An industrial intelligent control system and method

Abstract

The application discloses an industrial intelligent control system and method, relates to the technical field of industrial automatic control, and mainly solves the problems of low data cross-layer interaction efficiency, poor system global optimization, safety, expansion capability and function upgrade difficulty, and difficult fusion of AI control and traditional control in the existing industrial intelligent control scheme. The system comprises a basic control layer, an intelligent I / O layer and an intelligent control layer; the basic control layer comprises an expansion I / O cabinet, a common controller and a main control network A / B; the intelligent I / O layer comprises an intelligent I / O cabinet and an intelligent I / O switch; and the intelligent control layer comprises an intelligent controller, a credibility adjuster, a super-fusion module and an intelligent middle station. The system builds an intelligent collaborative architecture of end-edge-cloud, enhances the computing power of a DCS system, improves the development collaborative safety capability of a system intelligent algorithm, promotes the safe fusion of AI intelligent control and traditional control, and realizes smooth transition.

Description

Technical Field

[0001] This invention relates to the field of industrial automatic control technology, and in particular to an industrial intelligent control system and method. Background Technology

[0002] Industrial applications (such as thermal power generation, coal chemical industry, and metallurgy) commonly use DCS (Distributed Control Systems) to monitor and control production processes. Their traditional architecture typically employs a layered structure (field control layer and monitoring layer). With the development of the Industrial Internet and artificial intelligence technologies, intelligentization has become a crucial direction for improving industrial operational efficiency, safety, and flexibility.

[0003] Currently, industrial intelligent control solutions include: (1) optimization based on traditional DCS: adding optimization algorithm modules to the existing DCS architecture, such as using internal model control or intelligent algorithms (such as particle swarm optimization) to optimize controller parameters in order to improve the performance of specific control loops (such as coordinated control); (2) direct AI control: using independent AI controllers (such as controllers based on machine learning models) to directly output control commands, attempting to completely or partially replace traditional controllers. Although the above intelligent control solutions have their value to some extent, they still have problems such as low efficiency of cross-layer data interaction in rigid architecture, poor system global optimization, security, scalability and difficulty in function upgrades, and difficulty in integrating AI control and traditional control. Summary of the Invention

[0004] In view of this, the present invention provides an industrial intelligent control system and method, the main purpose of which is to solve the problems of low efficiency of cross-layer data interaction, poor system global optimization, security, scalability and difficulty in function upgrade, and difficulty in integrating AI control and traditional control in existing industrial intelligent control schemes.

[0005] According to one aspect of the present invention, an industrial intelligent control system is provided, comprising: a basic control layer, an intelligent I / O layer, and an intelligent control layer;

[0006] The basic control layer includes an expansion I / O cabinet, a general controller, and a main control network A / B; the expansion I / O cabinet is connected to the controller cabinet and is used to expand sensing devices and send the expanded sensing signals to the controller cabinet;

[0007] The intelligent I / O layer includes intelligent I / O cabinets and intelligent I / O switches; the intelligent I / O cabinets are used to convert field industrial signals into standard Ethernet signals and access them through the intelligent I / O switches; and access the intelligent control layer based on the intelligent I / O switches; the intelligent I / O switches connect the data from multiple intelligent I / O cabinets to the main control network A / B.

[0008] The intelligent control layer includes an intelligent controller, a reliability modifier, a hyperconverged module, and an intelligent middleware platform. The intelligent controller encapsulates machine learning algorithms to coordinate with the ordinary controllers in the basic control layer to process signal data that the ordinary controllers cannot handle. The reliability modifier receives the parallel outputs of the ordinary controllers and the intelligent controllers and dynamically determines control commands through fusion processing. The hyperconverged module integrates computing, storage, and network resources and pools these resources. The intelligent middleware platform dynamically and elastically allocates the pooled resources to multiple virtual machine services or containerized services to meet various service functions in the system.

[0009] Furthermore, the intelligent I / O layer also includes an intelligent I / O power supply module and an intelligent I / O dedicated network;

[0010] The intelligent I / O power supply module supplies power to the intelligent I / O cabinet through the intelligent I / O switch.

[0011] The dedicated intelligent I / O network modulates and encodes a traditional network to obtain an E / F network that integrates power supply and high-speed data communication, thereby meeting the power supply and high-speed data communication requirements of the intelligent I / O cabinet.

[0012] Furthermore, the virtual machine service includes:

[0013] Data platform services are used to store data and provide data support for the system.

[0014] The computing engine service is used to analyze and calculate data to determine the target information for optimal unit energy efficiency;

[0015] Time series model service, used to process time series data;

[0016] LLM model service for processing text data;

[0017] Industrial knowledge graph services are used to process industrial knowledge, including general knowledge and knowledge specific to each site.

[0018] Industrial digital modeling services establish digital models of industrial sites using a hybrid modeling approach that combines mechanistic and data-driven methods.

[0019] User behavior profiling service is used to record, analyze, and learn user operation records of intelligent control systems;

[0020] Application SDK services are used to provide development and runtime environments.

[0021] Furthermore, the containerized service includes:

[0022] The ordinary controllers in the basic control layer and the intelligent controllers in the intelligent control layer are containerized and saved as virtual controllers; the virtual controllers contain multiple controller instances;

[0023] The credibility modifier in the intelligent control layer is containerized and saved as a virtual credibility modifier.

