Network cooperative control and communication scheduling method for solid-state transformer cluster

Abstract

The invention discloses a network cooperative control and communication scheduling method for a solid-state transformer cluster, and belongs to the technical field of power system control and communication. The method comprises the following steps: realizing panoramic perception and task generation of a solid-state transformer cluster state through a cloud control center and an edge aggregator; using an improved hybrid heuristic algorithm to carry out collaborative optimization scheduling on communication and computing resources required by a control task; a layered instruction issuing mechanism is adopted; the solid-state transformer locally executes a hybrid strategy integrating autonomous communication-free control and communication-assisted optimization control, and has a mode smooth switching capability; and finally adjusting a scheduling strategy based on dynamic feedback. The system comprises a cloud control center, an edge aggregator, a solid-state transformer cluster and a high-speed communication network. According to the method, deep coordination of control and communication is realized, and the utilization efficiency of communication resources and the adaptive capability of the system are remarkably improved while the reliability and economy of cluster operation are guaranteed.

Description

Technical Field

[0001] This invention relates to the field of power system operation control and communication technology, and in particular to a network collaborative control and communication scheduling method for solid-state transformer clusters. Background Technology

[0002] With the integration of high proportions of renewable energy and diversified loads, future power distribution systems will take on a complex form, consisting of multiple AC / DC microgrids interconnected through solid-state transformers. Solid-state transformers, equipped with AC / DC ports at multiple voltage levels, are key devices for achieving flexible energy conversion and distribution. Clustering and interconnecting multiple solid-state transformers at both the physical and information levels enables energy sharing and synergistic optimization over a wider range, which is of great significance for improving system reliability, resilience, and the capacity to absorb renewable energy.

[0003] Current related technologies mainly focus on two independent directions: one is the control strategy of the solid-state transformer itself and the power coordination of the AC / DC microgrids it connects to; the other is the general communication task scheduling for power Internet of Things (IoT) scenarios. For example, existing technology one (CN115912519A) provides a power coordination control method based on solid-state transformers. By adopting specific droop control strategies at the low-voltage, isolation, and medium-voltage levels, it achieves communication-free power coordination and seamless mode switching between AC / DC microgrids and with the main grid. Although this method improves reliability, its "completely communication-free" premise limits the realization of the cluster's global optimization capabilities. Existing technology two (CN112653203A) proposes an autonomous power coordination control method for AC / DC hybrid power grids based on a single solid-state transformer, achieving power balance among four subgrids. However, it does not address cluster scenarios composed of multiple solid-state transformers, nor does it consider information interaction issues within the cluster and with upper-level networks. Existing technology 3 (CN114327878A) proposes a cloud-edge collaborative communication scheduling method for panoramic monitoring of UHV converter stations. It uses an improved artificial bee colony algorithm to optimize task allocation and solves the problems of transmission delay and uneven load. However, its application scenario and optimization objectives are fundamentally different from the operation and control of solid-state transformer clusters.

[0004] In summary, existing technologies have the following shortcomings: 1) For the collaborative operation of solid-state transformer clusters, there is a lack of a systematic solution that combines low-level fast power control with high-level optimization decision-making; 2) In cluster scenarios, limited communication resources are not fully considered in how to dynamically and efficiently serve diverse control and status awareness tasks, so as to optimize communication costs and reliability while ensuring control performance; 3) There is a lack of a mechanism that can adaptively adjust control modes and communication strategies according to the system's operating status. Therefore, there is an urgent need for an integrated solution for solid-state transformer clusters that can achieve deep coupling between networked collaborative control and efficient communication scheduling. Summary of the Invention

[0005] To address the technical problems existing in the prior art, this invention provides a network collaborative control and communication scheduling method for solid-state transformer clusters, applicable to a networked system including a cloud control center, multiple edge aggregators, and a cluster composed of multiple solid-state transformers, comprising the following steps: S101. Status Panoramic Perception and Task Generation: Based on the edge-cloud three-level architecture, the system collects the operating status data of the solid-state transformer cluster, performs global situation fusion in the cloud control center to generate a global situation panoramic view, and dynamically generates planned optimization tasks and event-driven control tasks based on the global situation panoramic view; S102. Coordinated Scheduling of Communication and Computing Resources: Establish a scheduling model with the optimization objectives of minimizing task completion delay, balancing resource load, and reducing system energy consumption. Use a hybrid heuristic algorithm to jointly solve the computation offloading decision, communication routing decision, and resource allocation decision for tasks, and generate a dynamic scheduling strategy table. S103. Layered collaborative control command issuance and execution: According to the dynamic scheduling strategy table, the control command is issued to the target solid-state transformer after security verification and coordination by the edge aggregator; S104. Execution of local hybrid control strategy for solid-state transformer: The local controller of the solid-state transformer runs a dual-mode adaptive control strategy, including an autonomous non-communication control mode based on the interconnected droop control network and a communication-assisted optimization control mode that introduces upper-level optimization instructions on this basis, and achieves smooth switching between modes through a mode manager; S105. Dynamic Feedback and Scheduling Strategy Adjustment: Monitor the execution effect of control commands and changes in system state. When performance indicators deviate from expectations or the system state changes abruptly, trigger a new round of collaborative scheduling of communication and computing resources to form closed-loop control.

