场总线的人工智能(AI)计算设备中的第一计算部件;
结果收集单元,用于通过所述现场总线接收所述AI计算设备发送的分析结果,所述分析结果由所述第一计算部件对所述数据进行分析得到;将所述分析结果提供给控制器中的决策装置,供所述决策装置生成用于自动化控制的控制指令。
可见,各实施例的控制装置通过使用连接到现场总线的AI计算设备对数据进行分析,利用分析结果生成自动化控制指令,使得在不需要更换已有设备、且不受已有设备的处理能力的限制的情况下,增强了控制器的处理能力,还可以实现实时的智能化闭环控制。
本申请实施例的工程师站,连接到工业自动化系统的现场总线,可以包括:处理器和存储器;所述存储器包括机器可读指令,所述指令可以由所述处理器执行用于:
提供设备配置接口;
从所述设备配置接口接收设备配置信息,所述设备配置信息包括人工智能(AI)计算设备的标识和所述AI计算设备中至少一个计算部件的计算部件标识,所述AI计算设备通过现场总线接口连接到所述现场总线;将所述设备配置信息发送到所述工业自动化 系统的控制器,以使所述控制器利用所述设备配置信息与所述AI计算设备进行通信;
获取所述AI计算设备对应的控制逻辑,从所述设备配置接口接收针对所述AI计算设备的控制配置信息,所述控制配置信息包括所述至少一个计算部件中第一计算部件的计算参数,将所述控制配置信息加载到所述控制逻辑中,将所述控制逻辑加载到所述控制器,所述控制逻辑用于使所述控制器对所述第一计算部件进行配置,将所述工业自动化网络中的数据发送给所述第一计算部件进行分析,并获取所述第一计算部件反馈的分析结果。
可见,各实施例的工程师站可以通过现场总线对控制器进行配置,从而实现控制器与AI计算设备的通信,从而在工业自动化系统中实现实时的智能化闭环控制。
本申请实施例的工业自动化系统,可以包括:工程师站(Engineer Station,ES)、控制器、生产设备和人工智能(AI)计算设备,以及连接各设备的现场总线;
所述ES,用于
提供设备配置接口,从所述设备配置接口接收配置信息,所述配置信息包括所述AI计算设备的信息,将所述配置信息加载到所述控制器;
获取所述AI计算设备对应的控制逻辑,将所述控制逻辑加载到所述控制器;
所述控制器,用于
执行所述控制逻辑,获取所述工业自动化系统中的复数个生产参数的值,将所述复数个生产参数的值发送给所述AI计算设备;接收所述AI计算设备发送的分析结果;
根据所述分析结果生成对所述生产设备的控制指令;
所述AI计算设备,用于
通过所述现场总线接收所述控制器发送的所述复数个生产参数的值,对所述复数个生产参数的值进行分析得到所述分析结果,将所述分析结果通过所述现场总线发送给所述控制器。
可见,各实施例的工业自动化系统通过采用连接到现场总线的AI计算设备,提高了系统的控制能力,同时,还能实现实时的智能化闭环控制。
本申请实施例还提供一种计算机可读存储介质,其中存储有机器可读指令,所述指令可以使处理器执行各实施例的控制方法。
附图说明
下面将通过参照附图详细描述本申请的优选实施例,使本领域的普通技术人员更清楚本申请的上述及其它特征和优点,附图中:
图1为本申请实施例的工业自动化系统的示意图。
图2为本申请实施例的一种AI计算设备的示意图。
图3为本申请实施例中AI计算设备内部数据通信的逻辑示意图。
图4为本申请实施例中AI计算设备内部数据处理的逻辑示意图。
图5为本申请实施例的一种AI计算架构的状态机示意图。
图6为本申请实施例的一种控制方法的流程图。
图7为本申请实施例的一种控制装置的示意图。
图8为本申请实施例的一种工程师站的示意图。
图9为本申请实施例的一种生产过程的示意图。
其中,附图标记如下:
序号 含义
10 工业自动化系统
20 AI计算设备
21 背板
211 背板总线
212 现场总线接口
22 通信部件
221 周期性数据
222 非周期性数据
223 诊断业务数据
23 计算部件
231 AI计算架构
232 配置数据存储模块
233 训练数据存储模块
235 状态切换模块
26 记录数据
27 IO通道
271 第一传输通道
272 第二传输通道
30 工程师站
31 控制逻辑
32 处理器
33 存储器
34 通信设备
35 操作系统
36 网络通信模块
37 管理模块
371 接口模块
372 设备配置模块
373 控制配置模块
40 控制器
41 控制装置
411 生产数据获取单元
412 任务发送单元
413 结果收集单元
414 配置单元
415 训练单元
42 决策装置
60 现场总线
S1 初始化完成
S2 准备训练
S3 训练
S4 训练错误
S5 准备操作
S6 操作
S7 操作错误
S61-S63 步骤
K1、K2、K3、V、T、Q1、Q2 参数
具体实施方式
为使本申请的目的、技术方案和优点更加清楚,以下举实施例对本申请进一步详细说明。
为了克服控制器计算力对工业自动化系统控制机制的限制,本申请实施例提出在工业自动化系统的控制闭环中加入具有人工智能计算能力的AI计算设备,从而增强系统的自主控制能力,并且在控制链中实现智能化。图1为本申请实施例的工业自动化系统的示意图。工业自动化系统10是采用自动控制和自动调整装置,用以代替人工操纵机器和机器体系进行加工生产的工业生产系统。如图1所示,该系统10可以包括:AI计算设备20、工程师站(Engineering Station,ES)30、控制器40,以及连接各设备的现场总线60。
现场总线60,也称为工业数据总线,用于实现工业生产现场的控制器40、数据采集设备(未示出)、执行器(未示出)等现场设备间的数字通信,以及这些现场设备和高级控制系统(如工程师站ES 30)之间的信息传递。现场总线60可以采用某种现场总线技术实现,例如ProfiBus、InterBus、控制器局域网(CAN)总线、可寻址远程传感器高速通道(HART)总线等。
控制器40可以是一个或复数个工业控制器,例如可编程逻辑(PLC)控制器、PC总线工业电脑(IPC)控制器、分散型控制系统(DCS)控制器、现场总线控制系统(FCS)控制器、数控系统(CNC)控制器,等。控制器40可以与生产现场的各种数据采集设备(例如智能化仪器仪表、传感器(sensor)等)、执行器(actuator)(如,电流调节阀、电压调节阀、进料阀等)进行通信,对生产设备的生产过程进行监测和控制。生产设备(未示出)是指用于工业加工、制造的一个或复数个设备的集合,例如机床、车床、流水线设备等。当生产设备为具有现场总线接口的程控设备时,控制器40还可以通过现场 总线60与生产设备进行通信,收集生产设备的过程信息,对生产设备的生产过程进行监测和控制。
控制器40可以获取生产设备的各种参数值(下文也称为生产参数的值、生产过程数据),并根据这些生产参数值对生产设备的运行施加控制。生产参数可以是任意与生产设备的生产过程有关的参数,例如生产设备的电压、电流、电机转速、原料的进料速度,等。生产参数的值可以通过数据采集设备获得,也可以从程控生产设备获得。控制器40还可以通过向执行器(actuator)发送控制指令,来调整生产参数的值。执行器可以接收控制器40的控制信号并执行相应的调整动作来改变生产参数的值。
AI计算设备20是一个具有AI计算能力的即插即用设备,可以通过现场总线接口连接到现场总线。AI计算设备20可以通过现场总线60接收控制器40中的控制装置41发送的复数个生产参数的值,对复数个生产参数的值进行分析得到分析结果,将分析结果通过现场总线60发送给控制器40,使控制器40中的决策装置42根据分析结果生成用于自动化控制的控制指令。一些实施例中,系统10可以连接复数个AI计算设备20。复数个AI计算设备20可以包括具有不同的AI功能的设备,也可以包括完全相同的几个设备用于提供冗余备份或进行负载分担。
为了使控制器40能够识别和使用AI计算设备20,可以在工程师站ES 30对控制器40进行针对AI计算设备20的配置。
工程师站ES 30是指供工业过程控制工程师使用的工业过程监控管理设备,可以对工业控制系统中的控制器40进行配置,使控制器40能够与生产现场的其它设备进行通信、执行数据处理、做出控制决策,从而实现对生产过程的监测和控制。
工程师站ES 30可以提供设备配置接口,从设备配置接口接收针对AI计算设备20的配置信息,并将配置信息加载到控制器40。控制器40利用配置信息与AI计算设备20进行通信。
一些实施例中,ES 30的设备配置接口可以包括人机交互界面。ES 30可以通过人机交互界面展示系统10中各设备的配置和运行情况,并通过人机交互界面展示配置接口。响应于操作员对人机交互界面中配置接口的操作,ES 30可以完成各种设备配置操作,例如添加设备、删除设备、添加设备配置信息、修改设备配置信息、删除设备配置信息,等。从设备配置接口接收到的配置信息可以包括操作员向ES 30输入的文本信息(例如AI计算设备的标识、地址,等),从操作员通过设备配置接口输入或选定的路径获取的配置文件、控制逻辑,等。操作员输入或选定的路径可以是ES 30内置的存储设备中的 存储路径,也可以是ES 30设备的外接扩展存储设备中的路径,还可以是网络中的位置(如URL,等)。一些实施例中,配置信息可以包括AI计算设备20的标识(例如媒体接入控制(MAC)地址、设备名称,等),使得控制器40可以与AI计算设备20通信。一些实施例中,配置信息还可以包括AI计算设备20执行AI计算过程所需要的复数个生产参数的信息,使得控制器40可以将通过传感器采集到的这些生产参数的值提供给AI计算设备20。