[0024] The system connects to an industrial intelligent agent, which represents a collaborative system composed of multiple containerized intelligent agents. The intelligent agent includes an AI controller, an optimization algorithm module, a diagnostic service, and an alarm subsystem.

[0025] Furthermore, the credibility modifier establishes an intelligent control evaluation model encompassing four dimensions: interpretability, rationality, security, and performance, as shown in the following formula:

[0026]

[0027] in, This is the result of a credibility assessment; For explainability and credibility; For rationality and credibility; For security and credibility; For performance reliability; 'a' represents the instruction; 'L' represents the decision logic.

[0028] The reliability of the output of the intelligent controller is evaluated based on the intelligent control evaluation model to obtain a reliability evaluation result corresponding to the intelligent controller.

[0029] Furthermore, the aforementioned interpretability credibility The intelligent platform uses an LLM model service to analyze whether the intelligent controller simultaneously outputs instruction a and decision logic L.

[0030] The rationality and credibility The decision logic L is evaluated to determine whether it conforms to physical laws and industrial standards by calling the LLM model service and industrial knowledge graph service of the intelligent platform.

[0031] The security and trustworthiness The predictive model of the controlled object is obtained by calling the industrial digital modeling service of the intelligent platform, and the timing model service of the intelligent platform is used to verify whether the instruction a has exceeded the safety red line.

[0032] The performance reliability By quantifying the control performance of instruction 'a', control can be transferred to a controller with better performance.

[0033] Furthermore, the intelligent middleware platform includes a coordinator, which is used for lifecycle management, resource scheduling, load balancing, and fault tolerance of each of the virtual machine services and each of the containerized services.

[0034] Furthermore, the system also includes an intelligent human-machine interaction layer, which includes an industrial control computer and various industrial control software deployed on the industrial control computer, used to monitor the controlled process and the status of the control system; configure card algorithms, control algorithms and modify algorithm parameters; record process data and alarm data and provide retrieval services; and realize intelligent interaction of the system by calling the services of the intelligent middleware platform.

[0035] According to another aspect of the present invention, an industrial intelligent control method is provided, applied to the aforementioned industrial intelligent control system. The method includes a two-line coordinated control process, comprising:

[0036] The ordinary controller in the basic control layer receives field data and performs logical operations to obtain the first control strategy of the controlled object; the first control strategy includes instruction b.

[0037] The intelligent controller in the intelligent control layer receives field data and performs logical operations to obtain a second control strategy for the controlled object; the second control strategy includes instruction a and decision logic L; the logical operations are executed by the field controller or by the virtual controller in the intelligent platform.

[0038] The first control strategy and the second control strategy are transmitted to the credibility adjuster, which performs credibility evaluation on the first control strategy and the second control strategy, and determines the weights of the ordinary controller and the intelligent controller based on the credibility evaluation results.

[0039] Based on the weights, a comprehensive control strategy for the controlled object is determined, and the comprehensive control strategy is sent to the general controller, which then sends control commands to the field devices.

[0040] Furthermore, the method also includes an intelligent I / O control process, comprising:

[0041] The system uses an intelligent I / O cabinet installed on-site to collect industrial signals and converts these signals into standard Ethernet signals for connection to the intelligent I / O switch.

[0042] The intelligent I / O switch connects the data from multiple intelligent I / O cabinets to the system's main control network A / B, so that the data from the multiple intelligent I / O cabinets can be transmitted to the intelligent middleware in the system's intelligent control layer based on the main control network A / B.

[0043] The virtual controller instance in the intelligent platform receives data and performs calculations to obtain the control strategy for the controlled object.

[0044] The control strategy of the controlled object is sent to the intelligent I / O cabinet via the system's main control network A / B, intelligent I / O switch and E / F network, and the intelligent I / O cabinet sends control commands to the field devices.

[0045] By employing the above-described technical solutions, the technical solutions provided by the embodiments of the present invention have at least the following advantages:

[0046] This invention provides an industrial intelligent control system and method. Compared with existing technologies, this invention adds an intelligent I / O layer containing intelligent I / O cabinets and intelligent I / O switches, as well as an intelligent control layer containing intelligent controllers, reliability modifiers, and hyper-converged HCI systems, to the traditional DCS architecture. This highly optimizes the computing power network of the DCS, the intelligent algorithm development and operation platform, and the collaboration mode of the AI ​​controller. It realizes an intelligent collaborative architecture of end (basic control layer) - edge (intelligent controller, reliable modifier) ​​- cloud (intelligent middleware), decouples the DCS controller module and I / O module, adjusts the reliability of the AI ​​controller, and realizes a microservice architecture system of virtualized data services, model services, knowledge graph services, application SDK services, and virtual controllers in the intelligent middleware. This enhances the computing power of the DCS system, improves the development and collaborative security capabilities of the system's intelligent algorithms, and promotes the safe integration and smooth transition of AI intelligent control with traditional control.

[0047] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of the present invention more apparent and understandable, specific embodiments of the present invention are described below. Attached Figure Description

[0048] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:

[0049] Figure 1 This invention provides a schematic diagram of the structure of an industrial intelligent control system according to an embodiment of the present invention.

[0050] Figure 2 A flowchart illustrating the execution process of a credibility modifier provided in an embodiment of the present invention is shown.