[0006] As a preferred embodiment of this application, in step S101, the three-tier architecture of end-edge-cloud specifically comprises: each solid-state transformer on the end side generates a snapshot of its device status, including core electrical status, device health status, and the status of the microgrid under its jurisdiction; the edge aggregator on the edge side performs timestamp alignment, data cleaning, and regional feature extraction on the snapshots of each solid-state transformer in its sub-cluster; the cloud control center on the cloud side integrates all edge data to form the global status panorama; the task generation includes planned optimization tasks based on predicted data and event-driven control tasks triggered by real-time events, and each task is assigned standardized attributes including task ID, type, priority, target device list, and expected maximum end-to-end latency.

[0007] As a preferred embodiment of this application, in step S102, the hybrid heuristic algorithm integrates the global exploration mechanism of the artificial bee colony algorithm and the local development mechanism of the particle swarm optimization algorithm; the solution encoding of the algorithm is a multi-dimensional vector containing task allocation identifier, communication path identifier and resource allocation vector; the algorithm adopts an adaptive weight adjustment mechanism, increasing the weight of random exploration in the early stage of iteration and increasing the weight of learning from the historical optimal solution and the global optimal solution in the later stage of iteration.

[0008] As a preferred embodiment of this application, in step S104, the autonomous communication-free control mode establishes a correlation between the droop coefficients of the low-voltage DC bus voltage, low-voltage AC frequency, medium-voltage DC bus voltage, and medium-voltage AC frequency to construct a voltage-frequency-power droop network, enabling the cluster to achieve self-organized power balance under communication-free conditions; the communication-assisted optimization control mode adopts a reference value fusion algorithm to perform weighted calculations of the upper-level optimization instructions and the local droop control output, and the fusion weights are smoothly transitioned according to the communication status and the validity of the instructions.

[0009] As a preferred embodiment of this application, in step S105, the conditions for triggering re-optimization include unsatisfactory control performance, sudden changes in system environment state, deterioration of key resource performance, or the insertion of a new task with higher priority; the re-optimization process is based on the latest system state and resource status, and generates a new dynamic scheduling strategy table to achieve adaptive adjustment.

[0010] This application also provides a network-coordinated control and communication scheduling system for solid-state transformer clusters, characterized in that it specifically includes: The cloud control center includes a global status database, a global optimization decision-making module, a resource collaborative scheduling module, and a task generation and instruction management module, which are used to perform global situational awareness, task generation, resource scheduling, and optimization decision-making. Multiple edge aggregators are communicatively connected to the cloud control center. Each edge aggregator includes a regional data aggregation module and a regional coordination module, which are used for data aggregation, instruction verification and forwarding, and have the edge autonomy function of collaborative optimization based on local data when communication is interrupted. A solid-state transformer cluster consists of multiple solid-state transformers, each of which has a built-in local controller. The controller includes a dual-mode control logic unit and a mode manager for executing the dual-mode adaptive control strategy. A high-speed, reliable communication network connects the cloud control center, edge aggregator, and solid-state transformer cluster, providing a communication channel for data and command transmission.

[0011] As a preferred embodiment of this application, the edge aggregator uses an industrial-grade embedded computer or a ruggedized server as hardware; the local controller of the solid-state transformer adopts a hardware architecture that combines DSP and FPGA, wherein the FPGA is used to generate PWM pulse modulation signals and the DSP is used to run control algorithms; the high-speed and reliable communication network supports fiber optic ring networks and 5G wireless private networks, has redundant communication paths, and supports differential quality of service assurance.

[0012] Compared to existing technologies, the advantages of this invention are as follows: The network collaborative control and communication scheduling scheme for solid-state transformer clusters provided in this application constructs a closed-loop control framework with deep "control-communication" collaboration, unifying and optimizing physical layer collaborative control and information layer resource scheduling, and adopting a local dual-mode adaptive control strategy. This not only improves the system's global control efficiency, resource utilization efficiency, and robustness in dealing with communication interruptions under complex operating scenarios, but also achieves low-latency, high-reliability execution of critical control tasks and seamless switching between different control modes, significantly enhancing the overall operating performance and adaptability of the solid-state transformer cluster. Attached Figure Description

[0013] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0014] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0015] Figure 1 The flowchart is shown below for the network collaborative control and communication scheduling method provided in the embodiments of the present invention.

[0016] Figure 2 This is a schematic diagram illustrating the state panorama perception and task generation provided in an embodiment of the present invention.

[0017] Figure 3 This is a schematic diagram illustrating the principle of collaborative scheduling of communication and computing resources provided in an embodiment of the present invention.

[0018] Figure 4 This is a schematic diagram illustrating the hierarchical collaborative control command issuance and execution provided in an embodiment of the present invention.

[0019] Figure 5 This is a schematic diagram illustrating the execution of a local hybrid control strategy for solid-state transformers provided in an embodiment of the present invention.

[0020] Figure 6This is a schematic diagram illustrating the dynamic feedback and scheduling strategy adjustment provided in an embodiment of the present invention.