一些实施例中,ES 30还可以获取AI计算设备20对应的控制逻辑31,将控制逻辑31加载到控制器40。例如,ES 30可以从操作员通过设备配置接口输入或选定的路径获取控制逻辑31。操作员输入或选定的路径可以是ES 30内置的存储设备中的存储路径,也可以是ES 30的外接扩展存储设备中的路径,还可以是网络中的位置(如URL,等)。ES 30还可以通过控制逻辑31提供的接口对控制逻辑31的功能进行配置和编辑。控制器40通过执行控制逻辑31实现对AI计算设备20的控制。上述复数个生产参数的信息也可以通过控制逻辑31提供的接口加载到控制逻辑31中,控制器40通过执行控制逻辑31来获取这些生产参数的值,并提供给AI计算设备20。一些实施例中,AI计算设备20可以包括通信部件22和一个或复数个计算部件23。AI计算设备20可以包括具有不同AI功能的复数个计算部件23,也可以具有复数个功能相同的计算部件。具有不同AI功能的计算部件23可以对应不同的控制逻辑31,具有不同AI功能的计算部件23可以由同一个控制逻辑31控制。
各实施例的工业自动化系统,通过采用具有现场总线接口的AI计算设备,使得在不需要更换已有设备、且不受已有设备的处理能力的限制的情况下,可以实现即插即用的智能控制功能,增强了控制系统的处理能力。同时,AI计算设备直接连接到现场总线,生产过程数据直接在现场进行分析处理,可以实现智能化闭环控制,提高控制效率。
为了实现即插即用,AI计算设备20具有可连接到现场总线的物理接口,并支持采用工业自动化系统使用的工业通信协议进行通信。下面以一个AI计算设备20的实现方式为例进行说明。图2为本申请实施例的一种AI计算设备的示意图。如图2所示,该AI计算设备20可以包括:背板21、通信部件22,以及计算部件23。
背板21,包括背板总线211和现场总线接口212。背板总线211用于连接通信部件22和计算部件23。现场总线接口212能够通过现场总线接口212与现场总线60连接通讯。现场总线接口212为符合现场总线60所采用的现场总线技术的接口,例如ProfiBus、 InterBus、CAN、HART等接口。背板总线211可以采用任意总线技术,如根据需要设计的总线技术,或已有的数据传输总线。例如,背板总线211提供的背板总线接口可以为低电压差分信号(Low-Voltage Differential Signaling)接口、S422接口、RS485接口,等。
通信部件22可以实现控制器40和计算部件23之间的数据交互。
计算部件23包括AI计算架构。一些实施例中,AI计算架构可以包括AI计算逻辑,如机器学习逻辑,神经网络算法。一些实施例中,AI计算架构还可以包括为AI计算定制的具有高计算性能的专用硬件(如GPU、FPGA、ASIC、神经网络处理器,等)。计算部件23可以通过通信部件22接收控制器40发送的数据,使用内嵌的AI计算架构对数据进行分析,将分析结果通过通信部件22发送给控制器40。
各实施例的AI计算设备20,由于具有现场总线接口,可以在工业自动化系统中实现即插即用的智能控制功能,增强了工业自动化系统的控制能力。同时,AI计算设备在生产现场对生产过程数据进行分析处理,可以实现智能化闭环控制,提高控制效率。
一些实施例中,为了实现AI计算设备20与控制器的通信,通信部件22可以利用工业通信协议(如PROFINET、EtherCat,等)定义的报文格式与现场总线60进行通信。该通信部件22可以使用该工业通信协议定义的报文格式对控制器40发送的报文进行解析,将解析得到的报文内容发送给计算部件23。该通信部件22还可以使用该报文格式对计算部件23发送的反馈数据进行封装,将封装得到的报文发送给控制器40。通过采用已有的工业通信协议报文与控制器40进行通信,使得AI计算设备20能够直接使用在现有的工业通信网络中,不需要修改网络,使用方便。一些实施例中,通信部件22可以由FPGA、ASIC、集成电路、工业通信芯片等实现。
AI计算设备20可以包括一个或者复数个计算部件23。当AI计算设备20中设置了一个计算部件23时,通信部件22可以在计算部件23和控制器40之间转发数据。当AI计算设备20包括复数个计算部件23时,通信部件22可以利用各计算部件23的计算部件标识来实现各计算部件23与控制器40通信。一些实施例中,通信部件22可以通过现场总线接口212接收控制器40发送的符合工业通信协议的报文(为了与AI计算设备20发出的报文进行区分,下文将其称为第一报文)。其中,第一报文包括报头(header)和负载(Payload),负载包括一个或者复数个计算部件23的标识(以下称为计算部件标识)以及各计算部件标识对应的报文数据或报文内容。计算部件23的计算部件标识用于在AI计算设备20中区别各个计算部件23,可以是名称、序列号等,可以是在生产时配置 在计算部件23中的设备识别码,也可以是控制器40分配的其它标识。通信部件22可以根据该工业通信协议定义的报文格式从第一报文中解析出计算部件23的标识以及和该标识对应的报文数据,并将报文数据通过背板总线211发送给与该标识对应的计算部件23。通信部件22还可以通过背板总线211从第一计算部件23接收反馈数据(第一计算部件23是复数个计算部件23中的一个计算部件),将反馈数据和第一计算部件23的计算部件标识作为报文数据生成具有该工业通信协议定义的报文格式的报文(为了与控制器40发出的报文进行区分,这里将其称为第二报文),将第二报文通过现场总线接口212发送给控制器40。其中,反馈数据是计算部件23对报文数据的处理结果,用于使控制器40掌握计算部件23的状态、获得分析结果,等,从而产生对AI计算设备20或其它设备(如生产设备、执行器等)的控制决定。例如,当报文数据是配置信息时,反馈数据可以是配置完成的确认消息;当报文数据是训练指令时,反馈数据可以是训练的结果信息;当报文数据是生产过程数据时,反馈数据可以是对生产过程数据的分析结果。这样,通过在通信中增加计算部件23的标识,使得控制器40能够与AI计算设备20中的复数个计算部件23进行通信,单个AI计算设备20中可以包括复数个计算部件23,大大提高了单个AI计算设备20的计算容量和计算能力。
一些实施例中,计算部件23和控制器40之间交互的数据可以分为复数种类型,例如可根据优先级不同、传输时间要求不同,等,将数据分为复数种类型。通信部件22可以根据接收到的数据的类型对应的预设处理策略来处理该数据。预设的处理策略可以包括针对数据处理顺序的策略、数据发送顺序的策略、数据的发送方式的策略,等。一些实施例中,可以为不同类型的数据预设不同类型的逻辑传输通道来处理的数据。通信部件22可以根据控制器40发送的第一报文中数据对应的数据通道标识确定数据的类型,并根据该类型对应的处理策略处理该数据;通过背板总线211获取第一计算部件23提供的反馈数据的类型,通过与反馈数据的类型具有预设对应关系的传输通道,将反馈数据对应的第二报文通过现场总线接口212发送给控制器40。
图3为本申请实施例中AI计算设备20中计算部件23与通信部件22之间内部数据通信的示意图。如图3所示,AI计算设备20中的一个或复数个计算部件23可以将反馈数据及类型发送给通信部件22。通信部件22根据反馈数据的类型将反馈数据映射到不同的数据通道,并通过现场总线接口212传输到现场总线60。该例子中,反馈数据的类型可以包括,周期性数据221、非周期性数据222、诊断业务数据223,等。
图4为本申请实施例中AI计算设备20中计算部件23使用不同的逻辑数据通道与控制器40进行数据通信的示意图。如图4所示,AI计算设备20与控制器40之间的逻辑通道(下文也称为IO通道27)分为第一传输通道271和第二传输通道272。第一传输通道271为非周期性传输通道,即其传输时机不具有周期性,仅在需要时传输。第二传输通道272为周期性传输通道,以固定的时间间隔传输。第一传输通道271和第二传输通道272可以通过通道标识进行区分。
图4中,计算部件23可以通过第一传输通道271与控制器40进行记录数据(record data)的交互。记录数据可以是一些重要性较低或者通信实时性要求较低的数据,例如配置信息(如AI计算架构的结构的配置信息、AI计算架构231的输入参数和输出参数的配置信息,等),训练数据,等。计算部件23可以将配置信息存储在内置的配置数据存储模块232中,将训练数据存储在训练数据存储模块233中。一些情况下,计算部件23可以将内置的配置数据存储模块232中的配置信息通过第一传输通道271发送给控制器40,控制器40可以根据这些配置信息来调整其它模块的计算架构。
图4中,计算部件23还可以通过第二传输通道272与控制器40进行控制数据的交互。例如,控制数据可以包括控制器40发送的对AI计算架构231的状态切换指令、控制器40发送的待分析的生产过程数据、计算部件23发送的AI计算架构231的状态报告,等。
计算部件23可以包括状态切换模块235,用于根据控制器40发送的状态切换指令改变AI计算架构231的运行状态。例如,状态切换模块235可以使用第二传输通道272,将AI计算架构231的当前运行状态通过通信部件22发送给控制器40,及通过通信部件22接收控制器40发送的状态切换指令,将AI计算架构231从第一运行状态切换到第二运行状态。例如,状态切换模块235可以通过输入状态字(ISW)将AI计算架构231的状态通过第二传输通道272发送给控制器40,根据控制器40发送的输出控制字(OCW)来切换AI计算架构231的状态。