[0051] Figure 3This diagram illustrates a process for evaluating the reliability of an intelligent controller according to an embodiment of the present invention.

[0052] Figure 4 A schematic diagram of the dual-line coordinated control process in an industrial intelligent control method provided by an embodiment of the present invention is shown.

[0053] Figure 5 This diagram illustrates the intelligent I / O control process in an industrial intelligent control method provided by an embodiment of the present invention. Detailed Implementation

[0054] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0055] This invention provides an industrial intelligent control system. The technical terms and definitions used in this invention are explained below:

[0056] DCS (Distributed Control System): A distributed control system is an automated control system widely used in industry. Its control functions are distributed among controllers in different locations and communicate and coordinate through a network.

[0057] Energy efficiency: Energy efficiency refers to the ratio of a system's effective output energy to its total input energy. In thermal power generation, it often refers to indicators such as power generation coal consumption rate and power supply coal consumption rate.

[0058] Closed-loop control: a control method in which the controller adjusts based on the deviation between the output feedback signal of the controlled object and the desired value, so that the system output automatically tracks the desired value.

[0059] Software-defined: A system design philosophy that emphasizes implementing and controlling hardware functions through software programming, thereby increasing the system's flexibility and scalability.

[0060] MMI (Man-Machine Interface): A human-machine interface, the interface through which operators and control systems interact.

[0061] A container is a lightweight, executable software package that contains everything needed to run a service: code, runtime, system tools, system libraries, and settings. It isolates software from its environment and ensures consistent performance across different computing environments.

[0062] Orchestrator: In a software-defined control system, this is a software tool or module responsible for the deployment, management, scaling, and networking of containers. It automatically manages the lifecycle of containerized applications.

[0063] Hyperconverged infrastructure (HCI): An IT infrastructure framework that integrates computing, storage, and networking capabilities, typically running on commercial servers and using software management to virtualize and pool all resources.

[0064] Industrial Agent: A software entity (usually containerized) with autonomy, social capabilities, responsiveness, and initiative that can interact and collaborate with other intelligent agents to accomplish specific tasks in an industrial environment.

[0065] like Figure 1 As shown, an industrial intelligent control system includes: a basic control layer, an intelligent I / O layer, and an intelligent control layer;

[0066] The basic control layer includes an expansion I / O cabinet, a general controller, and a main control network A / B; the expansion I / O cabinet is connected to the controller cabinet and is used to expand sensing devices and send the expanded sensing signals to the controller cabinet;

[0067] In embodiments of the present invention, such as Figure 1 As shown, an intelligent I / O layer and an intelligent control layer are added to the traditional DCS architecture (basic control layer). Furthermore, an expansion I / O cabinet is added to the traditional DCS architecture (basic control layer), connecting it to the controller cabinet where the ordinary controllers are located in the traditional DCS architecture. This expansion expands the sensing devices and sends the expanded sensing signals to the controller cabinet. The controller cabinet is located in the industrial production site or control electronics room, used to collect operational information (such as temperature, pressure, flow, etc.) from industrial production (such as thermal power generating units, coal chemical production lines), responsible for basic real-time logic control and signal processing, and executing control commands. It contains traditional ordinary controllers (such as DPUs), I / O modules, power supplies, etc., which are not specifically limited in this embodiment. The I / O modules use dedicated terminal blocks and traditional multi-core cables to connect to ordinary sensors (temperature sensors, pressure sensors, flow sensors, etc.) and actuators in the field.

[0068] It should be noted that the added expansion I / O cabinet does not include control functions, only I / O cards. It requires a separate power supply and is connected to the controller cabinet via I / O BUS cables to enable sensing and control of more sensors and equipment. It is typically located next to the control cabinet or in the industrial production site. The sensing devices may include video acquisition devices, sound acquisition devices, infrared sensors, etc., but this embodiment of the invention does not impose specific limitations.

[0069] The intelligent I / O layer includes intelligent I / O cabinets and intelligent I / O switches; the intelligent I / O cabinets are used to convert field industrial signals into standard Ethernet signals and access them through the intelligent I / O switches; and access the intelligent control layer based on the intelligent I / O switches; the intelligent I / O switches connect the data from multiple intelligent I / O cabinets to the main control network A / B.

[0070] In embodiments of the present invention, such as Figure 1 As shown, the newly added intelligent I / O layer of the system includes intelligent I / O cabinets and intelligent I / O switches. The intelligent I / O cabinets are responsible for converting field industrial signals into standard Ethernet signals, which are then connected to the intelligent I / O switches, and subsequently to the main control network A / B, communicating with the virtual controllers in the intelligent control layer. The intelligent I / O cabinets support programmable I / O modules, allowing for the definition of the type of each channel (e.g., AI / AO / DI / DO) for rapid deployment; they support the connection of field intelligent instruments (e.g., AUTBUS interfaces); and they support intrinsically safe design, allowing deployment in hazardous areas and meeting field protection standards. The intelligent I / O switches are primarily responsible for connecting data from multiple intelligent I / O cabinets to the main control network A / B.

