A hierarchical decoupling and elastic deployment method and system of a distributed system

Abstract

The application provides a layered decoupling and elastic deployment method and system of a distributed system, which decouples a monomer control application program into multiple logical function blocks that can be independently deployed and completes coarse distribution at a station level, generates a target deployment mapping table of the logical function blocks by a logical scheduling unit of each field control station based on a topology computing power parameter of an intelligent remote processing unit, adopts a layered progressive optimization algorithm, parallelly bears control logic execution by the distributed intelligent remote processing unit, completes cross-node data collaboration through a deterministic real-time industrial network, and simultaneously realizes global monitoring of a running state, system-level control coordination, and dynamic deployment adjustment of working condition adaptation by the logical scheduling unit, thereby achieving technical effects of horizontally elastic expansion of system computing power, natural isolation of faults to improve overall reliability, and global optimization of resources to reduce communication loss.

Description

Technical Field

[0001] This application relates to the field of industrial automation control technology, and in particular to a method and system for hierarchical decoupling and flexible deployment of distributed systems. Background Technology

[0002] Traditional distributed control systems (DCS) generally adopt a centralized logic execution architecture. The system consists of an upper-level management node, a main controller, and lower-level I / O modules. The main controller acts as a single control brain, centrally executing all control strategies. The I / O modules only have signal acquisition and conversion functions and do not have complex logic operation capabilities. The system executes a closed-loop process of "input scanning, centralized operation, and output refresh" in a fixed cycle, with the core guarantee being timing determinism rather than the dynamic elasticity of computing resources.

[0003] As industrial processes become larger, more complex, and more sophisticated, this centralized architecture has gradually revealed the following core flaws: The system's total processing capacity is physically fixed during the selection of the main controller hardware, making it impossible to flexibly expand with process upgrades. Upgrades require downtime for hardware replacement. The main controller is a single point of failure; even with redundant switching mechanisms, there is still a switching risk window, and the overall system reliability is limited by the highest reliability of a single hardware unit. Control logic is statically bound to a designated controller during the engineering configuration phase, making it impossible to migrate or adjust during operation. Fluctuations in operating conditions easily lead to uneven workloads, with some controllers overloaded while others are idle, resulting in low global resource utilization efficiency and an inability to flexibly optimize based on real-time operating conditions. The roles of the main controller and I / O units are strictly fixed; ordinary I / O lacks programmability, and intelligent I / O functions are mostly hardware-based. Complex control strategies must be centrally processed by the main controller, easily causing communication bus surges and computational congestion. Process changes and algorithm upgrades require overall adjustments to the controller program, resulting in long cycles, high risks, and poor flexibility. Summary of the Invention

[0004] In view of this, the purpose of this application is to provide a hierarchical decoupling and elastic deployment method and system for distributed systems. By decoupling the monolithic control application into multiple independently deployable logical functional blocks and completing station-level coarse allocation, the logical scheduling unit of each field control station generates a target deployment mapping table for the logical functional blocks based on the topology computing power parameters of the intelligent remote processing unit using a hierarchical progressive optimization algorithm. The distributed intelligent remote processing units carry out the parallel execution of control logic and complete cross-node data collaboration through a deterministic real-time industrial network. At the same time, the logical scheduling unit realizes global monitoring of the operating status, system-level control coordination, and dynamic deployment adjustment for adapting to operating conditions. This solves the problems of performance ceiling and single point of failure risk in the static coupling of computing resources in traditional centralized architecture, low utilization rate due to uneven resource availability caused by rigid logical deployment, and high system reconstruction cost and insufficient flexibility due to hardware function specialization. It achieves the technical effects of horizontal elastic expansion of system computing power, natural fault isolation to improve overall reliability, and global optimization of resource configuration to reduce communication loss.

[0005] This application provides a method for layered decoupling and elastic deployment of a distributed system, wherein the distributed system includes an engineering workstation, a deterministic real-time industrial network, and multiple field control stations; wherein each field control station is equipped with a logical scheduling unit and multiple intelligent remote processing units, and the method includes: In response to the engineer station receiving a single-unit control application, the engineer station's logic decoupling engine is used to decompose the single-unit control application into multiple logic function blocks, determine the remote processing unit range information corresponding to each logic function block, and based on the remote processing unit range information, allocate the logic function blocks to the corresponding field control stations to obtain a set of logic function blocks for each field control station. Within each of the field control stations, the logical scheduling unit generates a target deployment mapping table from each logical function block in the set of logical function blocks to the intelligent remote processing unit based on the topology computing power parameters of each of the intelligent remote processing units using a hierarchical progressive optimization algorithm. Each of the intelligent remote processing units executes the locally carried logical function blocks in parallel within each control cycle based on the target deployment mapping table, and completes cross-node data exchange through the deterministic real-time industrial network to realize the control closed loop of the distributed system. The logical scheduling unit acquires the operating status data of the intelligent remote processing unit and the control commands triggered by the engineer station or operator station in real time. Based on the operating status data and the control commands, it performs global control coordination on the field control station and dynamically adjusts the target deployment mapping table according to the changes in the operating conditions of the field control station.

[0006] Furthermore, the step of using the logic decoupling engine of the engineering station to decompose the monolithic control application into multiple logical function blocks and determining the remote processing unit range information corresponding to each logical function block includes: Using the logic decoupling engine of the engineering station, the monolithic control application is divided into multiple coarse-grained logic function blocks with the control logic function integrity of the monolithic control application as the decomposition granularity, and the data flow dependency relationship between each coarse-grained logic function block is determined. Based on the data flow dependency, the coarse-grained logic function blocks are initially divided according to the code size of the coarse-grained logic function blocks to obtain multiple preliminary subdivided logic function blocks, and the estimated running time of each preliminary subdivided logic function block is obtained through simulation. With the goal of maintaining a balance in the estimated running time of each of the initially subdivided logical function blocks, the initially subdivided logical function blocks are iteratively split and merged to obtain multiple logical function blocks with balanced granularity; wherein, each of the logical function blocks has clear input and output interfaces, execution cycle, computing power requirements and data flow dependency attributes; Based on the physical affiliation of the input / output channels associated with each logical function block, a three-level deployable resource pool is defined for each logical function block in descending order of communication cost, and the remote processing unit range information corresponding to each logical function block in the three-level deployable resource pool is determined. The first-level resource pool comprises the intelligent remote processing units within the same field control station, including associated input / output channels; the second-level resource pool comprises the remaining intelligent remote processing units within the same field control station; and the third-level resource pool comprises the intelligent remote processing units of the remaining field control stations. The deployment priority of the first-level resource pool is higher than that of the second-level resource pool, and the deployment priority of the second-level resource pool is higher than that of the third-level resource pool.

[0007] Furthermore, based on the remote processing unit range information, the logical function blocks are allocated to the corresponding field control stations to obtain a set of logical function blocks for each field control station, including: Based on the remote processing unit range information, the computing power weight and input / output home station attribute information of each logical function block, as well as the data flow dependency and logical correlation between the logical function blocks, are determined. An undirected weighted dependency graph is constructed with each of the logical functional blocks as a vertex with the computing power weight and the input / output home station attribute information, the data flow dependency as an edge, and the logical correlation degree as the edge weight. The edges are traversed in descending order of edge weight, and strongly correlated vertices are merged into supernodes. Supernodes with clearly defined input / output affiliations are preferentially assigned to their corresponding field control stations in descending order of total computing power. Cross-station supernodes are assigned to the field control station corresponding to the first load. If the computing power of the affiliation field control station is insufficient, the supernode is assigned to the field control station adjacent to that field control station. Under the pre-set rigid constraints, the improved Phidusia and Matthews method is used to iteratively migrate the supernodes of every two field control stations until no migration is gained or the number of iterations reaches a threshold, so as to obtain the set of logical function blocks of each field control station.

[0008] Furthermore, within each of the field control stations, the logical scheduling unit generates a target deployment mapping table from each logical function block in the logical function block set to the intelligent remote processing unit based on the topology computing power parameters of each intelligent remote processing unit, using a hierarchical progressive optimization algorithm, including: For each of the field control stations, decision variables and rigid hard constraints are defined; wherein, the decision variables include the deployment marker variable and the same deployment association variable of the logical function block; the rigid hard constraints include that each logical function block is deployed only in one intelligent remote processing unit, and the target computing power of a single intelligent remote processing unit does not exceed the upper limit of computing power after reserving the safety margin, and the logical function block is deployed only within the resource pool corresponding to the remote processing unit range information, and the same deployment association variable is consistent with the deployment marker variable; Based on the topology computing power parameters of each intelligent remote processing unit, under the rigid hard constraint, the decision variables are optimized in descending order of the preset hierarchical progressive optimization priority, and the objective optimization result of each layer of optimization is added as an equality constraint to the solution process of the next layer of optimization. In response to the completion of the single-objective optimization solution, the values ​​of the decision variables corresponding to the decision variables are obtained, and based on the values ​​of the decision variables, an initial deployment mapping table from the logical function block to the intelligent remote processing unit is generated; Each logical function block is sent to the corresponding intelligent remote processing unit and a trial run is started to obtain trial run data corresponding to each logical function block. Based on the trial run data, the initial deployment mapping table is iteratively optimized to obtain a target deployment mapping table from each logical function block in the set of logical function blocks to the intelligent remote processing unit.