状态切换模块235还可以根据AI计算架构(231)的当前运行状态,将控制器(40)提供的与当前运行状态对应的数据输入AI计算架构(231)。例如,当AI计算架构231处于训练状态时,状态切换模块235可以将控制器40通过第一传输通道271发送、并存储在训练数据存储模块233中的训练数据输入AI计算架构231,用于训练AI计算架构231。当AI计算架构231处于操作状态时,状态切换模块235可以将控制器40通过第二传输通道272周期性发送的待分析的数据输入AI计算架构231,使AI计算架构231输 出分析结果。AI计算架构231产生的分析结果也可以通过第二传输通道272反馈到控制器40。
这样,通过采用不同的传输通道来传输不同类型的数据,提高了AI计算设备20的数据响应能力和处理效率。
系统10中可以设置一个或复数个AI计算设备20,每个AI计算设备20可以包括一个或复数个计算部件23,不同的计算部件23可以具有不同功能的AI计算架构231。AI计算架构231可以实现的功能可以包括,但不限于,在线参数优化、过程监控、故障诊断,等。
例如,当AI计算架构231包括用于在线优化的优化计算过程时,AI计算架构231可以执行预设的优化计算过程对生产过程数据中的复数个生产参数进行优化,输出复数个生产参数的建议值。一些例子中,计算模块23还可以将建议值转换为控制器40可识别的生产调整建议作为分析结果,生产调整建议包括多少复数个生产参数中至少一个生产参数的建议值。一些例子中,可以预设针对建议值的处理策略。例如,可以预设一阈值,当参数的建议值与实际值的差异不大于该阈值时,可以在分析结果中不包括该参数的调整建议。这样可以避免频繁、不必要地改变生产参数。
又例如,当AI计算架构231包括用于状态监控的参数检测过程时,AI计算架构231可以执行预设的参数检测过程对生产过程数据中的复数个生产参数的值进行检测,输出参数检测结果,参数检测结果指示复数个生产参数的值是否正常。一些例子中,计算模块23还可以将参数检测结果转换为控制器40可识别的状态监控报告作为分析结果,状态监控报告包括用于指示生产设备的状态是否正常的信息。
再例如,当AI计算架构231包括故障诊断过程时,AI计算架构231可以执行预设的故障诊断过程利用生产过程数据中的复数个生产参数的值进行故障诊断,输出的故障诊断结果,故障诊断结果包括系统10中的一个部件(例如采集器、执行器、生产设备,等)的信息。一些例子中,计算模块23还可以将故障诊断结果转换为控制器40可识别的故障报告作为分析结果,故障报告包括部件的信息。
通过采用不同功能的AI计算架构231,可以增强工业自动化系统的控制能力。
一些实施例中,计算部件23还可以包括配置单元(未示出),用于对AI计算架构231进行配置。配置单元可以通过通信部件22接收控制器40发送的配置参数,配置参 数包括AI计算架构231的复数个属性值;将AI计算架构231对应的复数个属性的值设置为配置参数中的复数个属性值。例如,复数个属性值可以包括结构属性值,配置单元可以利用配置参数中的结构属性值,设置AI计算架构231的部件和连接方式,例如AI计算架构231中神经网络的层次结构。不同的AI功能和不同的生产工艺具有不同的计算需求,可以通过ES 30将所需的AI计算架构的结构信息配置到控制器40,由控制器40配置相应的计算部件23。又例如,复数个属性值可以包括第一参数和第二参数,第一参数为生产设备的一个或复数个生产参数。配置单元可以将第一参数设置为AI计算架构231的输入参数;将第二参数设置为将AI计算架构231的输出参数。不同的生产过程具有不同生产参数,可以通过ES 30将需要使用的生产参数的信息配置到控制器40,由控制器40配置相应的计算部件23。配置单元还可以将配置参数保存在配置数据存储模块232中。通过开放计算部件23的配置接口,可以扩大AI计算设备20的适用范围,避免研发不同场景下的专用AI计算设备20带来的成本过高的问题。
一些实施例中,状态切换模块235还可以利用状态机机制对AI计算架构231的运行状态进行管理。图5为本申请实施例的一种AI计算架构的状态机示意图。如图5所示,AI计算架构231可以有7种状态,分别是初始化完成(S1)、准备训练(S2)、训练(S3)、训练错误(S4)、准备操作(S5)、操作(S6),以及操作错误(S7)。操作是指AI计算架构231利用训练好的模型对生产过程数据进行分析。相应地,控制器40可以利用6种控制字来控制AI计算架构231的状态迁移。6种控制字用6种信号实现。例如,可以用具有6个比特的信息来表示6种控制字,每个比特位表示一个指令。例如,bit 1表示准备训练(从S1迁移到S2),bit2表示发起操作(从S1迁移到S5),bit3表示开始训练(从S2迁移到S3),bit4表示开始操作(从S5迁移到S6),bit5表示结束训练(从S2/S3迁移到S1),bit6表示结束操作(从S6/S5迁移到S1)。将某指令对应的比特位置位,则表示要求AI计算架构231进行该指令对应状态迁移操作。当AI计算架构231进入S4或S7的错误状态时,可以在向控制器40报告状态S4或S7后,不需等待控制器40的指令,直接进入S1。通过使用状态机来对AI计算架构231进行状态管理,使得AI计算架构231的运行控制更规范,便于管理。
控制器40还可以利用计算部件23的状态机进行复数个计算部件23的并行任务的同步。例如,控制器40把计算任务分配给复数个计算部件23,每个计算部件23计算完成都会把状态位设置为S1,那么控制器40只需要等待这些计算部件23都返回状态S1时,就确 定这些计算部件23当前已同步,可以开始下一个周期的操作。
一些实施例中,计算部件23还可以包括节能单元(未示出),用于根据控制器40的指令改变计算部件23的工作模式。节能单元可以通过通信部件22接收控制器40发送的第一指令,根据第一指令进入低功耗模式;通过通信部件22接收控制器40发送的第二指令,根据第二指令退出低功耗模式。例如,当采用PROFINET协议时,控制器40会在自动化系统进入或退出低功耗模式时,向所有设备发送进入睡眠或苏醒的PROFIenergy命令。计算部件23可以根据控制器40发送的PROFIenergy命令进入或退出低功耗模式。这样可以在自动化系统运行在低功耗模式时,降低AI计算设备20的能源消耗。
一些实施例中,背板21可以包括复数个插槽,计算部件23作为可插拔扩展卡通过插槽连接到背板总线211。这样,当需要扩充AI计算设备20的计算能力时,可以通过插槽接入新的计算部件23。当需要更换计算部件23(例如要将已有的进行参数优化的计算部件23替换为进行故障诊断的计算部件23)时,可以将旧的计算部件23从插槽移除,并插入新的计算部件23。这种可插拔的设计可以方便地扩展AI计算设备20的计算能力或改变AI计算设备20的AI功能。
下面对控制器40对AI计算设备20的控制方法进行说明。图6是本申请实施例的一种控制方法的流程图。该方法可以由控制器40中的控制装置41执行。该方法可以包括以下步骤。
步骤S61,获取工业自动化系统中的数据。
步骤S62,将数据通过工业自动化系统的现场总线60发送给连接到现场总线60的AI计算设备。
步骤S63,通过现场总线60接收AI计算设备20对该数据进行分析得到的分析结果,将分析结果提供给控制器40中的决策装置42,供决策装置生成对生产设备的控制指令。
各实施例的控制方法,通过利用具有现场总线接口AI计算设备对生产过程数据进行分析,可以在工业自动化系统中实现智能化的闭环控制,提高控制效率。
一些实施例中,为了确定需要提供给AI计算设备20的生产参数,控制装置41可以 将预设的第一配置信息中的复数个输入参数确定为复数个生产参数,从工业自动化系统中的数据采集设备提供的数据中获取复数个生产参数的值作为生产过程数据。一些实施例中,控制装置41可以每隔预设长度的时间段,从前一时间段内数据采集设备提供的数据中获取复数个生产参数的值作为生产过程数据。通过周期性地将生产过程数据发送给AI计算设备20进行分析,可以比较及时地获得当前生产过程的分析结果,便于及时调整生产参数,提高生产效率。
一些实施例中,将生产过程数据发送给AI计算设备20时,控制装置41可以从预设的第二配置信息中获取AI计算设备20中的第一计算部件23的信息;将生产过程数据发送给第一AI计算设备20中的第一计算部件23。通过从配置信息中获取计算部件23的信息,可以在计算部件23有更换、新增等情况下,仅通过更新配置信息就可以实现对控制装置41的适应性调整,方便灵活。
当AI计算设备20包括复数个计算部件23时,控制装置41需要获取这些计算部件的信息从而实现与各个计算部件的通信。一些实施例中,控制装置41可以获取第二配置信息,其中记录有复数个计算部件的信息。控制装置41可以从第二配置信息中获取复数个计算部件23的信息,并利用该信息向各计算模块23分配计算任务。
为了实现负载均衡,控制装置41可以从至少一个AI计算设备20获得复数个计算部件23的负载信息;根据负载信息从复数个计算部件23中选择一个计算部件23作为第一计算部件23,将第一计算部件23所属的AI计算设备20确定为AI计算设备20。例如,计算部件23可以将其负载信息(例如计算负荷、处理器的温度、计算带宽(每秒完成的运算次数),等)通过识别和维护(I&M)信息上传到控制器40,控制器40可以根据负载均衡算法(例如装箱算法,等)向计算部件23分配计算任务,发送该计算任务对应的生产过程数据。通过采用复数个计算部件23进行负载分担,可以在计算任务量较大的情况下仍然保证控制系统的平稳运行,提高运行效率。计算任务的分配方式根据使用的AI算法而定,例如复数个计算部件23可以处理不同的参数,或者处理不同时段的数据,或者并行处理同一数据集合,等。