[0071] It should be noted that the intelligent I / O layer also includes intelligent I / O power supply and a dedicated intelligent I / O network (E / F network). The intelligent I / O power supply is redundantly designed, providing power to the intelligent I / O cabinet through the intelligent I / O switch; no separate cabling is required, reducing deployment complexity. The dedicated intelligent I / O network (E / F network) modulates (e.g., Orthogonal Frequency Division Multiplexing (OFDM)) and encodes (e.g., based on Time-Sensitive Networking (TSN) and IPv6 protocol stack encoding techniques) traditional networks to create an integrated E / F network for power supply and high-speed data communication, meeting the power supply and high-speed data communication requirements of the intelligent I / O cabinet. Figure 1 As shown, the intelligent I / O layer also includes the G network, which uses the same modulation and coding technology as the E / F network and is mainly responsible for the access of field intelligent instruments.

[0072] The intelligent control layer includes an intelligent controller, a reliability modifier, a hyperconverged module, and an intelligent middleware platform. The intelligent controller encapsulates machine learning algorithms to coordinate with the ordinary controllers in the basic control layer to process signal data that the ordinary controllers cannot handle. The reliability modifier receives the parallel outputs of the ordinary controllers and the intelligent controllers and dynamically determines control commands through fusion processing. The hyperconverged module integrates computing, storage, and network resources and pools these resources. The intelligent middleware platform dynamically and elastically allocates the pooled resources to multiple virtual machine services or containerized services to meet various service functions in the system.

[0073] In this embodiment of the invention, the newly added intelligent control layer includes an intelligent controller, a reliability modifier, a hyper-converged module, and an intelligent middleware platform. The intelligent controller is an independent hardware device with a built-in dedicated processing chip (such as a CPU, GPU, NPU, or FPGA) and encapsulated advanced machine learning algorithms (such as deep learning and reinforcement learning models). It is used to coordinate with the ordinary controllers in the basic control layer to perform tasks such as condition prediction and optimized setpoint calculation. This intelligent controller is typically installed at the production site to collect and process signals that are difficult for traditional DCS to handle, such as video, vibration, and infrared images. This embodiment of the invention does not impose specific limitations. The execution process of the reliability modifier is as follows: Figure 2 As shown, it receives the parallel outputs of the ordinary controller in the traditional DCS architecture and the intelligent controller in the newly added intelligent control layer. Based on preset rules and the decisions of the multi-agent system, it dynamically calculates and outputs a weighted fusion final control command. For example, the command weight corresponding to the intelligent controller is 70%, and the command weight corresponding to the ordinary controller is 30%, etc. This embodiment of the invention does not impose specific limitations. In this embodiment, the weights can also be manually set by the engineer station or automatically adjusted by the system; this embodiment of the invention does not impose specific limitations. The hyper-converged infrastructure (HCI) integrates computing, storage, and network resources and pools these physical resources. The intelligent platform dynamically and elastically allocates the resources pooled by the HCI to various virtual machine services or containerized services to meet the basic platform needs of various functions of the intelligent control system.

[0074] The connection relationships and signal flow of the above system structure can be summarized as follows:

[0075] 1) Traditional controller cabinets acquire data from field devices and connect to the intelligent I / O layer, intelligent control layer, industrial computer and other DCS components via A / B network.

[0076] 2) The intelligent I / O cabinet obtains data from field devices, connects to the intelligent I / O switch via the E / F network, then connects to the A / B network, and connects to the virtual controller.

[0077] 3) The intelligent controller connects to the reliability regulator via the C / D network, while the traditional controller cabinet and industrial computer connect to the reliability regulator via the A / B network.

[0078] 4) The hyperconverged system connects and exchanges a large amount of data internally through the C / D network, accesses data from traditional controllers and intelligent I / O through the A / B network, and provides intelligent middleware services to the outside world through the C / D network.

[0079] 5) The credibility adjuster can be directly viewed and its weight set by the application and presentation layers.

[0080] 6) Containerized services and virtual machine services interact and collaborate internally and between each other through a microservice bus.

[0081] Furthermore, as a refinement and extension of the specific implementation methods described above, in order to achieve a unified, software-defined, and scalable intelligent middleware service method and system architecture through virtualization and microservice technologies, another industrial intelligent control system is provided, such as... Figure 1 As shown, the virtual machine service represents the various types of virtual machines running on the intelligent platform, including:

[0082] The data platform service is used to store data and provide data support for the system. The stored data includes time-series, relational, and other types of data from industrial sites, which support AI training, optimization calculations, and data analysis.

[0083] The computing engine service is used to analyze and calculate data to determine the target information for optimal unit energy efficiency. During data analysis and calculation, software such as data analysis, machine learning, or self-optimization algorithms can be deployed to analyze and calculate data in the database to determine the target information for optimal unit energy efficiency (such as optimal operating parameter values).

[0084] Time series model service, used to process time series data;

[0085] LLM model service is used to process text data, typically large language models;

[0086] Industrial knowledge graph services are used to process industrial knowledge, including general knowledge and knowledge specific to each site.

[0087] Industrial digital modeling services establish digital models of industrial sites using a hybrid modeling approach that combines mechanistic and data-driven methods.

[0088] User behavior profiling service is used to record, analyze, and learn user operation records of intelligent control systems;

[0089] Application SDK services are used to provide development and runtime environments to continuously expand the functionality of intelligent systems.