[0021] Figure 7 The hardware composition and network architecture diagram of the network collaborative control and communication scheduling system provided in the embodiments of the present invention. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] It should be noted that all directional indicators (such as up, down, left, right, front, back, etc.) in the embodiments of the present invention are only used to explain the relative positional relationship and movement of each component in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indicator will also change accordingly.

[0024] Furthermore, the descriptions involving "first," "second," etc., in this invention are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. Additionally, the technical solutions of the various embodiments can be combined with each other, but only on the basis of being achievable by those skilled in the art. When the combination of technical solutions is contradictory or impossible to implement, such a combination of technical solutions should be considered non-existent and not within the scope of protection claimed by this invention.

[0025] Example 1 This invention provides a network-coordinated control and communication scheduling method for solid-state transformer clusters, constructing a complete closed loop from global situational awareness, intelligent decision generation, dynamic resource scheduling to precise local execution. The core innovation of this method lies in unifying the physical layer coordinated control problem and the information layer resource optimization scheduling problem of solid-state transformer clusters under a single framework, achieving deep "control-communication" collaboration. Figure 1 As shown, the specific steps include: S101. Status Panorama Perception and Task Generation: Based on a three-tiered edge-cloud architecture, operational status data of the solid-state transformer cluster is collected. Global situational awareness is fused at the cloud control center to generate a global situational panorama. Based on this panorama, planned optimization tasks and event-driven control tasks are dynamically generated. The goal of this step is to construct a real-time digital image of the cluster's operation and, based on this image and optimization objectives, intelligently define the problems to be solved, providing precise input for subsequent decision-making and scheduling. Figure 2 As shown, the state panoramic perception and task generation are specifically as follows: 1) Multi-source data acquisition and hierarchical aggregation We adopt a three-tiered data collection and aggregation architecture of "end-edge-cloud" to ensure the comprehensiveness, real-time performance and reliability of the data.

[0026] End-side (Solid-state transformer): Each solid-state transformer acts as a smart edge node, with its local controller sampling electrical quantities at a high frequency and calculating and encapsulating a standardized "device state snapshot" at medium intervals. This snapshot contains: Core electrical status, including but not limited to the voltage, current, active / reactive power, frequency / phase of each port (low-voltage DC LD, low-voltage AC LA, medium-voltage DC MD, medium-voltage AC MA); transmission power and efficiency of the dual active bridge (DAB).

[0027] Equipment health status includes power module temperature, heat dissipation status, protection alarm signs, and controller operating mode (autonomous / optimized).

[0028] The status of the microgrid under its jurisdiction is obtained through internal communication or estimated based on local measurements, including the real-time state of charge of the energy storage units on the DC microgrid side, the instantaneous output and ultra-short-term forecast curves of distributed power sources, and the power of critical loads.

[0029] Edge Aggregator: Each edge aggregator manages a physically or electrically tightly coupled sub-cluster of solid-state transformers. Its primary responsibility is "regional status aggregation." It periodically collects "device status snapshots" from each subordinate solid-state transformer and performs the following operations: Data alignment and cleaning: All data is timestamped based on high-precision clock synchronization, and rules are applied to verify data validity.

[0030] Regional feature extraction is performed to calculate the aggregated power, average voltage / frequency, key imbalance indicators, average latency of communication links, and packet loss rate of the region.

[0031] Local caching and preprocessing: The processed data is cached in a local circular buffer, which serves two purposes: for quick local decision-making and for preparing for uploading.

[0032] Cloud-side (Cloud Control Center): Receives aggregated data streams uploaded from all edge aggregators. The data processing module of the cloud center is responsible for "global situational fusion," integrating all regional data into a unified spatiotemporal database to form a "global situational panorama" covering the entire network's electrical status, equipment status, communication network status, and renewable energy / load forecast information.

[0033] 2) Intelligent task generation mechanism The cloud control center's "task generation engine," based on a "global situational overview," dynamically generates two types of tasks with different time scales and objectives, combined with preset operational goals, constraints, and expert rules. Planned optimization tasks ( Based on forecast data, the goal is to achieve long-term optimization objectives such as economy and safety. It mainly includes economic dispatch tasks and preventive voltage control tasks. The economic dispatch task has an optimization cycle of 24 hours in the future and aims to minimize the total operating cost of the cluster (including electricity purchase cost, network loss and energy storage loss). The input is ultra-short-term wind and solar power output forecast, load forecast and time-of-use electricity price. The output is the day-ahead power plan curve of each solid-state transformer in each future time period. This task is computationally complex but not sensitive to time delay. It is usually executed in a rolling cycle of 15 minutes or 1 hour. The preventive voltage control task has an optimization cycle of 5-15 minutes in the future and aims to prevent voltage overruns and optimize reactive power distribution. The input is the state estimation results and ultra-short-term forecast. The output is the reactive power output or voltage reference setpoint of each solid-state transformer.

[0034] Event-driven control tasks ( Triggered by specific events detected in real time, this system aims to achieve rapid response to maintain system stability. It mainly includes four typical scenarios: When a sudden drop in main grid voltage or a feeder fault is detected, a fault-crossing collaborative support task is immediately triggered. By quickly adjusting the output characteristics of solid-state transformers in non-faulty areas, it provides voltage and frequency support for grid connection points or critical loads. This task has extremely high latency requirements for communication and control commands. If a severe and unpredictable fluctuation in the power of the source loads within the cluster leads to a serious power imbalance, a power mutation smoothing task is initiated. By quickly adjusting the power reference values ​​of each solid-state transformer, power balance is restored within seconds. When the network monitoring system detects a critical communication link interruption or severe performance degradation, a communication network self-healing task is triggered. By replanning the transmission path for high-priority control command streams, it ensures the reliability of the control link.