一些实施例中,控制装置41可以按照固定的周期,利用周期性数据来向计算部件23输入归一化的生产过程数据,并获取计算部件23归一化的输出数据。不同结构的计算部件23可能有不同数量的输入数据和输出数据。可以在控制装置41的配置信息中指明各计算部件23所采用的报文长度(如16字节、64字节,等)。配置信息可以通过例如配置文件、设备描述文件,等方式由ES 30提供给控制装置41。例如,配置文件可以 是GSDML文件,可以包括制造商信息、通讯端口信息、模块和子模块信息,及报警诊断信息,等。通过配置各计算部件23支持的报文长度,可以高效、灵活地实现控制装置41与各计算部件23的数据通信
一些实施例中,不同的计算部件可以具有不同功能的AI计算架构231,不同功能的AI计算架构231的输出数据也不同,提供给决策装置的分析结果也不同。
例如,当AI计算架构231包括用于在线优化的优化计算过程时,AI计算架构231对复数个生产参数进行优化。控制装置41可以获得AI计算架构231输出的复数个生产参数的建议值,将建议值转换为决策装置42可识别的生产调整建议作为提供给决策装置42的分析结果,生产调整建议包括多少复数个生产参数中至少一个生产参数的建议值。
又例如,当AI计算架构231包括用于状态监控的参数检测过程时,AI计算架构231中对复数个生产参数的值进行检测。控制装置41可以获得AI计算架构231输出的参数检测结果,参数检测结果指示复数个生产参数的值是否正常;将参数检测结果转换为决策装置42可识别的状态监控报告作为提供给决策装置42的分析结果,状态监控报告包括用于指示生产设备的状态是否正常的信息。
再例如,当AI计算架构231包括故障诊断过程时,AI计算架构231利用复数个生产参数的值进行故障诊断。控制装置41可以获得AI计算架构231输出的故障诊断结果,故障诊断结果包括生产设备的一个部件的信息;将故障诊断结果转换为决策装置42可识别的故障报告作为提供给决策装置42的分析结果,故障报告包括该部件的信息。
在控制装置41执行AI计算结果到分析结果的转换,可以简化计算部件23的实现。并且,由于控制装置41的控制过程可以通过配置和编程等方式调整,在控制装置41处理AI计算架构的输出数据,调整起来也比较容易。
一些实施例中,需要对计算部件的参数进行配置时,控制装置41可以从预设的第三配置信息中获取第一计算部件23的配置参数,配置参数包括第一计算部件23采用的AI计算架构的复数个属性值;将配置参数发送给第一计算部件23,使第一计算部件23将AI计算架构的复数个属性的值设置为复数个属性值。AI计算架构的复数个属性值可以包括结构属性值,或者输入参数、输出参数,等。
一些实施例中,AI计算架构的训练数据也可以从配置信息中获得。例如,控制装置41可以从预设的第四配置信息中获取训练数据,将训练数据发送给第一计算部件23,使第一计算部件23利用训练数据对第一计算部件23中的AI计算架构进行训练。
各实施例的控制方法可以由设置在控制器40中的控制装置41执行。图7为本申请实施例的一种控制装置的示意图。如图7所示,该控制装置41可以包括:生产数据获取单元411、任务发送单元412和结果收集单元413。
生产数据获取单元411可以获取工业自动化系统中生产设备的生产过程数据,生产过程数据包括生产设备的复数个生产参数的值。
任务发送单元412可以将生产过程数据通过工业自动化系统的现场总线60发送给连接到现场总线60的AI计算设备中的第一计算部件23。
结果收集单元413可以通过现场总线60接收AI计算设备20发送的分析结果,分析结果由第一计算部件23利用生产过程数据对生产设备的生产过程进行分析得到;将分析结果提供给控制器40中的决策装置,供决策装置生成对生产设备的控制指令。
一些实施例中,该控制装置还可以包括配置单元414。配置单元414从预设的第三配置信息中获取第一计算部件23的配置参数,配置参数包括第一计算部件23的AI计算架构的复数个属性值;将配置参数发送给第一计算部件23,从而将AI计算架构的复数个属性的值设置为复数个属性值。
一些实施例中,该控制装置还可以包括训练单元415。训练单元415可以从预设的第四配置信息中获取训练数据,将训练数据发送给第一计算部件23,以使第一计算部件23利用训练数据对AI计算架构进行训练。
本申请实施例的控制方法还可以由软件代码实现。该软件代码可以存储在计算机可读存储介质中。该软件代码可以为符合工业控制编程语言标准(如IEC61131-3)的机器可读指令。一些实施例中,该软件代码可以由ES 30从远程的存储设备、本地存储设备或者可移除存储设备(如光盘、闪存等)中读取,进行配置后加载到控制器40,由控制器40执行,以实现上述控制方法。
各实施例中,工程师站ES 30可以将对AI计算设备20进行控制所需的配置信息和控制逻辑提供给控制器40,从而使控制器40具有对AI计算设备20的控制能力。图8为本申请实施例的一种工程师站的示意图。如图5所示,该ES 30可以包括:处理器32、存储器33和通信设备34。通信设备34用于使ES 30与网络中的其它设备进行通信。存储器33可以包括管理模块37。
管理模块37包括接口模块371、设备配置模块372和控制配置模块373。
接口模块371可以提供设备配置接口。
设备配置模块372可以从设备配置接口接收设备配置信息,设备配置信息包括AI计算设备的标识和AI计算设备20中至少一个计算部件23的标识信息;将设备配置信息发送到工业自动化系统的控制器40,以使控制器40利用设备配置信息与AI计算设备20进行通信。
控制配置模块373可以获取AI计算设备20对应的控制逻辑31,从设备配置接口接收针对AI计算设备20的控制配置信息,控制配置信息包括至少一个计算部件23中第一计算部件23的计算参数;将控制配置信息加载到控制逻辑31中,将控制逻辑31加载到控制器40。该控制逻辑31可以使控制器40对第一计算部件23进行配置,将工业自动化网络中生产设备的生产过程数据发送给第一计算部件23进行分析,并获取第一计算部件23反馈的分析结果。
接口模块371、设备配置模块372和控制配置模块373可以由机器可读指令实现。
一些实施例中,存储器33还可以包括操作系统35和网络通信部件36。
一些实施例中,设备配置模块372可以从设备配置接口接收第一计算部件23的通道信息,通道信息包括一个或复数个通道的标识;将通道信息发送到控制器40,使控制器40将数据根据数据的类型对应的通道发送给第一计算部件23。
一些实施例中,控制配置模块373可以从设备配置接口接收第一计算部件23的架构信息,架构信息包括第一计算部件23中AI计算架构的部件和连接方式,将架构信息作为控制配置信息加载到控制逻辑中;从设备配置接口接收第一计算部件23的AI计算架构的输入参数和输出参数,将输入参数和输出参数作为控制配置信息加载到控制逻辑中;从设备配置接口接收第一计算部件23的训练数据,将训练数据作为控制配置信息加载到控制逻辑中。
下面举一个对工业生产过程进行控制的例子,使各实施例的工业自动化系统的控制机制更容易理解。这里仅仅是一个简单的例子,其它实施例中可能涉及更多的设备和生产参数。图9为本申请实施例的一种生产过程的示意图。如图9所示,控制器40通过3个伺服阀K1、K2、K3控制原料的流量,通过电机的转速V控制搅拌速度,通过恒温器设定点控制温度T。生产过程的输出参数是可以通过传感器测量得到的质量Q1和产量Q2。
一些实施例中,控制器40可以将生产参数K1、K2、K3、V、T、Q1、Q2的多组样本值发送到具有参数优化功能的计算部件23,配置K1、K2、K3、V、T为AI计算架构的输入参数,配置Q1、Q2为AI计算架构的输出参数。计算部件23可以通过神经网络训练AI计算架构,获得该生产过程对应的神经网络模型。训练结束后,控制器40将实际采集得到的生产参数K1、K2、K3、V、T、Q1、Q2的值发送到计算部件23。计算部件23的AI计算架构基于训练得到的模型,可以得到一组生产参数K1、K2、K3、V、T的建议值,采用该建议值可以使质量Q1和产量Q2达到最优。
一些实施例中,控制器40可以将生产参数K1、K2、K3、V、T、Q1、Q2的多组样本值发送到具有参数检测功能的计算部件23,配置K1、K2、K3、V、T为AI计算架构的输入参数,配置Q1、Q2为AI计算架构的输出参数。计算部件23可以通过神经网络训练AI计算架构,获得该生产过程对应的神经网络模型。训练结束后,控制器40将实际采集得到的生产参数K1、K2、K3、V、T、Q1、Q2的值发送到该计算部件23。计算部件23的AI计算架构基于训练得到的模型,可以得到K1、K2、K3、V、T的实际值对应的Q1、Q2的估计值,当Q1、Q2的实际值与估计值不符时,输出表示参数值不正常的检测结果。
一些实施例中,控制器40还可以将故障时采集得到的生产参数K1、K2、K3、V、T、Q1、Q2的值发送到具有故障诊断功能的计算部件23。该计算部件23通过训练得到的AI计算架构可以得到各个执行器故障的概率。如果K2故障的概率最大,可以输出表示K2伺服阀故障的故障诊断结果。
可见,使用本申请实施例的即插即用的AI计算设备20,可以方便地在现有的工业自动化控制系统中实现带有人工智能的闭环控制,从而提高生产效率。
以上仅为本申请的较佳实施例而已,并不用以限制本申请,凡在本申请的精神和原则之内,所作的任何修改、等同替换、改进等,均应包含在本申请的保护范围之内。