[0009] Furthermore, the hierarchical progressive optimization priority includes a first hierarchical priority, a second hierarchical priority, a third hierarchical priority, and a fourth hierarchical priority; the first hierarchical priority is to maximize the sum of the association weights of the logical functional blocks deployed in the same intelligent remote processing unit; the second hierarchical priority is to maximize the proportion of the logical functional blocks allocated to the first-level resource pool; the third hierarchical priority is to minimize the load rate difference of each intelligent remote processing unit; and the fourth hierarchical priority is to minimize the total communication traffic across the intelligent remote processing units. The step of distributing each logical function block to the corresponding intelligent remote processing unit and initiating trial operation, obtaining trial operation data corresponding to each logical function block, and iteratively optimizing the initial deployment mapping table based on the trial operation data to obtain a target deployment mapping table from each logical function block in the set of logical function blocks to the intelligent remote processing unit includes: The executable code and configuration parameters of each logical function block are sent to the corresponding intelligent remote processing unit using the deterministic real-time industrial network to start trial operation; During the trial operation, the actual operating performance datasets corresponding to each intelligent remote processing unit and each logical function block are collected in real time. Based on the actual operating performance datasets, the hierarchical progressive optimization algorithm is repeatedly called to update the initial deployment mapping table until the fluctuation of the actual operating performance dataset is lower than the preset fluctuation threshold, thereby obtaining the target deployment mapping table.

[0010] Furthermore, the step of each of the intelligent remote processing units executing the locally carried logical function blocks in parallel within each control cycle based on the target deployment mapping table, and completing cross-node data exchange through the deterministic real-time industrial network to realize the control closed loop of the distributed system includes: Each of the aforementioned intelligent remote processing units acquires the field input signals locally connected to the intelligent remote processing unit through a local I / O interface, and performs signal conditioning, isolation, and analog-to-digital conversion on the field input signals to obtain local input data; The logic function block code loaded in the memory of the intelligent remote processing unit is invoked. Based on the local input data and the input data received across nodes in the previous cycle, the local control logic operation of the logic function block is executed to obtain the intermediate result data and the target output value calculated by the logic function block. Based on the global clock synchronization signal of the distributed system, the intermediate result data of the local logical function block is sent to the corresponding intelligent remote processing unit through the deterministic real-time industrial network according to the predefined data flow dependency relationship, while receiving input data sent by the other intelligent remote processing units. The target output value is converted from digital to analog, isolated, and powered, and then output to the field actuator that is locally connected to the intelligent remote processing unit to realize the control closed loop of the distributed system.

[0011] Furthermore, the step of the logical scheduling unit acquiring in real time the operating status data of the intelligent remote processing unit and the control commands triggered by the engineer station or operator station, performing global control coordination on the field control station based on the operating status data and the control commands, and dynamically adjusting the target deployment mapping table according to the changes in the operating conditions of the field control station, includes: The logical scheduling unit acquires the running status data of the intelligent remote processing unit in real time, and generates a global running view based on the running status data; wherein, when a fault is detected in the intelligent remote processing unit, the logical function block carried by the intelligent remote processing unit is migrated to the intelligent remote processing unit in the same-site backup resource pool to achieve fault isolation; The logic scheduling unit obtains control commands triggered by the engineer or operator station based on the global operation view in real time. Based on the control commands, it sequentially performs global control mode switching, set value synchronization and configuration parameter consistency verification on the field control station, and ensures that the control logic of each field control station is coordinated and consistent. When the operating conditions of the field control station cause the distributed system to become unbalanced, the hierarchical progressive optimization algorithm is invoked again to migrate the non-strongly related logical function blocks online without interrupting control operation, so as to dynamically adjust the target deployment mapping table and maintain the load balance of the distributed system.

[0012] This application embodiment also provides a distributed system, which includes an engineering workstation, a deterministic real-time industrial network, and multiple field control stations; wherein each of the field control stations is equipped with a logical scheduling unit and multiple intelligent remote processing units; The engineering station has a built-in logic decoupling engine, which is used to decompose the single-unit control application into multiple logic function blocks in response to receiving the single-unit control application. The engine determines the remote processing unit range information corresponding to each logic function block, and allocates the logic function blocks to the corresponding field control stations based on the remote processing unit range information, thereby obtaining a set of logic function blocks for each field control station. The logical scheduling unit is used to generate a target deployment mapping table from each logical function block in the set of logical function blocks to the intelligent remote processing unit based on the topology computing power parameters of each intelligent remote processing unit and using a hierarchical progressive optimization algorithm; to acquire the operating status data of the intelligent remote processing unit and the control instructions triggered by the engineer station or operator station in real time; to perform global control coordination on the field control station based on the operating status data and the control instructions; and to dynamically adjust the target deployment mapping table according to the changes in the operating conditions of the field control station. The intelligent remote processing unit is a micro control node with independent computing capabilities. It is used to deploy a mapping table based on the target, execute the locally carried logical function blocks in parallel during each control cycle, and complete cross-node data exchange through the deterministic real-time industrial network to realize the control closed loop of the distributed system. The deterministic real-time industrial network is used to provide deterministic low-latency communication guarantees for each node in the distributed system, supporting cross-node data exchange and control command transmission.

[0013] This application also provides an electronic device, including: a processor, a memory, and a bus. The memory stores machine-readable instructions executable by the processor. When the electronic device is running, the processor communicates with the memory via the bus. When the machine-readable instructions are executed by the processor, the steps of the layered decoupling and elastic deployment method of the distributed system described above are performed.

[0014] This application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, performs the steps of the layered decoupling and elastic deployment method for distributed systems described above.

[0015] The layered decoupling and elastic deployment method and system for a distributed system provided in this application embodiment, wherein the distributed system includes an engineering workstation, a deterministic real-time industrial network, and multiple field control stations; wherein each of the field control stations is equipped with a logical scheduling unit and multiple intelligent remote processing units, and the method includes: In response to the engineering station receiving a single-unit control application, the engineering station's logic decoupling engine decomposes the single-unit control application into multiple logical function blocks, determines the remote processing unit range information corresponding to each logical function block, and allocates the logical function blocks to the corresponding field control stations based on the remote processing unit range information, obtaining a set of logical function blocks for each field control station; within each field control station, the logic scheduling unit, based on the topology computing power parameters of each intelligent remote processing unit, uses a hierarchical progressive optimization algorithm to generate each logical function block in the set of logical function blocks to the intelligent remote processing unit. A target deployment mapping table for the remote processing units is provided. Each intelligent remote processing unit, based on the target deployment mapping table, executes its locally hosted logical function blocks in parallel within each control cycle, completing cross-node data exchange through the deterministic real-time industrial network to achieve a control closed loop for the distributed system. The logical scheduling unit acquires the operating status data of the intelligent remote processing units and the control commands triggered by the engineer station or operator station in real time. Based on the operating status data and the control commands, it performs global control coordination on the field control station and dynamically adjusts the target deployment mapping table according to changes in the operating conditions of the field control station.

[0016] Compared to the traditional distributed control systems (DCS) in existing technologies that generally adopt a centralized logic execution architecture, consisting of upper-level management nodes, a main controller, and lower-level I / O modules, where the main controller acts as a single control brain, centrally executing all control strategies, this approach decouples the monolithic control application into multiple independently deployable logical function blocks and completes station-level coarse allocation. The logic scheduling unit of each field control station generates a target deployment mapping table for the logical function blocks based on the topology computing power parameters of the intelligent remote processing unit, using a hierarchical progressive optimization algorithm. The distributed intelligent remote processing units then carry out parallel execution of the control logic and achieve cross-node data collaboration through a deterministic real-time industrial network. Simultaneously, the logic scheduling unit performs global monitoring of the operating status, system-level control coordination, and dynamic deployment adjustments to adapt to operating conditions. This solves the problems of static coupling of computing resources in traditional centralized architectures, such as performance ceilings and single-point failure risks; rigid logic deployment leading to uneven resource utilization and low efficiency; and specialized hardware functions resulting in high system reconfiguration costs and insufficient flexibility. This approach achieves the technical effects of horizontally elastically scalable system computing power, naturally isolated faults improving overall reliability, and globally optimized resource configuration reducing communication losses.

[0017] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 A flowchart illustrating a layered decoupling and elastic deployment method for a distributed system provided in this application embodiment; Figure 2 This is a schematic diagram of the structure of a distributed system provided in an embodiment of this application; Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of this application. Based on the embodiments of this application, every other embodiment obtained by those skilled in the art without inventive effort falls within the scope of protection of this application.

[0021] Research has revealed that as industrial processes become increasingly large-scale, complex, and sophisticated, the centralized control architecture of traditional distributed control systems (DCS) is increasingly exhibiting performance bottlenecks, poor scalability, and single-point-of-failure risks. Current technologies, which centralize all core control logic in a central controller, limit the overall system capability to the performance of a single hardware component and cannot flexibly adjust to actual operating loads. Furthermore, the static binding of control logic to hardware location leads to low system resource utilization and results in high adjustment costs and insufficient flexibility when facing process changes.