[0035] Each generated task Each task will be assigned a standardized set of attribute descriptions, forming a "task work order," including: {task ID, type (G / E), priority (P1-P5), source data location, target device list, expected maximum end-to-end latency, computational complexity estimate, and data size}. This provides clear quantitative input for subsequent resource scheduling.

[0036] S102. Coordinated Scheduling of Communication and Computing Resources: Establish a scheduling model with the optimization objectives of minimizing task completion latency, balancing resource load, and reducing system energy consumption. A hybrid heuristic algorithm is used to jointly solve for task computation offloading decisions, communication routing decisions, and resource allocation decisions, generating a dynamic scheduling strategy table. This step is the intelligent transportation hub connecting upper-level decision-making and lower-level execution. Its core idea is to select the correct execution location and communication path, and allocate the correct resources for the correct task at the correct time. Figure 3 As shown, the coordinated scheduling of communication resources and computing resources is specifically as follows: 1) Modeling the scheduling problem The scheduling problem is formalized as a constrained optimization problem with the objective of maximizing the total utility of the system. The decision variables define the complete scheduling scheme: The calculation of offloading decisions, a binary variable matrix, determines whether each task is executed in the cloud or at an edge aggregator.

[0037] Communication routing decisions select a specific end-to-end network path for the data flow of each task (uplink status data and / or downlink control commands).

[0038] Resource allocation decisions allocate specific computing resources (such as the number of CPU cores and memory) and communication resources (such as link bandwidth share) to each task.

[0039] The optimization objective is a multi-objective function, aiming to: Minimize the overall task completion delay to ensure the timeliness of emergency control commands; Maximize the balance of resource utilization, avoid overload of some links or nodes, and improve the overall robustness of the system; Minimize the total system energy consumption, including communication transmission energy consumption and computing energy consumption, to improve operating economy.

[0040] The multi-objective problem is transformed into a single objective using a weighted summation method. Constraints ensure the feasibility of the solution, including the upper limit of node computing power, link bandwidth capacity, task deadline, and data flow conservation.

[0041] 2) Execution flow of the hybrid heuristic scheduling algorithm Since this problem is a typical NP-hard combinatorial optimization problem, this invention designs a hybrid heuristic scheduling algorithm. This algorithm simulates the foraging behavior of a bee colony (global exploration) and the social learning behavior of a particle swarm (local exploitation), and has a strong optimization capability.

[0042] Initialize by randomly generating a swarm of "bees / particles", each representing a randomly generated scheduling scheme (i.e., a potential solution).

[0043] Leading the bee / particle exploration, each individual attempts to search within the solution space. The search strategy is hybrid: One search direction is "random exploration," inspired by the artificial bee colony algorithm, which discovers new regions by comparing them with other random individuals, thus preventing premature convergence.

[0044] Another search direction is "experience-oriented," inspired by the particle swarm optimization algorithm, which guides individuals to learn from their own historical best position and the current global best position of the group, thereby accelerating convergence to the region of high-quality solutions.

[0045] Adaptive weights are introduced. In the early stages of iteration, the weight of "random exploration" is increased to broadly search the solution space; in the later stages of iteration, the weight of "experience-oriented" is increased to finely develop the optimal solution region.

[0046] Following the lead bee's selection and development, based on the "fitness" (i.e., the quality of the objective function value) of each solution, a roulette wheel strategy is used to select a batch of high-quality solutions for focused development (similar to bees following a lead bee to a high-quality nectar source). A more detailed search is then performed on the neighborhood of the selected solutions in order to find even better solutions.

[0047] The scout bee mechanism determines that a solution is trapped in a local optimum if it fails to improve after multiple iterations. This solution is then discarded and replaced by a newly generated, randomly selected solution, thus maintaining population diversity and preventing the population from falling into local optima.

[0048] Convergence and Output: The algorithm stops after iterating to the maximum number of times and outputs the globally optimal scheduling scheme found in the past. This scheme is decoded into a detailed "dynamic scheduling strategy table," which clearly specifies: where each task is computed, how data is transmitted, how instructions are issued, how many resources each stage occupies, and the precise time window arrangement.

[0049] S103, Layered Collaborative Control Command Issuance and Execution: Based on the dynamic scheduling strategy table, control commands are issued to the target solid-state transformer after security verification and coordination by the edge aggregator. This step is responsible for safely, reliably, and accurately transmitting the intent of the "decision-making brain" to the "execution limbs" through the optimized "information nerves." For example... Figure 4 As shown, the issuance and execution of the hierarchical collaborative control commands are as follows: 1) Command generation and secure encapsulation: The cloud control center or the edge aggregator performing the computation generates specific control commands based on the optimized computation results. The command content is encapsulated into standardized messages, including command sequence number, timestamp, command type, parameter value, validity period, and digital signature (for tamper prevention and authentication).