Claims (32)

  1. 人工智能(AI)计算设备(20),应用于工业自动化系统,包括:背板(21)、通信部件(22),以及计算部件(23);
    所述背板(21),包括背板总线(211)和现场总线接口(212),所述背板总线(211)连接所述通信部件(22)和所述计算部件(23),所述现场总线接口(212)能够通过所述现场总线接口与工业自动化系统的现场总线(60)连接通讯,其中,所述工业自动化系统包括至少一个控制器(40);
    所述通信部件(22),实现所述控制器(40)和所述计算部件(23)之间的数据交互;及
    所述计算部件(23),通过所述通信部件(22)接收所述控制器(40)发送的数据,利用内嵌的AI计算架构(231)对所述数据进行分析,以及将分析结果通过所述通信部件(22)发送给所述控制器(40)。
  2. 根据权利要求1所述的AI计算设备(20),其中,所述通信部件(22),
    所述通信部件(22)使用一种工业通信协议定义的报文格式对所述控制器(40)发送的报文进行解析,将解析得到的报文内容发送给所述计算部件(23);以及
    使用所述报文格式对所述计算部件(23)发送的反馈数据进行封装,将封装得到的报文发送给所述控制器(40)。
  3. 根据权利要求2所述的AI计算设备(20),其中,所述AI计算设备(20)包括复数个计算部件(23)时,其中每个所述计算部件(23)对应一个计算部件标识,所述通信部件(22)能够,
    通过所述现场总线接口(212)接收所述控制器(40)发送的第一报文;
    根据所述第一报文的报文格式从所述第一报文中解析出计算部件标识以及和所述计算部件标识对应的报文数据;及
    将所述报文数据通过所述背板总线发送给所述计算部件标识对应的计算部件(23)。
  4. 根据权利要求2或3所述的AI计算设备(20),其中,所述AI计算设备(20)包括复数个计算部件(23)时,所述通信部件(22)能够
    通过所述背板总线(211)从一个第一计算部件(23)接收反馈数据,其中所述第一计算部件(23)是所述复数个计算部件(23)中的一个计算部件(23);及
    将所述反馈数据和所述第一计算部件(23)的第一计算部件标识作为报文数据生成具有所述报文格式的第二报文,将所述第二报文通过所述现场总线接口(212)发送给所述 控制器(40)。
  5. 根据权利要求4所述的AI计算设备(20),其中,所述通信部件(22)能够,
    通过所述背板总线(211)获取所述第一计算部件(23)提供的所述反馈数据的类型;及
    通过与所述反馈数据的类型具有预设对应关系的传输通道,将所述第二报文通过所述现场总线接口(212)发送给所述控制器(40)。
  6. 根据权利要求1至5中任意一项所述的AI计算设备(20),其中,所述计算部件(23)进一步包括一个状态切换模块(235),所述状态切换模块(235)能够,
    根据所述计算部件(23)的当前运行状态,将所述控制器(40)提供的与所述当前运行状态对应的数据输入所述AI计算架构(231)。
  7. 根据权利要求6所述的AI计算设备(20),其中,所述状态切换模块(235)能够,
    将所述AI计算架构(231)的当前运行状态通过所述通信部件(22)发送给所述控制器(40);
    通过所述通信部件(22)接收所述控制器(40)发送的状态切换指令,将所述AI计算架构(231)从一个第一运行状态切换到一个第二运行状态。
  8. 根据权利要求6或7所述的AI计算设备(20),其中,所述状态切换模块(235)能够,
    当所述AI计算架构(231)的当前运行状态为训练状态时,将所述控制器(40)提供的训练数据输入所述AI计算架构(231),用于训练所述AI计算架构(231)。
  9. 根据权利要求6至8中任意一项所述的AI计算设备(20),其中,所述状态切换模块(235)能够,
    当所述AI计算架构(231)的当前运行状态为操作状态时,将所述数据中的复数个生产参数的值输入所述AI计算架构(231),执行所述AI计算架构(231)中预设的计算过程。
  10. 根据权利要求1至9中任意一项所述的AI计算设备(20),其中,所述AI计算架构(231)能够,
    执行预设的优化计算过程对所述数据中的复数个生产参数的值进行优化,输出所述复数个生产参数的建议值。
  11. 根据权利要求1至9中任意一项所述的AI计算设备(20),其中,所述AI计算架构(231)能够,
    执行预设的参数检测过程,对所述数据中的复数个生产参数的值进行检测,输出参数检测结果,所述参数检测结果指示所述复数个生产参数的值是否正常。
  12. 根据权利要求1至9中任意一项所述的AI计算设备(20),其中,所述AI计算架构(231)能够,
    执行预设的故障诊断过程,利用所述数据中的复数个生产参数的值进行故障诊断,输出故障诊断结果,所述故障诊断结果包括所述工业自动化系统中一个部件的信息。
  13. 根据权利要求1至12中任意一项所述的AI计算设备(20),其中,所述计算部件(23)进一步包括配置单元,所述配置单元能够,
    通过所述通信部件(22)接收所述控制器(40)发送的配置参数,所述配置参数包括所述AI计算架构(231)的结构属性值;及
    利用所述配置参数中的结构属性值,配置所述AI计算架构(231)的部件的结构属性。
  14. 根据权利要求13所述的AI计算设备(20),其中,所述配置单元能够,
    通过所述通信部件(22)接收所述控制器(40)发送的配置参数,其中所述配置参数包括第一参数和第二参数,所述第一参数为所述生产设备的一个或复数个生产参数;
    将所述AI计算架构(231)的输入参数配置为所述第一参数;及
    将所述AI计算架构(232)的输出参数配置为所述第二参数。
  15. 根据权利要求1至14中任意一项所述的AI计算设备(20),其中,所述背板(21)进一步包括插槽,所述计算部件(23)作为可插拔扩展卡通过所述插槽连接到所述背板总线(211)。
  16. 根据权利要求1至15中任意一项所述的AI计算设备(20),其中,所述计算部件(23)进一步包括节能单元,所述节能单元用于,
    通过所述通信部件(22)接收所述控制器(40)发送的第一指令,根据所述第一指令进入低功耗模式;及
    通过所述通信部件(22)接收所述控制器(40)发送的第二指令,根据所述第二指令退出低功耗模式。
  17. 控制方法,应用于工业自动化系统中的控制器,包括:
    获取工业自动化系统中的数据;
    将所述数据发送给连接到现场总线的人工智能(AI)计算设备;
    通过所述现场总线接收所述AI计算设备对所述数据进行分析得到的分析结果,将所述分析结果提供给控制器中的决策装置,供所述决策装置生成用于自动化控制的控制指令。
  18. 根据权利要求17所述的控制方法,其中,获取所述工业自动化系统中的所述数据包括:
    从预设的第一配置信息中确定复数个生产参数;
    从所述工业自动化系统中的数据采集设备提供的数据中获取所述复数个生产参数的值作为所述数据。
  19. 根据权利要求18所述的控制方法,其中,从所述工业自动化系统中的数据采集设备提供的数据中获取所述复数个生产参数的值包括:
    每隔预设长度的时间段,从前一时间段内所述数据采集设备提供的数据中获取所述复数个生产参数的值作为所述数据。
  20. 根据权利要求17-19中任一权利要求所述的控制方法,其中,将所述数据通过所述工业自动化系统的现场总线发送给连接到所述现场总线的AI计算设备包括:
    从预设的第二配置信息中获取所述AI计算设备中的第一计算部件的信息;
    利用所述信息将所述数据发送给所述第一AI计算设备中的所述第一计算部件。
  21. 根据权利要求20所述的控制方法,其中,从预设的第二配置信息中获取所述AI计算设备中的第一计算部件的信息块包括:
    从所述第二配置信息中获取复数个计算部件的信息;
    从至少一个AI计算设备获得所述复数个计算部件的负载信息;
    根据所述负载信息从所述复数个计算部件中选择一个计算部件作为所述第一计算部件,将所述第一计算部件所属的AI计算设备确定为所述AI计算设备。
  22. 根据权利要求17-21中任一权利要求所述的控制方法,其中,通过所述现场总线接收所述AI计算设备利用所述数据对所述生产设备的生产过程进行分析得到的分析结果,包括:
    获得所述AI计算设备的输出数据,将所述输出数据转换为所述决策装置可识别的分析报告作为所述分析结果。
  23. 根据权利要求20所述的控制方法,进一步包括:
    从预设的第三配置信息中获取所述第一计算部件的配置参数,所述配置参数包括所述第一计算部件采用的AI计算架构的复数个属性值;
    将所述配置参数发送给所述第一计算部件,使所述第一计算部件将所述AI计算架构 的复数个属性的值设置为所述复数个属性值。
  24. 根据权利要求20所述的控制方法,进一步包括:
    从预设的第四配置信息中获取训练数据,将所述训练数据发送给所述第一计算部件,使所述第一计算部件利用所述训练数据对所述第一计算部件中的AI计算架构进行训练。
  25. 控制装置(41),应用于工业自动化系统中的控制器(40),所述控制装置(41)包括:
    生产数据获取单元(411),用于获取工业自动化系统中的数据;
    任务发送单元(412),将所述数据发送给连接到现场总线(60)的人工智能(AI)计算设备(20)中的第一计算部件(23);
    结果收集单元(413),用于通过所述现场总线(60)接收所述AI计算设备(20)发送的分析结果,所述分析结果由所述第一计算部件(23)对所述数据进行分析得到;将所述分析结果提供给控制器(40)中的决策装置(42),供所述决策装置(42)用于自动化控制的控制指令。
  26. 根据权利要求25所述的控制装置(41),进一步包括:
    配置单元(414),用于从预设的第三配置信息中获取所述第一计算部件(23)的配置参数,所述配置参数包括所述第一计算部件(23)的AI计算架构(231)的复数个属性值,将所述配置参数发送给所述第一计算部件(23),从而将所述AI计算架构(231)的复数个属性的值设置为所述复数个属性值。
  27. 根据权利要求25所述的控制装置(41),进一步包括:
    训练单元(415),用于从预设的第四配置信息中获取训练数据,将所述训练数据发送给所述第一计算部件(23),以使所述第一计算部件(23)利用所述训练数据对所述AI计算架构(231)进行训练。
  28. 工程师站(Engineer Station,ES)(30),连接到工业自动化系统的现场总线(60),所述工程师站(30)包括:处理器和存储器;所述存储器包括机器可读指令,所述指令可以由所述处理器执行用于:
    提供设备配置接口;
    从所述设备配置接口接收设备配置信息,所述设备配置信息包括人工智能(AI)计算设备(20)的标识和所述AI计算设备(20)中至少一个计算部件(23)的计算部件标识, 所述AI计算设备(20)通过现场总线接口连接到所述现场总线(60);将所述设备配置信息发送到所述工业自动化系统的控制器(40),以使所述控制器(40)利用所述设备配置信息与所述AI计算设备(20)进行通信;
    获取所述AI计算设备(20)对应的控制逻辑,从所述设备配置接口接收针对所述AI计算设备(20)的控制配置信息,所述控制配置信息包括所述至少一个计算部件中第一计算部件(23)的计算参数,将所述控制配置信息加载到所述控制逻辑中,将所述控制逻辑加载到所述控制器(40),所述控制逻辑用于使所述控制器(40)对所述第一计算部件(23)进行配置,将所述工业自动化网络中的数据发送给所述第一计算部件(23)进行分析,并获取所述第一计算部件(23)反馈的分析结果。
  29. 根据权利要求28所述的ES(30),其中,所述指令可以由所述处理器执行用于:
    从所述设备配置接口接收所述第一计算部件(23)的通道信息,所述通道信息包括至少一个通道的信息;
    将所述通道信息发送到所述控制器(40),使所述控制器(40)将数据通过所述数据的类型对应的通道发送给所述第一计算部件(23)。
  30. 根据权利要求28或29所述的ES(30),其中,所述指令可以由所述处理器执行用于以下中的至少一个:
    从所述设备配置接口接收所述第一计算部件(23)的架构信息,所述架构信息包括所述第一计算部件(23)中AI计算架构(231)的部件的结构属性值,将所述架构信息作为所述控制配置信息加载到所述控制逻辑中;
    从所述设备配置接口接收所述第一计算部件(23)的所述AI计算架构(231)的输入参数和输出参数,将所述输入参数和所述输出参数作为所述控制配置信息加载到所述控制逻辑中;
    从所述设备配置接口接收所述第一计算部件(23)的训练数据,将所述训练数据作为所述控制配置信息加载到所述控制逻辑中。
  31. 一种工业自动化系统,包括:工程师站(Engineer Station,ES)(30)、控制器(40)、生产设备和人工智能(AI)计算设备(20),以及连接各设备的现场总线(60);
    所述ES(30),用于
    提供设备配置接口,从所述设备配置接口接收配置信息,所述配置信息包括所述 AI计算设备(20)的信息,将所述配置信息加载到所述控制器;
    获取所述AI计算设备(20)对应的控制逻辑,将所述控制逻辑加载到所述控制器(40);
    所述控制器(40),用于
    执行所述控制逻辑,获取所述工业自动化系统中的复数个生产参数的值,将所述复数个生产参数的值发送给所述AI计算设备(20);接收所述AI计算设备(20)发送的分析结果;
    根据所述分析结果生成对所述生产设备的控制指令;
    所述AI计算设备(20),用于
    通过所述现场总线(60)接收所述控制器(40)发送的所述复数个生产参数的值,对所述复数个生产参数的值进行分析得到所述分析结果,将所述分析结果通过所述现场总线(60)发送给所述控制器(40)。
  32. 一种计算机可读存储介质,存储有机器可读指令,所述指令可以使处理器执行权利要求17-24中任一权利要求所述的方法。
PCT/CN2018/101973 2018-08-23 2018-08-23 人工智能计算设备、控制方法及装置、工程师站及工业自动化系统 WO2020037608A1 (zh)

