Multi-loop power supply abnormity cooperative regulation and control system for oil storage station

Abstract

The invention discloses a multi-loop power supply abnormity cooperative regulation and control system for an oil storage station, and belongs to the technical field of industrial power supply and distribution safety and intellectualization. The system adopts a'cloud-side-end 'distributed intelligent architecture, and comprises a field sensing and execution layer, a regional cooperative control layer and a global optimization decision-making layer. Electrical and non-electrical quantity information is collected in real time through a multi-source heterogeneous data fusion sensing network; based on a dynamic topology and multi-agent cooperation mechanism, rapid abnormity diagnosis and flexible isolation are realized; fault clearing and seamless power supply switching are realized by using a software-defined flexible switch; based on a digital twinning and global optimization engine, a collaborative recovery strategy is generated and verified. The system has self-learning and ductile evolution capabilities, and can be continuously optimized from historical events. According to the invention, the crossing from the static hysteresis response to the dynamic cooperative active regulation is realized, and the safety, reliability and recovery capability of the power supply system of the oil storage station are remarkably improved.

Description

Technical Field

[0001] This invention belongs to the field of industrial power supply and distribution safety and intelligent technology, specifically relating to a multi-circuit power supply anomaly collaborative control system for oil storage stations. Background Technology

[0002] With the development of integrated energy stations, the power systems within these stations are becoming increasingly complex, placing higher demands on power supply security. Existing technologies, such as the control method disclosed in Chinese patent CN118899853A, provide a solution for ensuring power security in energy stations by establishing load models for refueling and gas filling equipment and verifying and adjusting power parameters based on power flow calculations.

[0003] However, this method has inherent limitations and is difficult to meet the requirements of industrial scenarios such as oil storage facilities that have extremely high requirements for real-time performance and security.

[0004] The contradiction between static analysis and dynamic response: This method relies on steady-state power flow calculation, which is essentially a post-event verification and static optimization. However, major risks in the power supply system (such as short circuits and voltage sags) occur in dynamic processes on the order of milliseconds to seconds. Static models cannot accurately describe and respond to these events in a timely manner, resulting in severe lag in regulation.

[0005] The disconnect between model adjustment and actual control: the control object is the parameters in the digital model, not the physical equipment. The solution does not address how the optimization results of the model parameters can be translated into real-time, precise control commands for field circuit breakers, compensation devices, and other equipment, thus lacking the ability to form a closed-loop safety control system.

[0006] The system suffers from a limited safety dimension: its constraints only address electrical quantities such as voltage and current, failing to integrate key non-electrical quantities such as equipment temperature, partial discharge, and ambient gas concentration. This prevents the system from achieving early, comprehensive risk assessment based on multi-source information.

[0007] The lack of local optimization and global coordination is a significant drawback: this method focuses on constraint verification of individual nodes. In real-world multi-loop power grids, faults propagate, and power restoration requires consideration of the globally optimal path. This scheme lacks the ability to identify abnormal coordinated propagation and a global coordinated restoration strategy that ensures core loads and minimizes the scope of power outages.

[0008] Therefore, existing technologies struggle to bridge the gap between static, lagging, and isolated model verification and dynamic, real-time, and collaborative proactive safety control. A novel, systematic solution is urgently needed to address the specific requirements of oil storage facilities. Summary of the Invention

[0009] To address the systemic deficiencies in the aforementioned background technologies, this invention proposes a multi-circuit power supply anomaly collaborative control system for oil storage stations. The core of this invention lies in constructing a closed-loop control system characterized by "real-time perception across the entire domain, dynamic collaborative cognition, global optimization decision-making, and flexible and precise execution," thereby achieving a fundamental shift in response to power supply anomalies from passive response to proactive immunity and collaborative recovery, thus resolving the technical problems in the aforementioned background technologies.

[0010] To achieve the above objectives, the present invention adopts the following technical solution: a multi-circuit power supply anomaly collaborative control system for oil storage stations. The system adopts a distributed intelligent architecture with "cloud-edge-device" collaboration, specifically divided into the following three layers:

[0011] 1. Field Sensing and Execution Layer (End Layer): This layer consists of various intelligent sensing devices deployed on the primary equipment side (such as synchronous phasor measurement units, temperature / partial discharge composite sensors, and gas detectors) and software-defined flexible switches as the core actuators. It is responsible for multi-dimensional data acquisition and control command execution.

[0012] 2. Regional Collaboration and Control Layer (Edge Layer): This layer consists of edge intelligent control cabinets deployed in various power distribution rooms or areas. Each control cabinet integrates a high-performance industrial computing unit, responsible for real-time fusion of multi-source data within its area, rapid anomaly diagnosis, and the judgment and execution of protection collaboration logic with adjacent areas.

[0013] 3. Global Optimization and Decision-Making Layer (Cloud Layer): This refers to the digital twin and collaborative control master station deployed in the station control center or cloud platform. It maintains a high-fidelity dynamic model synchronized with the physical system, and is responsible for the station-wide operational status assessment, resilience analysis, global optimization calculation of cross-regional collaborative recovery strategies, as well as the distribution and learning optimization of control strategies.

[0014] The composition includes:

[0015] 1. Synchronous Fusion Sensing Network for Multi-Source Heterogeneous Data

[0016] This addresses the issue of limited information dimensions in background technologies, providing the system with comprehensive "situational information" necessary for judgment.