[0022] Existing technologies are distributed control systems with a centralized logic execution architecture. The core of this approach is to centrally execute control strategies through a main controller (central controller / main DPU), scanning all I / O signals at fixed cycles to construct a "scan-operation-output" execution closed loop based on deterministic timing. Its focus is primarily on ensuring the periodic determinism and predictable timing of control logic execution, rather than the dynamic elasticity and distributed coordination of computing resources. Based on the above-mentioned traditional centralized DCS architecture, the following technical drawbacks exist: The static coupling of computing resources presents performance limitations and single-point-of-failure risks, hindering elastic scaling: Existing technologies heavily rely on a single central controller for control logic execution. The total processing capacity of the control system is physically fixed after controller selection, unable to adjust flexibly based on actual load during operation. When the process expands or complex control algorithms (such as Advanced Process Control, APC) are added, the CPU utilization of the central controller may approach its limit. The only solution is to shut down and replace with a higher-performance controller, or add redundant controllers for switching, but this does not improve the system's steady-state capacity. Simultaneously, the controller becomes a single point of failure; its failure will cause all control loops under its jurisdiction to fail. Although redundancy switching mechanisms can mitigate this, the switching process itself has a risk window, and spare parts are expensive. The consequences are: the system lacks horizontal scalability when facing increasing loads, exhibiting a clear performance bottleneck; the overall reliability of the system is limited by the highest reliability of a single hardware unit, failing to achieve decentralized system-level high availability.

[0023] The fixed deployment location of logic, unable to be flexibly optimized according to real-time status, leads to low resource utilization and local overload: In the engineering configuration phase, engineers must manually and statically bind control loops and sequential logic to a designated central controller. Once downloaded, this binding relationship is inconvenient to change during system operation. Even after the system is actually running, the computational load of each control loop may be extremely uneven (for example, some loops are in a rapid adjustment state and consume a lot of computation, while some loops are in a steady state). However, because the logic cannot be migrated, the controller with excessive load will still bear the pressure, while the controller with less load cannot share the burden. This is essentially a "static partitioning" resource allocation strategy, which cannot perform flexible load balancing according to real-time operating conditions. The consequence is low global utilization efficiency of the control system's computing resources and the existence of "uneven busy and idle" phenomena. When operating conditions fluctuate, local controllers may trigger performance alarms or even degrade control quality due to a surge in instantaneous computing demand, while other idle resources in the system cannot be effectively utilized to cope with this sudden situation.

[0024] Specialized hardware functions restrict the flexible deployment and collaborative execution of complex strategies, resulting in high system refactoring costs. In the existing architecture, the roles and functions of the central controller and I / O units are strictly separated. The central controller handles complex logic, while the I / O units are only responsible for signal conversion. For complex control strategies requiring strong real-time performance, tight multi-loop coupling, and frequent data exchange (such as anti-surge control and multi-variable coordinated control for large compressors), all signal acquisition, algorithm calculation, and output driving must be centrally processed by the central controller. This causes data surges on the communication bus and congestion within the controller. Even if some intelligent I / O modules possess simple local processing capabilities, their functions are usually fixed (e.g., only able to execute a single PID), unable to flexibly support user-defined, complex, and customized control algorithms. The consequences include: a rigid system architecture, unable to flexibly and efficiently support modern advanced control algorithms; poor system maintainability and scalability; and high costs, long cycles, and significant risks associated with changes.

[0025] Based on this, the embodiments of this application provide a hierarchical decoupling and elastic deployment method for distributed systems. By decoupling the monolithic control application into multiple independently deployable logical function blocks and completing station-level coarse allocation, the logical scheduling unit of each field control station generates a target deployment mapping table for the logical function blocks based on the topology computing power parameters of the intelligent remote processing unit using a hierarchical progressive optimization algorithm. The distributed intelligent remote processing units carry out the parallel execution of control logic and complete cross-node data collaboration through a deterministic real-time industrial network. At the same time, the logical scheduling unit realizes global monitoring of the operating status, system-level control coordination, and dynamic deployment adjustment for adapting to operating conditions. This solves the problems of performance ceiling and single point of failure risk in the static coupling of computing resources in traditional centralized architectures, low utilization rate due to uneven resource availability caused by rigid logical deployment, and high system reconstruction cost and insufficient flexibility due to hardware function specialization. It achieves the technical effects of horizontal elastic expansion of system computing power, natural fault isolation to improve overall reliability, and global optimization of resource configuration to reduce communication loss.

[0026] The core idea of ​​the method described in this application embodiment is to liberate the control application from the traditional form that is strongly coupled with the central controller, decoupling it into logical function blocks with appropriate granularity, clear interfaces, and independent deployment; and to build a distributed, software-defined control logic execution layer by utilizing the intelligent remote processing units with independent computing capabilities that are widely distributed in the field control station; and to realize the deployment and resource optimization of logical function blocks among the intelligent remote processing units through the main control unit, thereby constructing a decentralized, elastically scalable, and highly reliable new distributed system execution layer architecture.

[0027] Please see Figure 1 , Figure 1This is a flowchart illustrating a layered decoupling and elastic deployment method for a distributed system provided in an embodiment of this application. The distributed system includes an engineering workstation, a deterministic real-time industrial network, and multiple field control stations; each field control station is equipped with a logical scheduling unit and multiple intelligent remote processing units.

[0028] Here, the engineer station is the upper-level operation node in the distributed control system used for system configuration, logic programming, parameter configuration, and engineering debugging. It allows engineering technicians to write control logic, download configuration programs, and monitor the system's operating status. The deterministic real-time industrial network is an industrial communication network with deterministic transmission delay and precise timing scheduling capabilities. It ensures predictable data transmission delays and eliminates random conflicts, providing communication support for the real-time requirements of industrial control. It consists of two levels: an inter-station communication network and an intra-station backplane network. The field control station is a cluster of control units deployed at the production site in the distributed control system, directly connecting to field equipment. It is used to carry out control logic execution and... Field signal interaction serves as the on-site implementation platform for control functions. The logic scheduling unit, deployed within the field control station, employs redundant configurations to enhance reliability. It is the core of management and coordination of control resources within the station, responsible for the deployment and scheduling of logic function blocks, monitoring of operational status, system-level coordination, and dynamic optimization. It can also carry some global control logic as needed. The intelligent remote processing unit is a miniature control node with independent computing capabilities. Unlike traditional I / O modules that only have signal conversion functions, it combines I / O signal processing and complex logic execution capabilities. The logic functions it carries are entirely software-defined, rather than hardware-fixed, making it a distributed execution platform for control logic.

[0029] like Figure 1 As shown, the method includes: S101. In response to the engineer station receiving the single-unit control application, the engineer station uses its logic decoupling engine to decompose the single-unit control application into multiple logic function blocks, determines the remote processing unit range information corresponding to each logic function block, and allocates the logic function blocks to the corresponding field control stations based on the remote processing unit range information, thereby obtaining a set of logic function blocks for each field control station.

[0030] In this embodiment of the application, the core purpose of this step is to decompose the centralized monolithic control application into standardized logical function units that can be deployed independently, and to complete the initial allocation between stations across the entire plant. This decomposes the complex global problem into independent sub-problems of a single field control station, laying the foundation for subsequent fine-grained deployment within the station.

[0031] Here, a standalone control application refers to a complete set of control applications that are fully written and downloaded to a single main controller for execution under a traditional centralized architecture. It includes all control functions such as PID regulation loops, interlocking sequential control logic, process calculations, and advanced control algorithms.

[0032] In this embodiment of the application, the logic decoupling engine is a software functional module deployed in the engineering station, used to structurally decompose the monolithic control application and generate independent logical functional units with standardized interfaces.

[0033] Among them, the logic function block is the smallest deployable unit after the control logic is decomposed. It has clear input and output interfaces, independent execution cycle, quantifiable computing power requirements and clear data flow dependencies, and can be independently loaded into the intelligent remote processing unit for operation.

[0034] Here, the remote processing unit scope information refers to the set and priority information of the intelligent remote processing units that can be deployed in each logical function block. It is defined by the three-level deployable resource pool and is used to constrain the feasible domain of the deployment algorithm to ensure optimal communication cost.

[0035] In one possible implementation of this application, in specific implementation, the step S101, which uses the logic decoupling engine of the engineering station to decompose the monolithic control application into multiple logical function blocks and determines the remote processing unit range information corresponding to each logical function block, may include: S1011. Using the logic decoupling engine of the engineering station, the monolithic control application is divided into multiple coarse-grained logic function blocks with the control logic function integrity of the monolithic control application as the decomposition granularity, and the data flow dependency relationship between each coarse-grained logic function block is determined.

[0036] In this step, the functional integrity of the control logic is used as the granularity for decomposition. The decomposition is carried out according to the business boundaries of the process loop and functional unit. For example, a complete PID control loop and its supporting range conversion, interlocking protection, and output limiting logic are divided into a coarse-grained logic functional block. During the decomposition process, the data producer and consumer relationships between each coarse-grained logic functional block are sorted out to form a preliminary data flow dependency diagram.

[0037] This ensures that the decomposed functional units possess complete business semantics, avoids disrupting the inherent coherence of control logic, and provides a basic framework for subsequent refined decomposition. It forms clearly defined, coarse-grained functional units, guaranteeing the functional integrity of control logic and the traceability of dependencies.

[0038] For example, assuming that the control program for a certain chemical reactor is initially decomposed, the reactor temperature PID control loop, reactor pressure interlock logic, and feed flow rate regulation loop are divided into independent coarse-grained logic function blocks. At the same time, the interlock dependency relationship between reactor temperature output and feed flow rate regulation, and the control dependency relationship between reactor pressure interlock and feed valve are marked.