Priority Applications (6)

Application Number Priority Date Filing Date Title
KR1020217008642A KR102566324B1 (ko) 2018-08-23 2018-08-23 인공 지능 컴퓨팅 디바이스, 제어 방법 및 장치, 엔지니어 스테이션, 및 산업 자동화 시스템
JP2021510083A JP2021536055A (ja) 2018-08-23 2018-08-23 人工知能計算装置、制御方法および制御装置、エンジニアステーションおよび産業オートメーションシステム
EP18930654.1A EP3835898A4 (en) 2018-08-23 2018-08-23 ARTIFICIAL INTELLIGENCE COMPUTING DEVICE, CONTROL METHOD AND DEVICE, ENGINEERING STATION AND INDUSTRIAL AUTOMATION SYSTEM
CN201880095710.3A CN112424713A (zh) 2018-08-23 2018-08-23 人工智能计算设备、控制方法及装置、工程师站及工业自动化系统
PCT/CN2018/101973 WO2020037608A1 (zh) 2018-08-23 2018-08-23 人工智能计算设备、控制方法及装置、工程师站及工业自动化系统
US17/270,098 US20210181695A1 (en) 2018-08-23 2018-08-23 Artificial intelligence computing device, control method and apparatus, engineer station, and industrial automation system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2018/101973 WO2020037608A1 (zh) 2018-08-23 2018-08-23 人工智能计算设备、控制方法及装置、工程师站及工业自动化系统