[0090] like Figure 1 As shown, the containerized service in this embodiment represents various containers running on the intelligent middleware platform, including:

[0091] The ordinary controllers in the basic control layer and the intelligent controllers in the intelligent control layer are containerized and saved as virtual controllers. Each virtual controller contains multiple controller instances. In this embodiment, these controllers no longer depend on specific hardware; their functions are software-defined and can be dynamically created, migrated, redundantly configured, and destroyed on the hardware resources of the computing platform. Different instances can define different functions. For example, instance 1 can connect and control an intelligent I / O cabinet; instance 2 deploys encapsulated complex algorithm modules such as energy efficiency calculation and power consumption analysis, and the energy efficiency information of the computer group; instance 3 deploys intelligent algorithms such as AI inference engines and machine learning models to perform advanced intelligent control tasks such as predictive maintenance, intelligent optimization, and image recognition. This embodiment of the invention does not impose specific limitations.

[0092] The credibility modifier in the intelligent control layer is containerized and saved as a virtual credibility modifier.

[0093] An industrial intelligent agent is introduced, representing a collaborative system composed of multiple containerized intelligent agents. This intelligent agent includes an AI controller, an optimization algorithm module, a diagnostic service, and an alarm subsystem. In this embodiment, the intelligent agents can communicate and collaborate via a microservice bus to jointly complete complex industrial intelligent tasks, forming a multi-agent collaborative decision-making and control architecture.

[0094] In addition, this embodiment also configures a dedicated high-speed data network (C / D network) for the intelligent control layer, which adopts the same modulation and coding technology as the E / F network. It is used to connect various components in the intelligent control layer (such as database, computing platform, AI controller), carry a large amount of data interaction traffic, and is isolated from the A / B network to ensure the performance of the real-time control network.

[0095] It should be noted that this embodiment uses container technology (such as Docker) to implement software-defined functionality. Equivalent alternatives may include virtual machines (VMs), unikernels, and other application encapsulation and isolation technologies, as long as they achieve the effects of functional isolation, resource allocation, and rapid deployment.

[0096] Furthermore, as a refinement and extension of the specific implementation methods described above, another industrial intelligent control system is provided to improve the overall system performance, such as... Figure 1As shown, a coordinator is set up in the intelligent platform as the core management unit, responsible for the lifecycle management, resource scheduling, load balancing, and fault tolerance of all virtual machines and containerized services. The coordinator monitors hardware resource utilization and container performance, dynamically instantiating, replicating, or migrating containers between different computing nodes to ensure service continuity and optimal performance. For example, the coordinator continuously monitors the HCI system status. When a container or virtual machine is overloaded or its host node fails, the coordinator switches the active instance in its replicas or starts a new container instance on a node with sufficient resources, achieving seamless failover and load balancing. In addition, the coordinator can also achieve intelligent collaboration, enabling multiple virtual controllers or industrial agents to work collaboratively. For example, one may be responsible for load forecasting, while another is responsible for combustion optimization. The coordinator ensures that the resources required by them are guaranteed. This embodiment of the invention does not impose specific limitations.

[0097] It should be noted that in this embodiment, the coordinator can also be a different container orchestration platform (such as Kubernetes) or a customized version thereof, as long as it has similar scheduling, management and fault tolerance capabilities.

[0098] In this embodiment, the system further includes an intelligent human-machine interaction layer, which includes an industrial control computer and various industrial control software deployed on it. This layer is used to monitor the controlled process and the status of the control system; configure card algorithms, control algorithms, and modify algorithm parameters; record process data and alarm data and provide retrieval services; and achieve intelligent interaction of the system by calling the services of the intelligent platform. In this embodiment, the industrial control computer can also be divided according to roles, such as engineer workstations, history workstations, operator workstations, etc., but this embodiment of the invention does not impose specific limitations.

[0099] Furthermore, as a refinement and extension of the specific implementation methods described above, in order to promote the safe integration and smooth transition of AI intelligent control with traditional controllers, and to improve the reliability of control, another industrial intelligent control system is provided, such as... Figure 3 As shown, the system's credibility regulator establishes an intelligent control evaluation model encompassing four dimensions: interpretability, rationality, safety, and performance. It implements a "step-by-step screening" logic through sequential evaluation, as illustrated in the following formula:

[0100]

[0101] in, This is the result of a credibility assessment; For interpretability credibility, it is used to check the existence of logic; For the sake of reasonableness and credibility, it is used to assess logical compliance; For security and trustworthiness, it is used to verify security boundaries; For performance reliability, used to evaluate control efficiency; 'a' represents the instruction; 'L' represents the decision logic.

[0102] The reliability of the output of the intelligent controller is evaluated based on the intelligent control evaluation model to obtain a reliability evaluation result corresponding to the intelligent controller.

[0103] In this embodiment of the invention, the above-mentioned interpretability credibility By calling the LLM model service of the intelligent platform, we can analyze whether the intelligent controller simultaneously outputs the instruction 'a' and the decision logic 'L'; the following results are obtained:

[0104]

[0105] If interpretability credibility If the value is 1, the intelligent controller output is considered compliant, and the following evaluation will continue.

[0106] If interpretability credibility If the value is 0, the intelligent controller output is considered non-compliant, and the following evaluation will not be conducted.