[0039] S1012. Based on the data flow dependency relationship, the coarse-grained logic function block is initially divided according to the code size of the coarse-grained logic function block to obtain multiple preliminary subdivided logic function blocks, and the estimated running time of each preliminary subdivided logic function block is obtained through simulation.

[0040] S1013. With the goal of maintaining a balance in the estimated running time of each of the preliminary subdivided logical function blocks, perform iterative splitting and merging on the preliminary subdivided logical function blocks to obtain multiple logical function blocks with balanced granularity.

[0041] Each of the aforementioned logical functional blocks has clearly defined input / output interfaces, execution cycles, computing power requirements, and data flow dependency attributes.

[0042] In this step, the coarse-grained logic functional blocks are first initially divided according to code size, resulting in preliminary subdivided logic functional blocks. Then, a simulation tool is used to simulate the execution process of each preliminary subdivided logic functional block, and the estimated running time of each block is statistically obtained to quantify its computational requirements. With the goal of achieving a relatively balanced running time for all logic functional blocks, iterative adjustments are made: functional blocks with running times significantly exceeding the average are further split at weakly dependent interfaces; functional blocks with excessively short running times and close data interactions are appropriately merged, ultimately resulting in a standardized set of logic functional blocks with uniform granularity.

[0043] Here, each final logical function block has a clear input / output interface definition, a fixed execution cycle, quantifiable computing power requirements, and complete data flow dependency attributes.

[0044] This eliminates the computational power differences between functional blocks, avoids the overload of a single functional block's computational power, and provides a uniformly granular and attribute-standardized scheduling unit for subsequent deployment and scheduling. It achieves granular standardization and attribute quantification of logical functional blocks, ensuring the accuracy and feasibility of subsequent deployment optimization algorithms.

[0045] For example, if the estimated running time of a certain coarse-grained advanced control function block is 12ms, which is far greater than the average running time of 2ms for function blocks in the station, it can be decomposed into three sub-function blocks: parameter acquisition, model calculation, and output solution. Each sub-function block has a running time of about 3 to 4ms, thus achieving granularity balance.

[0046] S1014. Based on the physical affiliation of the input / output channels associated with each logical function block, a three-level deployable resource pool is defined for each logical function block in descending order of communication cost, and the remote processing unit range information corresponding to each logical function block in the three-level deployable resource pool is determined.

[0047] The first-level resource pool consists of the intelligent remote processing units within the same field control station, including associated input / output channels; the second-level resource pool consists of the remaining intelligent remote processing units within the same field control station; and the third-level resource pool consists of the intelligent remote processing units of the remaining field control stations.

[0048] It should be noted that the three-tier deployable resource pool is a three-tiered range of deployable intelligent remote processing units for each logical functional block, defined by communication costs from low to high. The deployment priority decreases as communication costs increase.

[0049] Here, the deployment priority of the first-level resource pool is greater than the deployment priority of the second-level resource pool, and the deployment priority of the second-level resource pool is greater than the deployment priority of the third-level resource pool.

[0050] Specifically, based on the physical location of the input and output channels associated with each logical function block, a three-level resource pool is defined for each logical function block. After the definition is completed, the remote processing unit range information corresponding to each logical function block is obtained.

[0051] The resource pool is divided into three levels: Level 1: Intelligent remote processing units within the same field control station that contain the input / output channels associated with the logical function block. When the function block is deployed at this level, I / O data can be directly read locally, resulting in the lowest communication load and the highest deployment priority. Level 2: Other intelligent remote processing units within the same field control station. When the function block is deployed at this level, I / O data needs to be transmitted through the station's backplane network, resulting in a medium communication load and a lower deployment priority. Level 3: Intelligent remote processing units from other field control stations. When the function block is deployed at this level, I / O data needs to be transmitted through the inter-station industrial network, resulting in the highest communication load and the lowest deployment priority.

[0052] This sets rigid constraints on the algorithm deployment, forcing functional blocks to be deployed closer to the data source, thus reducing communication latency and network load at the architectural level. The deployment rules ensure the principle of "data processing nearby," providing fundamental constraints for low-latency, high-real-time distributed control.

[0053] In one possible implementation of this application, in specific implementation, the step S101 of allocating the logical function blocks to the corresponding field control stations based on the remote processing unit range information, to obtain the logical function block set of each field control station, may include: S1015. Based on the remote processing unit range information, determine the computing power weight and input / output home station attribute information of each logical function block, as well as the data flow dependency and logical correlation between the logical function blocks.

[0054] Here, logical correlation is an indicator that quantifies the degree of data interaction between two logical functional blocks. The calculation formula is: "Logical correlation = amount of data interaction × interaction frequency × dependency coefficient". The higher the value, the stronger the correlation between the two functional blocks and the more suitable they are to be deployed on the same node.

[0055] S1016. Using each of the logical functional blocks as a vertex with the computing power weight and the input / output home station attribute information, the data flow dependency as an edge, and the logical correlation degree as the edge weight, construct an undirected weighted dependency graph.

[0056] Here, the undirected weighted dependency graph is used to describe the graph model of the relationship between logical functional blocks. The logical functional blocks are the vertices, the data flow dependencies between functional blocks are the edges, and the logical correlation degree is the edge weight.

[0057] In this step, an undirected weighted dependency graph is constructed based on the attributes and data flow dependencies of all logical functional blocks. Each vertex corresponds to a logical functional block, with an associated computing power weight (estimated running time) and input / output station attributes; if two functional blocks have a data flow dependency, they are connected by an edge, with the edge weight being the corresponding logical correlation.

[0058] S1017. Traverse the edges in descending order of edge weights, merge strongly related vertices into supernodes, and prioritize the allocation of supernodes with clearly defined input / output affiliations to their corresponding field control stations in descending order of total computing power. Allocate cross-station supernodes to the field control stations corresponding to the first load.

[0059] If the computing power of the assigned field control station is insufficient, the supernode will be assigned to a field control station adjacent to that field control station.

[0060] Here, a supernode is a composite node formed by merging multiple strongly related vertices during the graph coarsening process. It inherits attributes such as total computing power, I / O home station, and external edge weights from its child nodes, and is used to reduce the size of the graph and improve the efficiency of the partitioning algorithm.

[0061] In this step, all edges are traversed from highest to lowest weight. If the vertices at both ends of an edge have not been merged, and the total computing power of the merged supernode does not exceed 1 / 3 of the maximum computing power of a single field control station, then the two vertices are merged into one supernode. The supernode inherits the total computing power of all its child nodes, its I / O affiliation station (if all child nodes' I / O belong to the same station, it belongs to that station; otherwise, it is marked as cross-station), and the sum of its external edge weights with other supernodes. This coarsening process is iteratively performed until the graph size is reduced to 2 to 3 times the number of field control stations.

[0062] Furthermore, supernodes are sorted by their total computing power from highest to lowest, and allocation is performed sequentially. For supernodes with a clearly defined I / O home station, priority is given to allocating them to their home field control station; if the home station has sufficient remaining computing power, allocation is performed directly; otherwise, allocation is performed to an adjacent field control station. For cross-site supernodes, allocation is performed to the relevant field control station with the lowest current load. Throughout the allocation process, it is ensured that the total computing power of each field control station does not exceed its preset computing power limit.

[0063] S1018. Under the preset rigid constraints, the improved Phidusia and Matthews method is used to iteratively migrate the supernodes of every two field control stations until no benefit migration is achieved or the number of iterations reaches a threshold, so as to obtain the set of logical function blocks of each field control station.

[0064] Here, the improved Phidusia and Matthews method is based on the classic graph partitioning algorithm, and superimposed with an optimization algorithm for rigid constraints in industrial control scenarios. Under the premise of satisfying the constraints, it achieves the minimization of cross-site edge cutting weight and inter-site load balancing.

[0065] Here, an improved Phidussia and Matthews algorithm is used for iterative optimization, with rigid constraints set: nodes with clear I / O affiliation cannot migrate out of their home station, and strongly associated supernodes cannot be split within each other.

[0066] In this step, an iterative approach is adopted, traversing all field control station pairs in a loop. Each time, the migration of supernodes only occurs between two stations. The optimization priorities are as follows: the first priority is to minimize the sum of cross-station edge weights (reducing cross-station logical association losses), and the second priority is to reduce the difference in total computing power between stations. Iteration continues until no more profitable migrations are possible, or the preset maximum number of iterations is reached, and the final station-level allocation result is output, which is the set of logical function blocks corresponding to each field control station.

[0067] This approach achieves plant-wide functional block inter-station allocation, maximizing the proportion of strongly correlated functional blocks deployed on the same station while satisfying computing power constraints and resource pool priorities. It decomposes the globally complex problem into independent sub-problems on each station, reducing the computational complexity of subsequent fine-grained deployment within each station. This results in optimal station-level partitioning across the entire plant, reducing cross-station communication traffic at the station level while ensuring initial load balancing between stations, laying a high-quality foundation for fine-grained deployment within stations.