Publications (1)

Publication Number Publication Date
WO2020037608A1 true WO2020037608A1 (zh) 2020-02-27

Family

ID=69592207

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2018/101973 WO2020037608A1 (zh) 2018-08-23 2018-08-23 人工智能计算设备、控制方法及装置、工程师站及工业自动化系统

Country Status (6)

Country Link
US (1) US20210181695A1 (zh)
EP (1) EP3835898A4 (zh)
JP (1) JP2021536055A (zh)
KR (1) KR102566324B1 (zh)
CN (1) CN112424713A (zh)
WO (1) WO2020037608A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2022035998A (ja) * 2020-08-19 2022-03-04 ヒタチ・エナジー・スウィツァーランド・アクチェンゲゼルシャフト コントローラの意思決定論理の性能を監視するための方法および監視システム

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
BR112021010468A2 (pt) * 2018-12-31 2021-08-24 Intel Corporation Sistemas de segurança que empregam inteligência artificial
US20220121597A1 (en) * 2019-04-26 2022-04-21 Mitsubishi Electric Corporation Data processing device, data processing method, and program
DE102020102863A1 (de) * 2020-02-05 2021-08-05 Festo Se & Co. Kg Parametrierung einer Komponente in der Automatisierungsanlage
US20220187773A1 (en) * 2020-12-14 2022-06-16 Micron Technology, Inc. Managing energy using artificial intelligence
CN117136348A (zh) * 2021-05-25 2023-11-28 西门子股份公司 Ai推理设备和自动化控制器的数据交互方法、装置和系统
CN117312388B (zh) * 2023-10-08 2024-03-19 江苏泰赋星信息技术有限公司 一种人工智能模型控制系统

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2005029314A1 (en) * 2003-08-21 2005-03-31 Microsoft Corporation Storage platform for organizing, searching, and sharing data
US20110246427A1 (en) * 2010-03-31 2011-10-06 Sachin Modak Computer File Storage, Backup, Restore and Retrieval
CN102508456A (zh) * 2011-09-29 2012-06-20 东南大学 振动监视保护装置的背板总线式结构及其通信控制方法
CN103217935A (zh) * 2011-11-11 2013-07-24 洛克威尔自动控制技术股份有限公司 用于访问和传递数据的、集成且可扩展的架构
CN105243047A (zh) * 2015-09-09 2016-01-13 浪潮(北京)电子信息产业有限公司 一种服务器架构
CN105721546A (zh) * 2016-01-20 2016-06-29 冶金自动化研究设计院 一种基于工业物联通讯的多数据集成服务平台
CN107491044A (zh) * 2016-06-09 2017-12-19 罗克韦尔自动化技术公司 用于自动控制系统的可扩展分析架构