[0107] In the embodiments of the present invention, the above-mentioned rationality and credibility The decision logic L is evaluated for compliance with physical laws and industrial standards by invoking the LLM model service and industrial knowledge graph service of the intelligent platform; it satisfies the rule that "illogical logic reduces credibility". An illogical logic database is constructed ( Implement compliance checks – database of unreasonable logic ( Defined by domain experts, it includes all known logic types such as "physical contradictions, irrelevant features, and dangerous operations," and the model performs real-time checks. Whether or not The matching is shown in the following formula:

[0108]

[0109] in, Penalty value (initial value) (This can be adjusted by experts) is used to reduce the credibility of illogical logic, but retains a small amount of weight (to avoid extreme switching).

[0110] In this embodiment of the invention, Example of an illogical database:

[0111] (1) Physical contradiction rule: IF temp>110℃ THEN increase_heater = 1.0 (Heating is still performed when the temperature is too high, which violates the law of thermal equilibrium);

[0112] (2) Irrelevant feature rule: IF pressure_sensor_status = OK THEN valve_opening = 0.5 (the sensor status and valve opening are not causally related);

[0113] (3) Dangerous operation rule: IF flow<10 m³ / h THEN close_valve = 1.0 (closing the valve when the flow rate is too low can easily lead to negative pressure in the pipeline).

[0114] Penalty value (initial value) (This can be adjusted by experts) is used to reduce the credibility of illogical logic, but retains a small amount of weight (to avoid extreme switching).

[0115] In this embodiment of the invention, the above-mentioned security reliability The system obtains a predictive model of the controlled object by calling the industrial digital modeling service of the intelligent platform, and verifies whether the instruction a exceeds the safety red line by calling the timing model service of the intelligent platform; thus realizing "one-vote veto for unsafe actions" and ensuring that the system operates within the safety boundary.

[0116] In this embodiment, the intelligent middleware timing model service can be based on the robustness of Signal Timing Logic (STL). Quantifying safety margins—STL is the standard language used in industry to define safety rules (such as "temperature always below 120°C"), robustness. Reflecting "action" In state The degree to which the STL rules are satisfied is normalized to the sigmoid function. Interval.

[0117]

[0118] in, For STL robustness; Explain the action Meet safety rules The larger the value, the higher the safety margin (e.g., "temperature 110℃, rule is temp < 120℃"). = (”); Explain the action Violation of safety rules The smaller the value, the higher the level of danger (e.g., "temperature 125℃, rule is temp < 120℃"). In this embodiment, common STL security rule examples are as follows: ("From now until the next T milliseconds, the temperature will remain below 120°C", symbol "") This means that within the specified time period, the subsequent conditions must be true at every moment; there can be no moment when they are violated. (“The pressure will drop below 4.5 MPa at least once within the next 50 milliseconds,” symbol “ "This means that the condition must be true at least once within the specified time period; it does not need to be true all the time."

[0119] in, Safety gain factor (recommended) ,like ),make sure When it approaches 0 (near the safety boundary). The value drops precipitously to 0, achieving a "one-vote veto." In this embodiment, the Sigmoid function's role is to... The infinite range of values ​​( Normalization to This meets the quantitative requirements for credibility.

[0120] In this embodiment of the invention, the above-mentioned performance reliability By quantifying the control performance of instruction 'a', control can be transferred to a controller with superior performance. Control performance includes factors such as economic efficiency and tracking accuracy, which are not specifically limited in this embodiment of the invention. For example, an object prediction model Q-value function can be used. (Reaction action) The expected cumulative future revenue, based on the "performance baseline of the common controller" ( Based on ), normalized to using the Sigmoid function. The interval. The core formula is shown below:

[0121]

[0122] in, For state Next action The expected cumulative future revenue (from offline secure RL training in Phase 1) directly reflects the performance quality—the higher the Q value, the better the long-term economic benefits of the action (e.g., "higher output and lower energy consumption"). For ordinary controllers The performance baseline is a constant pre-calculated for the offline phase—through The average Q value on historical safety datasets (e.g., "average Q value of PID controller = 50") is used as a benchmark for performance comparison. This is a performance gain coefficient (adjusted according to the scenario), which regulates the sensitivity of performance differences to credibility. The larger the value, the greater the difference for the same Q value. The more significant the change.

[0123] It should be noted that in this embodiment, the reliability modifier can also use an application-specific integrated circuit (ASIC) or FPGA to implement its core algorithm in order to pursue the ultimate real-time performance and reliability. It can also be embedded as a software module into existing controllers or host computers to reduce costs and adapt to different application scenarios. It only needs to be able to complete the weight calculation and fusion output of multiple input instructions.

[0124] This invention provides an industrial intelligent control system. Compared with existing technologies, this invention adds an intelligent I / O layer containing intelligent I / O cabinets and intelligent I / O switches, as well as an intelligent control layer containing intelligent controllers, reliability regulators, and hyper-converged HCI systems, to the traditional DCS architecture. This highly optimizes the computing power network of the DCS, the intelligent algorithm development and operation platform, and the collaboration mode of the AI ​​controller. It realizes an intelligent collaborative architecture of end (basic control layer) - edge (intelligent controller, reliable regulator) - cloud (intelligent middleware), decouples the DCS controller module and I / O module, regulates the reliability of the AI ​​controller, and realizes a microservice architecture system of virtualized data services, model services, knowledge graph services, application SDK services, and virtual controllers in the intelligent middleware. This enhances the computing power of the DCS system, improves the development and collaborative security capabilities of the system's intelligent algorithms, and promotes the safe integration and smooth transition of AI intelligent control with traditional control.