[0050] 2) Layered distribution based on scheduling strategy Command issuance strictly follows the communication path specified in the "Dynamic Scheduling Strategy Table," even for extremely urgent situations. (e.g., fault support) The policy table may specify a low-latency, high-priority direct or backup path.

[0051] In most cases, instructions are first issued to the edge aggregator in the target area. The edge aggregator acts as both a "security sentinel" and a "coordinator." Security verification verifies the digital signature and validity of the command.

[0052] Local coordination involves checking for conflicts in instructions issued to multiple solid-state transformers within the region (e.g., whether the total power exceeds the limit), and making minor adjustments if necessary.

[0053] Protocol conversion transforms generic protocol messages from the cloud into industrial protocol messages that can be recognized by solid-state transformer devices.

[0054] 3) Reliable transmission and acknowledgment: A reliable transmission mechanism (such as TCP or application layer acknowledgment mechanism) is adopted. After issuing the instruction, the sender starts a timer. If no acknowledgment signal is received from the receiver within the timeout period, retransmission or route switching is initiated according to the backup plan in the scheduling policy table.

[0055] S104. Execution of Local Hybrid Control Strategy for Solid-State Transformers: The local controller of the solid-state transformer operates a dual-mode adaptive control strategy, including an autonomous, non-communication-based control mode based on an interconnected droop control network and a communication-assisted optimization control mode that introduces upper-level optimization instructions. Smooth switching between modes is achieved through a mode manager. This step is the final implementation stage of the control action. Each solid-state transformer's local controller, as an autonomous intelligent agent, runs a "dual-mode adaptive control strategy," perfectly combining the autonomous robustness under non-communication conditions with global optimality under communication conditions. For example... Figure 5 As shown, the local hybrid control strategy for the solid-state transformer is executed as follows: 1) Autonomous, non-communication control mode (basic layer): This mode serves as the physical inertia and safety foundation of the system, ensuring continuous operation under any circumstances, including complete communication interruption. At its core is a meticulously designed interconnected droop control network.

[0056] Control principle: By associating electrical quantities (voltage, frequency) at different ports with a drooping coefficient, a virtual "power-voltage-frequency" mapping network is established. For example: low voltage DC voltage The increase in [a certain value] means that the local DC microgrid has a power surplus, which will be addressed through [a certain factor]. This leads to low-voltage AC frequency Increase.

[0057] frequency The increase in voltage will be transmitted to adjacent solid-state transformers through the AC bus, and ultimately affect the medium-voltage AC frequency. .

[0058] Similarly, medium-voltage DC voltage With medium voltage AC frequency Also through coefficients Related.

[0059] Autonomous effect: This mapping allows adjacent solid-state transformers to automatically determine the power abundance or deficiency in the region simply by sensing the voltage and frequency of the common connection point, without needing to communicate. They can then spontaneously provide power support or absorption, achieving basic power balance and voltage support at the cluster level, much like an "intelligent organism".

[0060] 2) Communication-assisted optimized control mode (optimization layer) When the local controller receives and verifies the upper-layer optimization instruction, it switches to this mode while ensuring a smooth transition.

[0061] The instruction fusion algorithm does not simply replace the autonomous control output, but rather employs strategies such as "reference value fusion" or "droop coefficient correction." For example, it integrates the optimized power reference value sent down from the upper layer. Power calculated by autonomous drooping Perform weighted fusion: The weight β smoothly transitions from 1 to a small value close to 0 during the switching process, achieving a disturbance-free switching.

[0062] Advanced function execution allows the execution of specific advanced control algorithms issued by the upper layer, such as harmonic compensation and resonance suppression. The controller will load the corresponding algorithm module and execute it.

[0063] Seamless pre-synchronization: When operating in parallel with the main grid or other solid-state transformers, the built-in pre-synchronization module actively adjusts the amplitude and phase of the unit's output voltage to synchronize it with the target bus. Only after the synchronization conditions are met will the circuit breaker close and fully accept optimization commands, completely eliminating parallel impact.

[0064] 3) Mode Manager: An intelligent state machine that continuously monitors communication link quality, command validity, and local electrical status. Based on a set of explicit rules (e.g., normal communication and valid command? -> optimized mode; communication timeout or invalid command? -> autonomous mode), it automatically determines the current operating mode and manages the smooth switching process to ensure the system always operates in the most suitable state.

[0065] S105. Dynamic Feedback and Scheduling Strategy Adjustment: Monitor the execution effect of control commands and changes in system state. When performance indicators deviate from expectations or the system state changes abruptly, trigger a new round of collaborative scheduling of communication and computing resources, forming a closed-loop control. This step constitutes the system's "learning and adaptation loop," ensuring that the method can dynamically respond to internal and external changes. Figure 6 As shown, the dynamic feedback and scheduling strategy adjustment are specifically as follows: 1) Closed-loop monitoring: The cloud control center and edge aggregator continuously compare the "system state prediction before the instruction is issued" with the "actual system state feedback after the instruction is executed".