[0068] For example, suppose a plant-wide system contains 3 field control stations with a total of 120 logical function blocks. After graph coarsening, these are merged into 7 supernodes. After greedy allocation, station 1 is allocated 3 supernodes with a total of 42 function blocks, station 2 is allocated 2 supernodes with a total of 38 function blocks, and station 3 is allocated 2 supernodes with a total of 40 function blocks. After refinement and adjustment by the improved algorithm, the total weight of cross-station edges is reduced by 18%, and the maximum computing power difference between stations is reduced from 12% to 5%, achieving optimal allocation at the station level.

[0069] S102. Within each of the field control stations, the logic scheduling unit generates a target deployment mapping table from each of the logical function blocks in the set of logical function blocks to the intelligent remote processing unit based on the topology computing power parameters of each of the intelligent remote processing units using a hierarchical progressive optimization algorithm.

[0070] Among them, the topology computing power parameters refer to the hardware attribute parameters such as the physical location, interconnection relationship, link bandwidth, and single-node computing power limit of the intelligent remote processing units within the station, which serve as the hardware constraint basis for deployment optimization; the hierarchical progressive optimization algorithm is a multi-objective optimization algorithm that solves the single-objective optimization problem layer by layer according to the priority from high to low, and adds the optimal result of the previous layer as a rigid constraint to the solution of the next layer, ensuring that high-priority indicators are not sacrificed by low-priority objectives; the target deployment mapping table is a mapping relationship table that records the intelligent remote processing unit numbers deployed corresponding to each logical function block, which serves as the basis for the issuance and operation of function blocks.

[0071] In this embodiment of the application, the core objective of this step is to generate a precise deployment mapping table from logical function blocks to intelligent remote processing units based on the hardware topology and computing power parameters of the intelligent remote processing unit within a single field control station, using a hierarchical and progressive optimization algorithm, and to obtain a final stable deployment scheme through trial operation and iterative optimization.

[0072] In one possible implementation of this application, step S102 may include: S1021. Define decision variables and rigid hard constraints for each of the field control stations.

[0073] The decision variables include the deployment marker variable and the co-deployment association variable of the logical function block; the rigid hard constraints include that each logical function block is deployed only in one intelligent remote processing unit, and the target computing power of a single intelligent remote processing unit does not exceed the upper limit of computing power after reserving the safety margin, and the logical function block is deployed only within the resource pool corresponding to the remote processing unit range information, and the co-deployment association variable is consistent with the deployment marker variable.

[0074] Specifically, the decision variables include two categories: one is the deployment marker variable, which is a binary variable. A value of 1 indicates that the corresponding logical function block is deployed on the corresponding intelligent remote processing unit; a value of 0 indicates that it is not deployed. The other is the same deployment association variable, which is a binary variable. A value of 1 indicates that the two logical function blocks are deployed on the same intelligent remote processing unit; a value of 0 indicates that they are not in the same unit.

[0075] The rigid hard constraints include: allocation uniqueness constraint: each logical function block must be deployed on one and only one intelligent remote processing unit; computing power limit constraint: the total computing power requirement of all logical function blocks carried by a single intelligent remote processing unit shall not exceed the computing power limit after reserving a 30% safety margin for that unit; resource pool range constraint: each logical function block can only be deployed on intelligent remote processing units within its own three-level resource pool range; consistency constraint: the values ​​of deployment-related variables must be consistent with the calculation results of deployment marker variables.

[0076] This approach clearly defines the decision space and rigid boundary of the optimization problem, ensuring that the solution meets engineering feasibility requirements. A standardized integer programming solution model is constructed, providing a standardized solution framework for subsequent hierarchical optimization.

[0077] S1022. Based on the topology computing power parameters of each of the intelligent remote processing units, under the rigid hard constraint conditions, single-objective optimization is performed on the decision variables in descending order of the preset hierarchical progressive optimization priority, and the objective optimization result of each layer of optimization is added as an equality constraint to the solution process of the next layer of optimization.

[0078] In this step, four layers of single-objective optimization are executed in sequence from high to low according to the preset optimization priority. After the optimal value is obtained at each layer, the optimal value is added as an equality constraint to the optimization model of the next layer to ensure that the optimization of low-priority objectives does not sacrifice high-priority indicators.

[0079] The hierarchical progressive optimization priority includes a first hierarchical priority, a second hierarchical priority, a third hierarchical priority, and a fourth hierarchical priority; the first hierarchical priority is to maximize the sum of the association weights of the logical function blocks deployed in the same intelligent remote processing unit; the second hierarchical priority is to maximize the proportion of the logical function blocks allocated to the first-level resource pool; the third hierarchical priority is to minimize the load rate difference of each intelligent remote processing unit; and the fourth hierarchical priority is to minimize the total communication traffic across the intelligent remote processing units.

[0080] Specifically, the four-layer optimization priorities and objectives include: First layer priority (highest): Maximize the sum of association weights of logical function blocks deployed in the same intelligent remote processing unit, i.e., minimize the logical association loss across intelligent remote processing units, and prioritize the deployment of strongly associated function blocks on the same node; Second layer priority: Maximize the proportion of logical function blocks allocated to the first-level resource pool, and prioritize the deployment of function blocks close to the I / O data source to reduce data read latency; Third layer priority: Minimize the load rate difference among intelligent remote processing units to achieve on-site computing power load balancing and avoid local overload; Fourth layer priority (lowest): Minimize the total communication traffic across intelligent remote processing units to further optimize the on-site network load.

[0081] In this way, in multi-objective optimization scenarios, the optimality of core priority indicators is ensured through a hierarchical and progressive approach, avoiding the damage to core indicators caused by weighted summation of multiple objectives. This achieves multi-dimensional hierarchical optimization of on-site deployment, prioritizing communication performance while also considering resource proximity, load balancing, and network optimization.

[0082] S1023. In response to the completion of the single-objective optimization solution, obtain the values ​​of the decision variables corresponding to the decision variables, and generate an initial deployment mapping table from the logical function block to the intelligent remote processing unit based on the values ​​of the decision variables.

[0083] In this step, after the four-layer optimization solution is completed, an initial deployment mapping table from the logical function blocks of the field control station to the intelligent remote processing unit is generated based on the final values ​​of the deployment marker variables, thus clarifying the target deployment node for each function block.

[0084] In this way, an executable initial deployment plan is output, providing a basis for the subsequent distribution and trial operation of functional blocks, and obtaining the theoretically optimal deployment plan based on simulation parameters.

[0085] S1024. Each of the logical function blocks is sent to the corresponding intelligent remote processing unit and trial operation is started to obtain trial operation data corresponding to each of the logical function blocks. Based on the trial operation data, the initial deployment mapping table is iteratively optimized to obtain the target deployment mapping table from each of the logical function blocks in the set of logical function blocks to the intelligent remote processing unit.

[0086] In one possible implementation of this application, step S1024 may include: S10241. Using the deterministic real-time industrial network, the executable code and configuration parameters of each logical function block are sent to the corresponding intelligent remote processing unit to start trial operation.

[0087] In this step, the logical scheduling unit distributes the executable code, configuration parameters, and operating configuration of each logical function block to the intelligent remote processing unit corresponding to the initial deployment mapping table through the deterministic real-time industrial network of the station backplane; after all units complete the loading verification, the system enters the trial operation state.

[0088] S10242. During the trial operation, the actual operating performance dataset corresponding to each intelligent remote processing unit and each logical function block is collected in real time. Based on the actual operating performance dataset, the hierarchical progressive optimization algorithm is repeatedly called to update the initial deployment mapping table until the fluctuation of the actual operating performance dataset is lower than the preset fluctuation threshold, and the target deployment mapping table is obtained.

[0089] During this step, the logical scheduling unit collects resource status data such as CPU utilization, memory usage, and network load of each intelligent remote processing unit in real time, using control cycles as the unit, as well as operational performance data such as actual running time, execution cycle deviation, and output result correctness of each logical function block, forming an actual operational performance dataset.

[0090] Furthermore, the simulation-estimated parameters during the initial optimization are replaced with actual operating performance data, and the hierarchical progressive optimization algorithm is repeatedly called to iteratively update the initial deployment mapping table. The iteration continues until the fluctuation of the system's operating status is lower than the preset fluctuation threshold (e.g., load fluctuation is less than 2%, execution cycle deviation is less than 0.1ms), then the trial run ends, the final target deployment mapping table is output, and the system enters formal operation.

[0091] In this way, simulation estimation errors are corrected using real-world operational data, ensuring that the deployment plan reaches its optimal state in actual operating scenarios. This eliminates the deviation between simulation parameters and actual operation, guarantees the actual optimality of the deployment plan, and improves the stability and performance of system operation.

[0092] For example, suppose the simulation estimated running time of a certain function block is 2ms, but the actual running time measured during the trial run is 2.8ms, resulting in a high load on the unit where it is located. After iterative optimization, the two low-association function blocks in this unit are migrated to the adjacent low-load unit. Finally, the load rate of all units is stabilized within the range of 60%±2%, reaching the stable operation standard.

[0093] S103. Each of the intelligent remote processing units executes the locally carried logical function blocks in parallel within each control cycle based on the target deployment mapping table, and completes cross-node data exchange through the deterministic real-time industrial network to realize the control closed loop of the distributed system.

[0094] In this embodiment of the application, the core purpose of this step is that after the system is officially running, all intelligent remote processing units will execute their respective control logic independently and in parallel based on the target deployment mapping table, and complete cross-node data collaboration through the deterministic real-time industrial network to achieve a complete distributed control closed loop.