[0125] As a response to the above Figure 1 The implementation of the system shown in this invention provides an industrial intelligent control method, such as... Figure 4 As shown, this method includes a two-line coordinated control process, the specific process of which is as follows:

[0126] 101. The ordinary controller in the basic control layer receives field data and performs logical operations to obtain the first control strategy of the controlled object; the first control strategy includes instruction b;

[0127] 102. The intelligent controller in the intelligent control layer receives field data and performs logical operations to obtain a second control strategy for the controlled object; the second control strategy includes instruction a and decision logic L; the logical operations are executed by the field controller or the virtual controller in the intelligent platform.

[0128] 103. The first control strategy and the second control strategy are transmitted to the credibility adjuster, which performs credibility evaluation on the first control strategy and the second control strategy, and determines the weights of the ordinary controller and the intelligent controller based on the credibility evaluation results.

[0129] 104. Determine the comprehensive control strategy for the controlled object based on the weights, and send the comprehensive control strategy to the general controller, which then sends control commands to the field equipment.

[0130] In this embodiment, the above steps are for the reasonable operation of the intelligent collaborative architecture of end (basic control layer) - edge (intelligent controller, trusted regulator) - cloud (intelligent middleware), and gradually realize the safe integration and smooth transition of AI control and traditional control through the trust assessment method.

[0131] It should be noted that scenarios without the participation of a traditional intelligent controller can also be implemented using the system of this invention. The specific process is as follows:

[0132] The control cabinet collects operational data → via A / B network → stored in the intelligent control layer / intelligent middle platform / data middle platform service → the intelligent control layer's computing engine service accesses the data through the internal bus to perform optimization calculations → the calculation results (target information) are sent to the traditional controller via A / B network → the traditional controller calculates and outputs according to the target → the final output is sent to the field equipment via the traditional controller, forming a closed-loop control.

[0133] Furthermore, as a refinement and extension of the specific implementation of the above embodiments, in order to achieve decoupling between the DCS controller module and the I / O module, another industrial intelligent control method is provided, such as... Figure 5 As shown, this method includes an intelligent I / O control process, the specific process of which is as follows:

[0134] 201. Use an intelligent I / O cabinet installed on site to collect on-site industrial signals; and convert the on-site industrial signals into standard Ethernet signals to access the intelligent I / O switch;

[0135] 202. The intelligent I / O switch connects the data of multiple intelligent I / O cabinets to the main control network A / B of the system, so that the data of multiple intelligent I / O cabinets can be transmitted to the intelligent middleware in the intelligent control layer of the system based on the main control network A / B.

[0136] 203. The virtual controller instance in the intelligent platform receives data and performs calculations to obtain the control strategy for the controlled object;

[0137] 204. The control strategy of the controlled object is sent to the intelligent I / O cabinet via the system's main control network A / B, intelligent I / O switch and E / F network, and the intelligent I / O cabinet sends control commands to the field devices.

[0138] This invention provides an industrial intelligent control method. Compared with existing technologies, this invention adds an intelligent I / O layer containing intelligent I / O cabinets and intelligent I / O switches, as well as an intelligent control layer containing intelligent controllers, reliability modifiers, and hyper-converged HCI systems, to the traditional DCS architecture. This highly optimizes the computing power network of the DCS, the intelligent algorithm development and operation platform, and the collaboration mode of the AI ​​controller. It realizes an intelligent collaborative architecture of end (basic control layer) - edge (intelligent controller, reliable modifier) ​​- cloud (intelligent middleware), decouples the DCS controller module and I / O module, adjusts the reliability of the AI ​​controller, and realizes a microservice architecture system of virtualized data services, model services, knowledge graph services, application SDK services, and virtual controllers in the intelligent middleware. This enhances the computing power of the DCS system, improves the development and collaborative security capabilities of the system's intelligent algorithms, and promotes the safe integration and smooth transition of AI intelligent control with traditional control.

[0139] It is obvious to those skilled in the art that the modules or steps of the present invention described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. Optionally, they can be implemented using computer-executable program code, thereby storing them in a storage device for execution by a computing device. In some cases, the steps shown or described can be performed in a different order than those presented herein, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, the present invention is not limited to any particular combination of hardware and software.