[0066] 2) Performance evaluation and triggering conditions: Calculate deviations in key performance indicators, such as power point tracking error and voltage recovery time. A new round of scheduling optimization is immediately triggered (jumping back to S102) when the following conditions occur: The control measures failed to meet the standards, and the actual results deviated significantly from expectations. Sudden changes in environmental conditions, such as new faults, network topology changes, or drastic and unpredictable power fluctuations; Deteriorating resource conditions, excessive load on critical computing nodes, or significant degradation in communication link performance; New tasks are inserted, and higher-priority urgent tasks are generated.

[0067] 3) Iterative evolution of strategies: New scheduling optimization processes will be based on the latest and most realistic system state and resource conditions. This means that the scheduling strategy is "time-varying" and "adaptive." For example, if a sharp increase in latency is detected on a fiber optic link, the new scheduling strategy will immediately migrate traffic for critical control commands to a backup 5G link. Through this continuous dynamic adjustment, the entire system possesses strong anti-disturbance capabilities and environmental adaptability.

[0068] Example 2 This invention also provides a network-based collaborative control and communication scheduling system for solid-state transformer clusters, which is a comprehensive hardware and software system integrating sensing, decision-making, communication, and execution. For example... Figure 7 As shown, the system specifically includes interconnected systems: Cloud Control Center As the "brain" of the entire collaborative control system, the cloud control center is usually deployed in the power company's main data center or regional dispatch center, undertaking the core functions of overall decision-making, resource scheduling, and system monitoring.

[0069] The hardware relies on a robust infrastructure consisting of high-performance server clusters, high-speed network switches, large-capacity storage arrays, and forward and reverse isolation devices that ensure secure connections to the production control area to provide the necessary computing, storage, and security isolation capabilities.

[0070] At the software level, the cloud control center comprises a series of key functional modules that collectively enable its intelligent decision-making and management capabilities. The global state database, serving as the system's data foundation, uses time-series and relational databases to store historical and real-time data uploaded by all edge aggregators, supporting the construction of an accurate digital twin model of the system. The global optimization decision module is the core algorithm engine, typically including a state estimation and prediction submodule for data cleaning and ultra-short-term prediction, an optimization calculation submodule with embedded multiple planning solvers to handle periodic optimization tasks, and a rapid analysis submodule for quickly responding to event-triggered tasks. The communication resource management module maintains the global communication network topology, monitors the bandwidth, latency, and other performance indicators of each link in real time, and provides crucial network state input for the scheduling algorithm. The resource collaborative scheduling module runs advanced scheduling algorithms, outputting the optimal scheduling strategy table based on the task list and real-time network status. Furthermore, the task generation and instruction management module manages the task queue and completes the encapsulation, signing, and distribution of instructions according to the scheduling strategy. The system also provides operators with a comprehensive human-computer interaction and visualization platform for panoramic monitoring, policy issuance, parameter configuration, and report generation.

[0071] Regarding external interfaces, the cloud control center connects to various edge aggregators through the power dispatch data network to issue commands and collect status data; simultaneously, it interacts with the information management area through a firewall to receive external information such as weather and electricity prices, thereby making more comprehensive decisions. This architecture enables unified scheduling and management of resources between the control system and the intelligent computing platform.

[0072] Edge aggregator Edge aggregators play a crucial bridging role between the "cloud" and the "edge" in the collaborative control system of solid-state transformer clusters. They are typically deployed in substations, switching stations, or important power distribution nodes, undertaking the core task of connecting the upper and lower layers.

[0073] In terms of hardware configuration, the edge aggregator uses industrial-grade embedded computers or ruggedized servers, equipped with multi-core CPUs, adequate memory, and solid-state drives to ensure stable operation in harsh industrial environments. Its communication interfaces are diverse, including Gigabit Ethernet, Industrial Ethernet, fiber optic, and 5G CPE, to enable heterogeneous network access with the cloud and underlying solid-state transformers.

[0074] At the software functionality level, the edge aggregator comprises several core modules. The regional data aggregation module is responsible for periodically collecting and caching data from subordinate solid-state transformers, performing local time synchronization and simple preprocessing (such as limit checks), and uploading data according to cloud center requirements or event triggers. The regional coordination module, as the core of edge intelligence, is responsible for command verification, coordination, and forwarding when cloud-edge communication is normal; when cloud-edge communication is interrupted or latency is excessive, it automatically upgrades to an "edge autonomous brain," using built-in lightweight local optimization algorithms (such as consensus-based distributed algorithms), utilizing locally stored historical models and real-time regional data to independently calculate and issue collaborative control commands, ensuring autonomous regional operation. The communication gateway module implements the conversion between different communication protocols (e.g., converting MQTT / HTTP messages from the cloud into ModbusTCP / IEC61850 messages to the solid-state transformers), and manages local communication links and packet routing and forwarding. Furthermore, the local resource monitor continuously monitors the device's CPU, memory, and storage usage, as well as the quality of the communication link to each solid-state transformer, reporting this information to the cloud center as an important basis for resource scheduling.

[0075] In terms of external interfaces, the edge aggregator connects to the cloud control center via the power backbone network (such as fiber optics) uplink and to the local solid-state transformer cluster via an industrial-grade LAN downlink. This design enables it to efficiently integrate data from end devices and achieve intelligent collaboration between the cloud and the edge.