[0095] Among them, the control closed loop refers to the complete closed-loop process of the control system from field signal acquisition, control logic operation to control output execution, and is the core process of the control system to achieve automatic control; the global clock synchronization signal is a synchronization clock signal issued by the system's global clock source, used to ensure the execution timing alignment of all distributed nodes and ensure the determinism of distributed execution.

[0096] In one possible implementation of this application, step S103 may include: S1031. Each of the intelligent remote processing units acquires the field input signal locally connected to the intelligent remote processing unit through the local I / O interface, and performs signal conditioning, isolation and analog-to-digital conversion on the field input signal to obtain local input data.

[0097] In this step, at the beginning of each control cycle, each intelligent remote processing unit acquires the field input signals output by the field sensors and transmitters directly connected to it through the local I / O interface, and completes signal conditioning, electrical isolation and analog-to-digital conversion, converting the analog or switch field signals into local digital input data that the system can process.

[0098] This approach completes the acquisition of input signals for the control closed loop, allowing for local data collection from the field and reducing signal transmission loss and delay. It also enables local digital processing of field signals, avoiding the interference and delay issues associated with long-distance signal transmission in centralized architectures.

[0099] S1032. Call the logic function block code loaded in the memory of the intelligent remote processing unit, and execute the local control logic operation of the logic function block based on the local input data and the input data received across nodes in the previous cycle, to obtain the intermediate result data and the target output value calculated by the logic function block.

[0100] In this step, the intelligent remote processing unit calls the executable code of the logic function block loaded in its memory, and performs all local control logic operations such as PID adjustment, logic judgment, process calculation, and advanced control algorithm based on local input data and input data from other function blocks received across nodes in the previous cycle, to obtain intermediate result data and the final target output value generated by the operation.

[0101] In this way, distributed parallel computation of control logic is completed, realizing distributed delivery of control functions. All intelligent remote processing units execute computations in parallel, and the total processing capacity of the system is the sum of the computing power of all units, breaking through the performance ceiling of a single controller.

[0102] S1033. Based on the global clock synchronization signal of the distributed system, the intermediate result data of the local logical function block is sent to the corresponding intelligent remote processing unit through the deterministic real-time industrial network according to the predefined data flow dependency relationship, while receiving the input data sent by the other intelligent remote processing units.

[0103] In this step, based on the global clock synchronization signal of the distributed system, within a preset communication time slot of each control cycle, each intelligent remote processing unit sends intermediate result data generated by its local logic function block operations, which needs to be used by other nodes, to the corresponding target intelligent remote processing unit through the deterministic real-time industrial network according to predefined data flow dependencies; simultaneously, it receives input data required by its own node's function block from other intelligent remote processing units. The communication process is guaranteed by deterministic network scheduling, ensuring no conflicts and fixed latency.

[0104] This enables data collaboration among distributed nodes, ensuring data integrity and timing consistency across node control loops. It achieves deterministic data interaction in a distributed environment, replicating the deterministic control timing of a centralized architecture and guaranteeing control quality.

[0105] For example, suppose the main loop function block of a cascade control loop is deployed in unit 1 and the secondary loop function block is deployed in unit 2. In each control cycle, the main loop output value calculated by unit 1 is sent to unit 2 through the backplane network within a fixed time slot as the set value of the secondary loop. The transmission delay is fixed at 0.2ms, which meets the real-time requirements of cascade control.

[0106] S1034. The target output value is converted from digital to analog, isolated, and powered, and then output to the field actuator that is locally connected to the intelligent remote processing unit to realize the control closed loop of the distributed system.

[0107] In this step, the intelligent remote processing unit performs digital-to-analog conversion, electrical isolation, and power drive on the final target output value obtained from the calculation, converting it into a signal that can be received by the field actuator, and outputting it to the locally connected field actuator (such as a regulating valve, motor, contactor, etc.) to complete the control of the field equipment.

[0108] This completes the final output of the control closed loop, achieving automatic control of the production process. Outputting control signals from the nearest point reduces transmission loss and interference in the output channel, improving control response speed.

[0109] S104. The logic scheduling unit obtains the operating status data of the intelligent remote processing unit and the control instructions triggered by the engineer station or operator station in real time. Based on the operating status data and the control instructions, it performs global control coordination on the field control station and dynamically adjusts the target deployment mapping table according to the changes in the operating conditions of the field control station.

[0110] In this embodiment of the application, the core purpose of this step is for the logical scheduling unit to assume the role of global management within the station, realize operation status monitoring, fault isolation, system-level coordination and dynamic load optimization, ensure stable and reliable operation of the system, and maintain elastic optimization capabilities.

[0111] Among them, the control commands are issued by the engineer station or operator station and are used to control the system's operating mode, adjust process parameters, and upgrade configuration logic. The global operation view is an overview of the entire station's operating status generated by the logic scheduling unit after summarizing the operating status of all nodes. It is used to display the overall operating status of the system. Fault isolation refers to the rapid migration of the functions carried by a node to a backup node when a node fails, limiting the impact of the fault to a minimum and preventing the fault from spreading and causing system paralysis.

[0112] In one possible implementation of this application, step S104 may include: S1041. The logical scheduling unit obtains the running status data of the intelligent remote processing unit in real time, and generates a global running view based on the running status data.

[0113] When a fault is detected in the intelligent remote processing unit, the logical function block carried by the intelligent remote processing unit is migrated to the intelligent remote processing unit in the same site backup resource pool to achieve fault isolation.

[0114] In this step, the logical scheduling unit continuously collects the heartbeat status, resource utilization, task execution status, and intra-site network link status of all intelligent remote processing units, and summarizes them to generate a global operational view of the entire site for monitoring and display by upper-level sites. When a hardware failure or software anomaly is detected in an intelligent remote processing unit, the fault handling mechanism is immediately triggered, and all logical function blocks carried by the faulty unit are quickly migrated to idle intelligent remote processing units in the same site's backup resource pool. The switching process ensures no disturbance or only a short interruption, limiting the impact of the fault to a local area and achieving fault isolation.

[0115] This enables global, visualized control over the system's operational status, rapid handling of node failures, and improved overall system reliability. A single node failure only affects a local loop and will not cause a complete system outage; therefore, the overall system reliability is far superior to the single-controller, single-point-of-failure mode of a centralized architecture.

[0116] S1042. The logic scheduling unit obtains the control instructions triggered by the engineer or operator station based on the global operation view in real time. Based on the control instructions, it sequentially performs global control mode switching, set value synchronization and configuration parameter consistency verification on the field control station, and ensures that the control logic of each field control station is coordinated and consistent.

[0117] In this step, the logic scheduling unit receives control commands issued by the engineer station or operator station and executes system-level coordination operations across the entire station, including: global control mode switching, such as automatic / manual switching and local / remote switching, which are synchronously sent to all relevant intelligent remote processing units to ensure consistent switching timing and no logical conflicts; synchronous distribution of setpoints, which synchronously distributes the setpoint parameters of process adjustments to the corresponding control loops to ensure parameter consistency; and configuration parameter consistency verification, which periodically verifies the configuration parameter version of each node to ensure that the logic version of the entire station is unified.

[0118] This achieves unified global management and control within a distributed architecture, ensuring the coordination and consistency of control logic across the entire site. It solves the problem of difficult global operation coordination in a distributed architecture, balancing the performance advantages of distributed execution with the convenience of centralized management.

[0119] S1043. When the operating conditions of the field control station cause the distributed system to become unbalanced, the hierarchical progressive optimization algorithm is called again to migrate the non-strongly related logical function blocks online without interrupting the control operation, so as to dynamically adjust the target deployment mapping table and maintain the load balance of the distributed system.

[0120] In this step, when changes occur in the field conditions, leading to increased load or load imbalance in some intelligent remote processing units, or when the system is expanded to add new logical function blocks, the logical scheduling unit re-invokes the hierarchical progressive optimization algorithm. Without interrupting the system control operation, it performs online migration and adjustment of non-strongly related logical function blocks, dynamically updates the target deployment mapping table, and maintains the system's load balance and optimal communication performance.

[0121] This enables flexible resource scheduling during operation, adapting to fluctuations in operating conditions and system expansion, and continuously maintaining the system in its optimal operating state. It truly realizes a software-defined flexible control architecture, allowing the system to dynamically adjust resource allocation online without requiring downtime hardware modifications, significantly improving the system's flexibility and scalability.

[0122] For example, suppose a production process is expanded and four new control loop function blocks are added. The logic scheduling unit allocates the new function blocks to low-load units online and fine-tunes the deployment positions of the three existing function blocks. The entire process does not require downtime, and the system always maintains a load-balanced state.

[0123] In summary, the method described in this application has the following significant technical advantages compared with the methods under existing centralized distributed system architectures: Overcoming performance and reliability bottlenecks, the system achieves elastic scalability and high availability. The total system control processing capacity is expanded from the performance of a single controller to the sum of the computing power of all intelligent remote processing units, and horizontal linear expansion can be achieved by adding nodes. The control functions are deployed in a distributed manner, and the failure of a single node only affects a local loop, naturally achieving fault isolation and eliminating the single point of failure risk of centralized architecture, thus significantly improving the overall reliability of the system.