Family Cites Families (26)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5341496A (en) * 1990-08-29 1994-08-23 The Foxboro Company Apparatus and method for interfacing host computer and computer nodes using redundant gateway data lists of accessible computer node data
DE19933924A1 (de) * 1999-04-29 2000-11-02 Loher Ag Mikrosystem zur lokalen Zustandsüberwachung und Zustandsdiagnose von Maschinen, Anlagen und/oder Baugruppen, insbesondere von Antriebssystemen
DE10161669A1 (de) 2001-12-14 2003-06-26 Bosch Gmbh Robert Verfahren und Vorrichtung zum Aktivieren bzw. Deaktivieren verteilter Steuereinheiten
US7062580B2 (en) * 2002-09-20 2006-06-13 Smar Research Corporation Logic arrangement, system and method for configuration and control in fieldbus applications
CN1419170A (zh) * 2002-12-17 2003-05-21 白凤双 通用智能自动化系统
JP2008511938A (ja) * 2004-08-31 2008-04-17 ワットロー・エレクトリック・マニュファクチャリング・カンパニー 動作システムの分散された診断システム
US8055814B2 (en) * 2005-03-18 2011-11-08 Rockwell Automation Technologies, Inc. Universal safety I/O module
US7853677B2 (en) * 2005-09-12 2010-12-14 Rockwell Automation Technologies, Inc. Transparent bridging and routing in an industrial automation environment
US9411769B2 (en) * 2006-09-19 2016-08-09 Fisher-Rosemount Systems, Inc. Apparatus and methods to communicatively couple field devices to controllers in a process control system
US8832579B2 (en) * 2008-08-12 2014-09-09 Rockwell Automation Technologies, Inc. System for creation and management of industrial automation and information solutions and services
JP5339071B2 (ja) * 2009-04-02 2013-11-13 横河電機株式会社 制御装置及び通信管理方法
JP5549455B2 (ja) * 2010-07-21 2014-07-16 富士電機株式会社 プログラマブルコントローラシステムおよびそのプログラム更新方法
JP2012118715A (ja) * 2010-11-30 2012-06-21 Fuji Electric Co Ltd Plcシステム、その開発支援装置、プログラム
DE102011107321A1 (de) * 2011-07-06 2013-01-10 Abb Ag System und Verfahren zur Parametrierung von Feldgeräten eines Automatisierungs- oder Steuerungssystems
JP5911439B2 (ja) * 2013-01-28 2016-04-27 三菱電機株式会社 監視制御システム
CN105324728B (zh) * 2013-06-11 2019-06-07 西门子公司 用于监控自动化工厂的工业控制系统
JP6475469B2 (ja) * 2014-10-29 2019-02-27 株式会社日立製作所 診断ジョブ生成システム、診断ジョブ生成方法及び診断ジョブ生成表示方法
US9471452B2 (en) * 2014-12-01 2016-10-18 Uptake Technologies, Inc. Adaptive handling of operating data
JP6626313B2 (ja) * 2015-10-21 2019-12-25 株式会社キーエンス プログラマブル・ロジック・コントローラ、拡張ユニット、制御方法、プログラム作成支援装置、プログラム作成支援方法およびプログラム
WO2017168706A1 (ja) * 2016-03-31 2017-10-05 三菱電機株式会社 ユニット及び制御システム
JP6546566B2 (ja) * 2016-06-15 2019-07-17 日本電信電話株式会社 並列負荷分散システム、並列負荷分散方法、sdnコントローラホスト及びプログラム
JP6436148B2 (ja) * 2016-11-18 2018-12-12 横河電機株式会社 情報処理装置、保全機器、情報処理方法、情報処理プログラム及び記録媒体
JP6468313B2 (ja) * 2017-06-08 2019-02-13 オムロン株式会社 制御装置、制御プログラムおよび制御方法
JP6740920B2 (ja) * 2017-02-01 2020-08-19 株式会社デンソー 演算処理装置
EP3493000B1 (de) * 2017-12-04 2023-06-14 Siemens Aktiengesellschaft Verfahren zum fehlersicheren erfassen eines messwertes und automatisierungssystem
US11249469B2 (en) * 2018-09-28 2022-02-15 Rockwell Automation Technologies, Inc. Systems and methods for locally modeling a target variable

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2005029314A1 (en) * 2003-08-21 2005-03-31 Microsoft Corporation Storage platform for organizing, searching, and sharing data
US20110246427A1 (en) * 2010-03-31 2011-10-06 Sachin Modak Computer File Storage, Backup, Restore and Retrieval
CN102508456A (zh) * 2011-09-29 2012-06-20 东南大学 振动监视保护装置的背板总线式结构及其通信控制方法
CN103217935A (zh) * 2011-11-11 2013-07-24 洛克威尔自动控制技术股份有限公司 用于访问和传递数据的、集成且可扩展的架构
CN105243047A (zh) * 2015-09-09 2016-01-13 浪潮(北京)电子信息产业有限公司 一种服务器架构
CN105721546A (zh) * 2016-01-20 2016-06-29 冶金自动化研究设计院 一种基于工业物联通讯的多数据集成服务平台
CN107491044A (zh) * 2016-06-09 2017-12-19 罗克韦尔自动化技术公司 用于自动控制系统的可扩展分析架构

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2022035998A (ja) * 2020-08-19 2022-03-04 ヒタチ・エナジー・スウィツァーランド・アクチェンゲゼルシャフト コントローラの意思決定論理の性能を監視するための方法および監視システム
JP7257455B2 (ja) 2020-08-19 2023-04-13 ヒタチ・エナジー・スウィツァーランド・アクチェンゲゼルシャフト コントローラの意思決定論理の性能を監視するための方法および監視システム

Also Published As

Publication number Publication date
US20210181695A1 (en) 2021-06-17
KR102566324B1 (ko) 2023-08-14
CN112424713A (zh) 2021-02-26
EP3835898A1 (en) 2021-06-16
EP3835898A4 (en) 2022-07-27
KR20210045476A (ko) 2021-04-26
JP2021536055A (ja) 2021-12-23

Similar Documents

Publication Publication Date Title
WO2020037608A1 (zh) 人工智能计算设备、控制方法及装置、工程师站及工业自动化系统
US9529351B2 (en) System and method for parameterizing field devices of an automation or control system
WO2019109502A1 (zh) 一种基于物联网的数控机床智能监控网关
JP3993243B2 (ja) プロセス制御ネットワーク用のネットワークアクセス可能なインタフェース
CN111464339B (zh) 一种基于动态重构的异构工业网络互联方法及通用有线通信模块
CN112615935B (zh) 一种终端设备联网参考系统及其交互方法
CN101989085A (zh) 控制器
CN110663222B (zh) 处理过程数据
US20120330478A1 (en) Energy Saving at a Network Node in an Automation Network
US8630723B2 (en) Method for controlling behavioral intervention of a submodule
CN111061669A (zh) 一种工位监控方法、计算机可读存储介质和系统
CN100426167C (zh) 控制装置、控制模块、模块电池和控制系统
US9989940B2 (en) Method for engineering a distributed control system and an engineering tool thereof
CN108363368B (zh) 运行自动化系统的方法及自动化系统、现场设备和控制器
US10938967B2 (en) Method and apparatus for communication in a motor drive application
CN100420213C (zh) 车间局域网内分布式数控机床的实时监控系统
CN110300055B (zh) 异构现场总线用网关系统
TWI615692B (zh) 工廠監視控制系統用資料再生裝置
Xing et al. Motion control system using sercos over ethercat
CN109991924B (zh) 一种用于数控系统远程客户端实时监控的通讯方法及系统
CN102576222B (zh) 用于操作现场总线接口的方法
CN113589743A (zh) 一种现场总线系统
Fang et al. Development of multifunctional controller based on PROFIBUS-DP
Miclot Distribution of control via automation networks using the producer/consumer model
Liu et al. Design of PROFINET-based communication system for rice processing line

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 18930654

Country of ref document: EP

Kind code of ref document: A1

ENP Entry into the national phase

Ref document number: 2021510083

Country of ref document: JP

Kind code of ref document: A

NENP Non-entry into the national phase

Ref country code: DE

ENP Entry into the national phase

Ref document number: 2018930654

Country of ref document: EP

Effective date: 20210308

ENP Entry into the national phase

Ref document number: 20217008642

Country of ref document: KR

Kind code of ref document: A