[0076] Solid-state transformer cluster The solid-state transformer cluster, as the core execution unit of the system, consists of multiple intelligent solid-state transformer devices with standardized communication and control interfaces. Its core value lies in integrating multiple functions such as traditional transformer, reactive power compensation, and active filtering into one, realizing the leap from "passive power supply equipment" to "active smart grid node".

[0077] Solid-state transformers typically employ a fully functional three-stage topology. The input stage (AC / DC) rectifies medium-voltage AC to DC; the isolation stage (DC / DC) uses a high-frequency transformer for electrical isolation and voltage conversion, which is crucial for significantly reducing size and weight; the output stage (DC / AC) inverts the electrical energy into the desired form of AC or DC. This design allows it to simultaneously provide four ports: medium-voltage AC, medium-voltage DC, low-voltage AC, and low-voltage DC, perfectly adapting to the needs of mixed AC / DC distribution networks. Its power semiconductor devices commonly use silicon carbide MOSFETs or IGBTs, especially silicon carbide devices, which, due to their high voltage withstand and low loss characteristics, can support system operation at higher frequencies, thereby improving efficiency and power density.

[0078] The local controller is the "brain" that enables its intelligent and autonomous operation capabilities, typically employing a collaborative computing architecture of "DSP+FPGA". In this architecture, a high-performance digital signal processor is responsible for executing complex control algorithm calculations, while the field-programmable gate array (FPGA) leverages its parallel processing capabilities to efficiently generate multiple precise PWM pulse modulation signals. The controller's hardware foundation includes high-speed AD sampling circuits, memory, and various communication interfaces such as Ethernet and fiber optics. At the software level, its firmware contains several key functional units: a data acquisition and processing unit responsible for real-time calculation of electrical quantities such as active power, reactive power, and harmonics; a dual-mode control logic unit embedding algorithms such as autonomous droop control and capable of smoothly switching between different control modes; and a protection and fault recording unit ensuring rapid and safe response of the equipment under fault conditions such as overcurrent and overvoltage.

[0079] In terms of external interfaces, the solid-state transformer connects to the medium-voltage power grid, DC bus and low-voltage AC / DC microgrid through a variety of electrical ports; it also connects to the industrial communication network through communication ports to receive commands, report status and interact with upper-layer edge aggregators, thereby supporting cluster collaborative control.

[0080] High-speed and reliable communication network To ensure the effectiveness of the collaborative control system, a high-speed and reliable communication network is needed as a "neural network." This network adopts a hierarchical and partitioned architecture: the backbone layer (cloud-edge) relies on the power SDH fiber optic ring network or SPN network, using IP / MPLS technology to provide high-bandwidth, low-latency, and highly reliable primary and backup channels; the access layer (edge-end) establishes an industrial Ethernet ring network within the substation or distribution area, and for locations with difficult cabling, 5G wireless private networks or power line carriers can be used as supplements, and it must support precise clock synchronization protocols such as IEEE 1588PTP.

[0081] Communication networks must possess four key characteristics: First, redundancy, ensuring physical or logical redundancy along critical paths so that single-point failures do not affect core services; second, quality of service assurance, requiring network equipment to support differential services, allocate high-priority queues for control command flows, and reserve bandwidth; third, security, meeting the security protection regulations of power monitoring systems, employing measures such as vertical encryption and access control to ensure data security; and finally, manageability, supporting centralized network management for real-time monitoring of traffic, latency, and topology status, and enabling flexible traffic scheduling using software-defined networking technology.

[0082] After the system powers on, each solid-state transformer's local controller initiates autonomous control mode to maintain basic operation. Edge aggregators establish connections with the solid-state transformers, begin collecting data, and upload it to the cloud control center. The cloud center aggregates global information, generates optimized tasks, and dynamically schedules resources using the HHSO algorithm to optimize computational tasks and communication paths. Subsequently, control commands are distributed layer by layer along the optimized paths. The solid-state transformer's local controller smoothly switches to optimized control mode based on the commands, executing the global optimization strategy. Simultaneously, each layer continuously monitors feedback information, forming a closed loop of "perception-decision-scheduling-execution-evaluation," ensuring the entire solid-state transformer cluster remains in a collaborative, optimized, and reliable operating state.

[0083] Example 3 The present invention also provides an electronic device, including: a processor, a transmitting device, an input device, an output device, and a memory. The processor may be implemented using a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit, or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this application. The memory may be implemented using a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM), and is used to store computer program code. The computer program code includes computer instructions. When the processor executes the computer instructions, the electronic device executes a method as described in any of the above possible implementation methods.

[0084] Example 4 The present invention also provides a computer-readable storage medium storing a computer program, the computer program including program instructions, which, when executed by a processor of an electronic device, cause the processor to perform a method as described in any of the above possible implementations.

[0085] In the description of this specification, references to terms such as "an embodiment," "example," and "specific example" indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0086] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the present invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features claimed herein.