[0124] This enables software-defined deployment optimization, improving resource utilization efficiency and real-time performance. Through topology-aware intelligent deployment algorithms, functional block locations are optimized based on data flow dependencies and network topology from the configuration stage. Strongly correlated functional blocks are prioritized for deployment nearby, significantly reducing cross-node communication latency and network load. A load balancing mechanism avoids uneven resource utilization, resulting in a significant improvement in overall resource utilization efficiency.

[0125] Enhance flexibility throughout the entire lifecycle and support the agile implementation of complex strategies. The control logic is completely decoupled from the hardware, and functional blocks can be upgraded independently and maintained online. System reconstruction requires no hardware modifications. Complex and advanced control algorithms can be decomposed and distributed and deployed on nodes close to the data source, significantly reducing implementation costs and timelines, and adapting to the refined control needs of modern industry.

[0126] The layered decoupling and elastic deployment method for distributed systems provided in this application decouples the monolithic control application into multiple independently deployable logical function blocks and completes station-level coarse allocation. Based on the topology computing power parameters of the intelligent remote processing unit, the logical scheduling unit of each field control station generates a target deployment mapping table for the logical function blocks using a layered progressive optimization algorithm. The distributed intelligent remote processing units carry out the parallel execution of control logic and complete cross-node data collaboration through a deterministic real-time industrial network. At the same time, the logical scheduling unit realizes global monitoring of the operating status, system-level control coordination, and dynamic deployment adjustment for working conditions. This solves the problems of performance ceiling and single point of failure risk in the static coupling of computing resources in traditional centralized architecture, low utilization rate due to uneven resource availability caused by rigid logical deployment, and high system reconstruction cost and insufficient flexibility due to hardware function specialization. It achieves the technical effects of horizontal elastic expansion of system computing power, natural fault isolation to improve overall reliability, and global optimization of resource configuration to reduce communication loss.

[0127] Please see Figure 2 , Figure 2 This is a schematic diagram of the structure of a distributed system provided in an embodiment of this application. Figure 2 As shown, the distributed system 200 includes an engineer station 210, a deterministic real-time industrial network 220, and multiple field control stations 230; wherein each of the field control stations 230 is equipped with a logical scheduling unit 231 and multiple intelligent remote processing units 232; The engineer station 210 has a built-in logic decoupling engine, which is used to decompose the single-unit control application into multiple logic function blocks in response to receiving the single-unit control application. The logic decoupling engine determines the remote processing unit range information corresponding to each logic function block, and based on the remote processing unit range information, the logic function blocks are allocated to the corresponding field control station 230 to obtain the logic function block set of each field control station 230. The logical scheduling unit 231 is used to generate a target deployment mapping table from each logical function block in the set of logical function blocks to the intelligent remote processing unit 232 based on the topology computing power parameters of each intelligent remote processing unit 232 using a hierarchical progressive optimization algorithm; to acquire the operating status data of the intelligent remote processing unit 232 and the control instructions triggered by the engineer station 210 or operator station in real time; to perform global control coordination on the field control station 230 based on the operating status data and the control instructions; and to dynamically adjust the target deployment mapping table according to the changes in the operating conditions of the field control station 230. The intelligent remote processing unit 232 is a micro control node with independent computing capabilities. It is used to deploy the target mapping table, execute the locally carried logical function blocks in parallel in each control cycle, and complete cross-node data exchange through the deterministic real-time industrial network 220 to realize the control closed loop of the distributed system 200. The deterministic real-time industrial network 220 is used to provide deterministic low-latency communication guarantees for each node in the distributed system 200, supporting cross-node data exchange and control command transmission.

[0128] Furthermore, each of the intelligent remote processing units 232 integrates a microcontroller unit, memory, non-volatile memory, local input / output interface, and network interface. The microcontroller unit executes assigned control logic tasks, the memory stores runtime variables and intermediate calculation results, the non-volatile memory stores logic function block code and configuration parameters, the local input / output interface connects to field sensing devices and actuators, and the network interface accesses the deterministic real-time industrial network. The intelligent remote processing unit 232 is a programmable general-purpose execution unit, with functions defined by software, unlike traditional intelligent I / O with fixed functions.

[0129] Furthermore, the deterministic real-time industrial network 220 includes an inter-station deterministic industrial network and an intra-station backplane deterministic network; the inter-station deterministic industrial network is used to connect each of the field control stations 230 with the upper-level engineer station 210 and operator station, and the intra-station backplane deterministic network is used to connect the logical scheduling unit 231 within the field control station 230 with each of the intelligent remote processing units 232.

[0130] Among them, the Engineer Station 210 is an industrial-grade host computer device with built-in configuration editing software and a logic decoupling engine. Its core functions are: receiving individual control applications written by engineers, and completing the decomposition of functional blocks, resource pool allocation, and station-level coarse allocation through the logic decoupling engine; it also has human-computer interaction functions such as system monitoring, historical data storage, and parameter distribution, and is the core node for interaction between engineers and the system.

[0131] In this embodiment, the logic scheduling unit 231 is deployed in the cabinets of each field control station, employing a redundant dual configuration and equipped with an industrial-grade processor and memory storage. Its core functions are: executing deployment optimization algorithms based on station topology awareness, generating and updating deployment mapping tables; distributing function block codes and configuration parameters; monitoring the operational status of all station nodes, and performing fault migration, global coordination, and dynamic load optimization. It can also handle some global-level control logic as needed, thus fulfilling both computational and management roles.

[0132] The distributed system provided in this application decouples the monolithic control application into multiple independently deployable logical function blocks and completes station-level coarse allocation. The logical scheduling unit of each field control station generates a target deployment mapping table for the logical function blocks based on the topology computing power parameters of the intelligent remote processing unit and adopts a hierarchical progressive optimization algorithm. The distributed intelligent remote processing units carry out the parallel execution of control logic and complete cross-node data collaboration through a deterministic real-time industrial network. At the same time, the logical scheduling unit realizes global monitoring of the operating status, system-level control coordination, and dynamic deployment adjustment for working condition adaptation. This solves the problems of static coupling of computing resources in traditional centralized architecture, which has performance ceilings and single-point failure risks; rigid logical deployment leading to uneven resource utilization and low efficiency; and hardware function specialization causing high system reconstruction costs and insufficient flexibility. It achieves the technical effects of horizontal elastic expansion of system computing power, natural fault isolation to improve overall reliability, and global optimization of resource configuration to reduce communication loss.

[0133] Please see Figure 3 , Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Figure 3 As shown, the electronic device 300 includes a processor 310, a memory 320, and a bus 330.

[0134] The memory 320 stores machine-readable instructions executable by the processor 310. When the electronic device 300 is running, the processor 310 and the memory 320 communicate via the bus 330. When the machine-readable instructions are executed by the processor 310, they can perform the operations described above. Figure 1 The steps of the layered decoupling and elastic deployment method for the distributed system in the method embodiment shown are described in detail in the method embodiment, and will not be repeated here.

[0135] This application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, can perform the above-described actions. Figure 1 The steps of the layered decoupling and elastic deployment method for the distributed system in the method embodiment shown are described in detail in the method embodiment, and will not be repeated here.

[0136] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0137] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the shown or discussed mutual couplings, direct couplings, or communication connections may be through some communication interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

[0138] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0139] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0140] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a processor-executable, non-volatile, computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0141] Finally, it should be noted that the above-described embodiments are merely specific implementations of this application, used to illustrate the technical solutions of this application, and not to limit them. The scope of protection of this application is not limited thereto. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some of the technical features, within the scope of the technology disclosed in this application. Such modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for layered decoupling and elastic deployment of a distributed system, characterized in that, The distributed system includes an engineering workstation, a deterministic real-time industrial network, and multiple field control stations; wherein each field control station is equipped with a logical scheduling unit and multiple intelligent remote processing units, and the method includes: In response to the engineer station receiving a single-unit control application, the engineer station's logic decoupling engine is used to decompose the single-unit control application into multiple logic function blocks, determine the remote processing unit range information corresponding to each logic function block, and based on the remote processing unit range information, allocate the logic function blocks to the corresponding field control stations to obtain a set of logic function blocks for each field control station. Within each of the field control stations, the logical scheduling unit generates a target deployment mapping table from each logical function block in the set of logical function blocks to the intelligent remote processing unit based on the topology computing power parameters of each of the intelligent remote processing units using a hierarchical progressive optimization algorithm. Each of the intelligent remote processing units executes the locally carried logical function blocks in parallel within each control cycle based on the target deployment mapping table, and completes cross-node data exchange through the deterministic real-time industrial network to realize the control closed loop of the distributed system. The logical scheduling unit acquires the operating status data of the intelligent remote processing unit and the control commands triggered by the engineer station or operator station in real time. Based on the operating status data and the control commands, it performs global control coordination on the field control station and dynamically adjusts the target deployment mapping table according to the changes in the operating conditions of the field control station.