[0140] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. An industrial intelligent control system, characterized in that, The application relates to a control system, which comprises a basic control layer, an intelligent I / O layer and an intelligent control layer. The basic control layer comprises an extended I / O cabinet, a common controller and a main control network A / B; the extended I / O cabinet is connected with a controller cabinet and is used for extending a sensing device and sending an extended sensing signal to the controller cabinet. The intelligent I / O layer comprises an intelligent I / O cabinet and an intelligent I / O switch; the intelligent I / O cabinet is used for converting field industrial signals into standard Ethernet signals and accessing the intelligent I / O switch; the intelligent I / O switch accesses the intelligent control layer based on the intelligent I / O switch; and the intelligent I / O switch accesses data of a plurality of intelligent I / O cabinets to the main control network A / B. The intelligent control layer comprises an intelligent controller, a credibility adjuster, a super-fusion module and an intelligent middle station; the intelligent controller is encapsulated with a machine learning algorithm and is used for cooperating with a common controller in the basic control layer to process signal data that cannot be processed by the common controller; the credibility adjuster receives parallel outputs of the common controller and the intelligent controller and dynamically determines a control instruction through fusion processing; the super-fusion module is used for integrating computing, storage and network resources and pooling the resources; and the intelligent middle station is used for dynamically and elastically allocating the pooled resources to a plurality of virtual machine services or containerized services to meet various service functions in the system. The credibility adjuster establishes an intelligent control evaluation model comprising four dimensions of explainability, rationality, safety and performance, and the formula is as follows: The intelligent control evaluation model is used for credibility evaluation processing of outputs of the intelligent controller to obtain a credibility evaluation result corresponding to the intelligent controller. wherein, is a result of the trustworthiness evaluation; is an explainability trustworthiness; is a reasonableness trustworthiness; is a security trustworthiness; is a performance trustworthiness; a is an instruction; L is a decision logic; The intelligent I / O layer further comprises an intelligent I / O power supply module and an intelligent I / O special network. The explainability credibility analyzing, by the LLM model service of the intelligent middle station, whether the intelligent controller simultaneously outputs the instruction a and the decision logic L; The rationality credibility The decision logic L is evaluated whether it conforms to physical laws and industrial specifications by calling the LLM model service and the industrial knowledge graph service of the intelligent middle platform. The security credibility A prediction model of the controlled object is obtained by calling an industrial digital modeling service of the intelligent middle platform, and whether the instruction a breaks a safety red line is verified by calling a timing model service of the intelligent middle platform. the performance credibility The control performance of the instructions a is quantified so that the control is migrated to the controller with better performance.

2. The system of claim 1, wherein, The intelligent I / O power supply module provides power supply for the intelligent I / O cabinet through the intelligent I / O switch. The intelligent I / O special network obtains an E / F network used for power supply and high-speed data communication integration through modulation and coding of a traditional network to meet power supply and high-speed data communication requirements of the intelligent I / O cabinet. The virtual machine service comprises a data middle station service used for storing data and providing data support for the system, a computing engine service used for analyzing and calculating data to determine target information of optimal unit energy efficiency, a time sequence model service used for processing time sequence data, an LLM model service used for processing text data, an industrial knowledge graph service used for processing industrial knowledge, including general knowledge and knowledge specific to each field, an industrial digital modeling service used for establishing an industrial field digital model through a hybrid modeling method of mechanism and data driving, a user behavior portrait service used for recording, analyzing and learning operation records of a user on the intelligent control system and an application SDK service used for providing a development and running environment.

3. The system of claim 1, wherein, The containerized service comprises containerization processing of the common controller in the basic control layer and the intelligent controller in the intelligent control layer and saving as a virtual controller; and the virtual controller comprises a plurality of controller instances. ​ ​ ​ ​ ​ ​ ​ ​ 4. The system of claim 1, wherein, ​ ​ Containerize the trustworthiness adjuster in the intelligent control layer and save as a virtual trustworthiness adjuster; Access an industrial intelligent agent, the industrial intelligent agent characterizes a collaborative system composed of multiple containerized intelligent agents, and the intelligent agents include an AI controller, an optimization algorithm module, a diagnosis service, and an alarm subsystem.

5. The system of any one of claims 1 to 4, wherein, The intelligent middle station includes a coordinator, which is configured to perform lifecycle management, resource scheduling, load balancing, and fault tolerance on each virtual machine service and each containerized service.

6. The system of claim 5, wherein, The system further includes an intelligent human-machine interaction layer, which includes an industrial computer and various industrial software deployed on the industrial computer, is configured to monitor the state of a controlled process and a control system, configure card algorithm, control algorithm, and modify algorithm parameters, record process data and alarm data, and provide retrieval services, and realize intelligent interaction of the system by calling services of the intelligent middle station.

7. An industrial intelligent control method, characterized by, The method is applied to the industrial intelligent control system of claim 1 and includes a double-line coordinated control process, which includes: The ordinary controller in the basic control layer receives field data and performs logical operation to obtain a first control strategy of a controlled object; the first control strategy includes an instruction b; The intelligent controller in the intelligent control layer receives field data and performs logical operation to obtain a second control strategy of the controlled object; the second control strategy includes an instruction a and a decision logic L; the logical operation is performed by a field controller or a virtual controller in the intelligent middle station; The first control strategy and the second control strategy are transmitted to a trustworthiness adjuster, the trustworthiness adjuster performs trustworthiness evaluation on the first control strategy and the second control strategy, and determines the weights of the ordinary controller and the intelligent controller based on the trustworthiness evaluation result; Based on the weights, a comprehensive control strategy of the controlled object is determined, and the comprehensive control strategy is sent to the ordinary controller, which sends a control instruction to a field device.

8. The method of claim 7, wherein, The method further includes an intelligent I / O control process, which includes: An intelligent I / O cabinet installed in the field is used to collect field industrial signals; and the field industrial signals are converted into standard Ethernet signals and connected to an intelligent I / O switch; The intelligent I / O switch connects data of multiple intelligent I / O cabinets to a main control network A / B of the system, so that the data of the multiple intelligent I / O cabinets are transmitted to an intelligent middle station in the intelligent control layer of the system based on the main control network A / B; A virtual controller instance in the intelligent middle station receives data and performs calculation to obtain a control strategy of a controlled object; The control strategy of the controlled object is sent to an intelligent I / O cabinet through the main control network A / B, the intelligent I / O switch, and an E / F network of the system, and a control instruction is sent to a field device by the intelligent I / O cabinet.

Images