Claims

1. A network-based collaborative control and communication scheduling method for solid-state transformer clusters, characterized in that... The method, applied to networked systems including cloud control centers, multiple edge aggregators, and clusters composed of multiple solid-state transformers, includes the following steps: S101. Status Panoramic Perception and Task Generation: Based on the edge-cloud three-level architecture, the system collects the operating status data of the solid-state transformer cluster, performs global situation fusion in the cloud control center to generate a global situation panoramic view, and dynamically generates planned optimization tasks and event-driven control tasks based on the global situation panoramic view; S102. Coordinated Scheduling of Communication and Computing Resources: Establish a scheduling model with the optimization objectives of minimizing task completion delay, balancing resource load, and reducing system energy consumption. A hybrid heuristic algorithm is used to jointly solve the computation offloading decision, communication routing decision, and resource allocation decision for the planned optimization task and the event-driven control task, generating a dynamic scheduling strategy table. S103. Layered collaborative control command issuance and execution: According to the dynamic scheduling strategy table, the control command is issued to the target solid-state transformer after security verification and coordination by the edge aggregator; S104. Execution of local hybrid control strategy for solid-state transformer: The local controller of the solid-state transformer runs a dual-mode adaptive control strategy. The dual modes include an autonomous non-communication control mode based on the interconnected droop control network and a communication-assisted optimization control mode that introduces upper-level optimization instructions on the basis of the autonomous non-communication control mode. Smooth switching between modes is achieved through a mode manager. S105. Dynamic feedback and scheduling strategy adjustment: Monitor the execution effect of control commands and changes in system state. When performance indicators deviate from expectations or the system state changes abruptly, trigger the communication and computing resource collaborative scheduling steps for re-optimization to form closed-loop control.

2. The method according to claim 1, characterized in that... In step S101, the three-tier architecture of end-edge-cloud specifically includes: On the end side, each solid-state transformer generates a snapshot of its device status, including its core electrical state, device health status, and the status of the microgrid it governs; On the side, the edge aggregator performs timestamp alignment and data cleaning on the device status snapshots of each solid-state transformer within its sub-cluster, and extracts regional features; On the cloud side, the cloud control center integrates data uploaded by all edge aggregators to form the aforementioned global situational overview.

3. The method according to claim 1, characterized in that... In step S102, the hybrid heuristic algorithm integrates the global exploration mechanism of the artificial bee colony algorithm and the local development mechanism of the particle swarm optimization algorithm; the solution encoding of the hybrid heuristic algorithm is a multi-dimensional vector containing task allocation identifier, communication path identifier and resource allocation vector; the hybrid heuristic algorithm adopts an adaptive weight adjustment mechanism, increasing the weight of random exploration in the early stage of iteration and increasing the weight of learning from the historical optimal solution and the global optimal solution in the later stage of iteration.

4. The method according to claim 1, characterized in that... In step S104, the autonomous non-communication control mode establishes a correlation between the droop coefficients of the low-voltage DC bus voltage, the low-voltage AC frequency, the medium-voltage DC bus voltage, and the medium-voltage AC frequency, and constructs a voltage-frequency-power droop network, enabling the solid-state transformer cluster to achieve self-organized power balance and voltage support under non-communication conditions.

5. The method according to claim 1, characterized in that... In step S104, the communication-assisted optimization control mode adopts a reference value fusion algorithm to perform weighted calculation of the optimization instructions issued by the upper layer and the local droop control output. The fusion weight is smoothly transitioned according to the communication status and the validity of the instructions to achieve a disturbance-free switch.

6. The method according to claim 1, characterized in that... The method further includes a communication service quality mapping step, which assigns a service quality level to each task based on the task type and urgency; and in the communication and computing resource collaborative scheduling step, high-priority communication paths and computing resources are allocated to tasks with high service quality levels.

7. A network-coordinated control and communication scheduling system for solid-state transformer clusters, employing the method described in any one of claims 1 to 6, characterized in that... ,include: The cloud control center includes a global state database, a global optimization decision module, a resource collaborative scheduling module, and a task generation and instruction management module, which are used to execute the following parts of the method: step S101 state panoramic perception and task generation, step S102 communication and computing resource collaborative scheduling, and step S103 dynamic feedback and scheduling strategy adjustment. Multiple edge aggregators are communicatively connected to the cloud control center. Each edge aggregator includes a regional data aggregation module and a regional coordination module, used to perform data aggregation, instruction verification, and forwarding. A solid-state transformer cluster, consisting of multiple solid-state transformers, each solid-state transformer having a built-in local controller. The local controller includes a dual-mode control logic unit and a mode manager, used to execute step S104 of the method: solid-state transformer local hybrid control strategy execution. A high-speed and reliable communication network connects the cloud control center, the edge aggregator, and the solid-state transformer cluster, providing a communication channel for data and command transmission.

8. The system according to claim 7, characterized in that... The edge aggregator also has edge autonomy functionality. When a communication interruption with the cloud control center is detected, it can switch to an edge autonomy mode based on local data for collaborative optimization. The local controller adopts a hardware architecture that combines DSP and FPGA. The FPGA is used to generate PWM pulse modulation signals, and the DSP is used to run control algorithms. The high-speed and reliable communication network supports at least one of fiber optic ring networks and 5G wireless private networks, and has redundant communication paths.

9. An electronic device, comprising a processor, a memory, and a computer program stored in the memory, characterized in that... When the processor executes the computer program, it implements the method as described in any one of claims 1-6.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that... When the program is executed by the processor, it implements the method as described in any one of claims 1-6.

Images