2. The method according to claim 1, characterized in that, The process of using the logic decoupling engine of the engineering station to decompose the monolithic control application into multiple logical function blocks and determining the remote processing unit range information corresponding to each logical function block includes: Using the logic decoupling engine of the engineering station, the monolithic control application is divided into multiple coarse-grained logic function blocks with the control logic function integrity of the monolithic control application as the decomposition granularity, and the data flow dependency relationship between each coarse-grained logic function block is determined. Based on the data flow dependency, the coarse-grained logic function blocks are initially divided according to the code size of the coarse-grained logic function blocks to obtain multiple preliminary subdivided logic function blocks, and the estimated running time of each preliminary subdivided logic function block is obtained through simulation. With the goal of maintaining a balance in the estimated running time of each of the initially subdivided logical function blocks, the initially subdivided logical function blocks are iteratively split and merged to obtain multiple logical function blocks with balanced granularity; wherein, each of the logical function blocks has clear input and output interfaces, execution cycle, computing power requirements and data flow dependency attributes; Based on the physical affiliation of the input / output channels associated with each logical function block, a three-level deployable resource pool is defined for each logical function block in descending order of communication cost, and the remote processing unit range information corresponding to each logical function block in the three-level deployable resource pool is determined. The first-level resource pool comprises the intelligent remote processing units within the same field control station, including associated input / output channels; the second-level resource pool comprises the remaining intelligent remote processing units within the same field control station; and the third-level resource pool comprises the intelligent remote processing units of the remaining field control stations. The deployment priority of the first-level resource pool is higher than that of the second-level resource pool, and the deployment priority of the second-level resource pool is higher than that of the third-level resource pool.

3. The method according to claim 1, characterized in that, Based on the remote processing unit range information, the logical function blocks are allocated to the corresponding field control stations, resulting in a set of logical function blocks for each field control station, including: Based on the remote processing unit range information, the computing power weight and input / output home station attribute information of each logical function block, as well as the data flow dependency and logical correlation between the logical function blocks, are determined. An undirected weighted dependency graph is constructed with each of the logical functional blocks as a vertex with the computing power weight and the input / output home station attribute information, the data flow dependency as an edge, and the logical correlation degree as the edge weight. The edges are traversed in descending order of edge weight, and strongly correlated vertices are merged into supernodes. Supernodes with clearly defined input / output affiliations are preferentially assigned to their corresponding field control stations in descending order of total computing power. Cross-station supernodes are assigned to the field control station corresponding to the first load. If the computing power of the affiliation field control station is insufficient, the supernode is assigned to the field control station adjacent to that field control station. Under the pre-set rigid constraints, the improved Phidusia and Matthews method is used to iteratively migrate the supernodes of every two field control stations until no migration is gained or the number of iterations reaches a threshold, so as to obtain the set of logical function blocks of each field control station.

4. The method according to claim 1, characterized in that, Within each of the field control stations, the logical scheduling unit, based on the topology computing power parameters of each intelligent remote processing unit, generates a target deployment mapping table from each logical function block in the logical function block set to the intelligent remote processing unit using a hierarchical progressive optimization algorithm, including: For each of the field control stations, decision variables and rigid hard constraints are defined; wherein, the decision variables include the deployment marker variable and the same deployment association variable of the logical function block; the rigid hard constraints include that each logical function block is deployed only in one intelligent remote processing unit, and the target computing power of a single intelligent remote processing unit does not exceed the upper limit of computing power after reserving the safety margin, and the logical function block is deployed only within the resource pool corresponding to the remote processing unit range information, and the same deployment association variable is consistent with the deployment marker variable; Based on the topology computing power parameters of each intelligent remote processing unit, under the rigid hard constraint, the decision variables are optimized in descending order of the preset hierarchical progressive optimization priority, and the objective optimization result of each layer of optimization is added as an equality constraint to the solution process of the next layer of optimization. In response to the completion of the single-objective optimization solution, the values ​​of the decision variables corresponding to the decision variables are obtained, and based on the values ​​of the decision variables, an initial deployment mapping table from the logical function block to the intelligent remote processing unit is generated; Each logical function block is sent to the corresponding intelligent remote processing unit and a trial run is started to obtain trial run data corresponding to each logical function block. Based on the trial run data, the initial deployment mapping table is iteratively optimized to obtain a target deployment mapping table from each logical function block in the set of logical function blocks to the intelligent remote processing unit.

5. The method according to claim 4, characterized in that, The hierarchical progressive optimization priority includes a first hierarchical priority, a second hierarchical priority, a third hierarchical priority, and a fourth hierarchical priority; the first hierarchical priority is to maximize the sum of the association weights of the logical functional blocks deployed in the same intelligent remote processing unit; the second hierarchical priority is to maximize the proportion of the logical functional blocks allocated to the first-level resource pool; the third hierarchical priority is to minimize the load rate difference of each intelligent remote processing unit; and the fourth hierarchical priority is to minimize the total communication traffic across the intelligent remote processing units. The step of distributing each logical function block to the corresponding intelligent remote processing unit and initiating trial operation, obtaining trial operation data corresponding to each logical function block, and iteratively optimizing the initial deployment mapping table based on the trial operation data to obtain a target deployment mapping table from each logical function block in the set of logical function blocks to the intelligent remote processing unit includes: The executable code and configuration parameters of each logical function block are sent to the corresponding intelligent remote processing unit using the deterministic real-time industrial network to start trial operation; During the trial operation, the actual operating performance datasets corresponding to each intelligent remote processing unit and each logical function block are collected in real time. Based on the actual operating performance datasets, the hierarchical progressive optimization algorithm is repeatedly called to update the initial deployment mapping table until the fluctuation of the actual operating performance dataset is lower than the preset fluctuation threshold, thereby obtaining the target deployment mapping table.

6. The method according to claim 1, characterized in that, The process involves each intelligent remote processing unit executing its locally hosted logical function blocks in parallel within each control cycle based on the target deployment mapping table, and completing cross-node data exchange through the deterministic real-time industrial network to achieve a control closed loop for the distributed system, including: Each of the aforementioned intelligent remote processing units acquires the field input signals locally connected to the intelligent remote processing unit through a local I / O interface, and performs signal conditioning, isolation, and analog-to-digital conversion on the field input signals to obtain local input data; The logic function block code loaded in the memory of the intelligent remote processing unit is invoked. Based on the local input data and the input data received across nodes in the previous cycle, the local control logic operation of the logic function block is executed to obtain the intermediate result data and the target output value calculated by the logic function block. Based on the global clock synchronization signal of the distributed system, the intermediate result data of the local logical function block is sent to the corresponding intelligent remote processing unit through the deterministic real-time industrial network according to the predefined data flow dependency relationship, while receiving input data sent by the other intelligent remote processing units. The target output value is converted from digital to analog, isolated, and powered, and then output to the field actuator that is locally connected to the intelligent remote processing unit to realize the control closed loop of the distributed system.

7. The method according to claim 1, characterized in that, The logical scheduling unit acquires in real time the operating status data of the intelligent remote processing unit and the control commands triggered by the engineer station or operator station. Based on the operating status data and the control commands, it performs global control coordination on the field control station and dynamically adjusts the target deployment mapping table according to the changes in the operating conditions of the field control station, including: The logical scheduling unit acquires the running status data of the intelligent remote processing unit in real time, and generates a global running view based on the running status data; wherein, when a fault is detected in the intelligent remote processing unit, the logical function block carried by the intelligent remote processing unit is migrated to the intelligent remote processing unit in the same-site backup resource pool to achieve fault isolation; The logic scheduling unit obtains control commands triggered by the engineer or operator station based on the global operation view in real time. Based on the control commands, it sequentially performs global control mode switching, set value synchronization and configuration parameter consistency verification on the field control station, and ensures that the control logic of each field control station is coordinated and consistent. When the operating conditions of the field control station cause the distributed system to become unbalanced, the hierarchical progressive optimization algorithm is invoked again to migrate the non-strongly related logical function blocks online without interrupting control operation, so as to dynamically adjust the target deployment mapping table and maintain the load balance of the distributed system.

8. A distributed control system, characterized in that, The distributed system includes an engineering workstation, a deterministic real-time industrial network, and multiple field control stations; each of the field control stations is equipped with a logical scheduling unit and multiple intelligent remote processing units. The engineering station has a built-in logic decoupling engine, which is used to decompose the single-unit control application into multiple logic function blocks in response to receiving the single-unit control application. The engine determines the remote processing unit range information corresponding to each logic function block, and allocates the logic function blocks to the corresponding field control stations based on the remote processing unit range information, thereby obtaining a set of logic function blocks for each field control station. The logical scheduling unit is used to generate a target deployment mapping table from each logical function block in the set of logical function blocks to the intelligent remote processing unit based on the topology computing power parameters of each intelligent remote processing unit and using a hierarchical progressive optimization algorithm; to acquire the operating status data of the intelligent remote processing unit and the control instructions triggered by the engineer station or operator station in real time; to perform global control coordination on the field control station based on the operating status data and the control instructions; and to dynamically adjust the target deployment mapping table according to the changes in the operating conditions of the field control station. The intelligent remote processing unit is a micro control node with independent computing capabilities. It is used to deploy a mapping table based on the target, execute the locally carried logical function blocks in parallel during each control cycle, and complete cross-node data exchange through the deterministic real-time industrial network to realize the control closed loop of the distributed system. The deterministic real-time industrial network is used to provide deterministic low-latency communication guarantees for each node in the distributed system, supporting cross-node data exchange and control command transmission.

9. An electronic device, characterized in that, include: The device includes a processor, a memory, and a bus. The memory stores machine-readable instructions executable by the processor. When the electronic device is running, the processor communicates with the memory via the bus. The machine-readable instructions are executed by the processor to perform the steps of the layered decoupling and elastic deployment method for a distributed system as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, performs the steps of the layered decoupling and elastic deployment method for a distributed system as described in any one of claims 1 to 7.

Images