[0017] Composition: In addition to traditional electrical quantity transformers, wired or wireless temperature sensors, ultra-high frequency partial discharge sensors, and acoustic vibration sensors are added at key locations such as cable joints, switch cabinets, and transformers. Combustible gas concentration detectors are deployed near the process area. All sensors are connected to the network via an industrial IoT gateway based on precision clock synchronization (such as IEEE 1588 PTP).

[0018] The system can not only "see" voltage and current, but also "sense" the thermal state of equipment, insulation state, and the concentration of combustibles in the environment, realizing synchronous panoramic monitoring of the electrical, physical, and chemical safety status of the power supply system, and providing a multi-dimensional data foundation for early warning.

[0019] 2. Rapid anomaly recognition and isolation based on dynamic topology analysis and multi-agent collaboration

[0020] This addresses the issues of "node independence" constraints and the inability to handle dynamic faults in the background technology.

[0021] Composition: Dynamic Topology Real-Time Identification Module (located at the edge layer): Based on real-time acquired switch position signals and electrical quantities, it automatically generates and updates the real-time electrical connection topology diagram of the entire station. This topology is dynamic and changes with the operating mode.

[0022] Distributed area protection coordination unit (located at the edge layer): This eliminates the need for single threshold protection. When any intelligent sensor detects an anomaly (such as a sudden current surge or temperature rise), it immediately broadcasts the event and characteristic data via a low-latency industrial network to adjacent area control units associated with its electrical topology. Each area unit, based on shared synchronous phasor data and dynamic topology, runs distributed algorithms (such as longitudinal comparison and direction determination) to collaboratively determine the exact location and nature of the anomaly (e.g., "the fault is located on the line segment between cabinet A and cabinet B").

[0023] Collaborative flexible isolation: After a fault is detected, the edge control unit of the relevant area no longer only commands the switch on its own side to trip, but negotiates to determine an optimal sequence of isolation points, and may command the nearest software-defined flexible switch to perform precise disconnection at the current zero crossing point, which greatly reduces operating overvoltage and arc, and achieves "flexible fault clearing".

[0024] By upgrading anomaly identification and isolation from local judgments that rely on fixed values ​​to wide-area collaborative intelligent judgments based on real-time topology and shared information, faster and more selective fault isolation is achieved, effectively suppressing fault propagation.

[0025] 3. Software-defined flexible switches as the core actuator

[0026] This addresses the issues of disconnect between the control object (model parameters) and physical execution in the background technology, as well as the large impact and inability to achieve uninterrupted switching in traditional mechanical switch operations.

[0027] Composition: This switch is essentially a combination of a solid-state circuit breaker made of fully controlled power electronic devices (such as SiC MOSFETs or IGBTs) and a voltage source converter. It is connected in series in the circuit that requires flexible control.

[0028] Normal state: in low-loss conduction mode, equivalent to a wire.

[0029] In case of a fault: As a solid-state circuit breaker, it can quickly shut off and interrupt the fault current.

[0030] During power switching: As an inverter, its control core receives a "pre-synchronization grid connection" command from the clouds. It can actively adjust the amplitude, phase, and frequency of its output voltage to achieve precise synchronization with the target power supply side, and then smoothly transfer the load current from the original power supply to the target power supply at a controllable ramp rate, with no voltage drop and no current surge throughout the entire process.

[0031] It enables software-based, flexible, and precise control of the power grid connection status, providing a physical basis for "zero flicker" power supply switching and flexible fault isolation, and directly and losslessly applying the results of optimization decisions to the physical power grid.

[0032] 4. Collaborative recovery strategy engine based on digital twin and global optimization

[0033] This addresses the lack of a global collaborative recovery strategy in the background technology, enabling a leap from local parameter adjustment to global power supply reconfiguration.

[0034] Composition (located in the cloud layer):

[0035] Real-time synchronized digital twin: Establish a hybrid electromagnetic-electromechanical transient simulation model for the entire station, including detailed equipment models, line parameters, and load dynamic characteristics. This model maintains synchronization with the physical system by receiving real-time status (switch positions, voltage phasors) from the edge layer.

[0036] Collaborative Recovery Strategy Optimizer: Once a fault is isolated, the optimizer starts with the current digital twin state as the initial condition. Its optimization objective is not single-load power supply safety, but rather a comprehensive optimization of multiple objectives: ① Minimize the weighted outage time of the core load; ② Minimize the switching operation sequence and electrical impact (prioritizing the use of software-defined flexible switches); ③ Maximize the static and transient stability margins of the restored system. The optimizer utilizes parallel computing and heuristic algorithms to quickly search for the optimal power restoration path and corresponding equipment operation sequence ("operation ticket") that satisfies the above objectives.

[0037] Strategy verification and deployment: The generated strategy is first simulated and verified in the digital twin. After confirming that there is no risk, it is then deployed to the corresponding edge control unit and software-defined flexible switch for execution.

[0038] It achieves global coordination, decision optimization, and pre-verification of operations in the power restoration process after a power outage, significantly shortening the restoration time, maximizing the power supply to core loads, and ensuring the safety of the restoration operation itself.

[0039] 5. System's self-learning and resilient evolution mechanism

[0040] This enables the system to continuously improve based on operational experience, thereby enhancing its ability to handle complex anomalies.

[0041] The system stores all abnormal events, control strategies implemented, and their final effects as "cases" in a knowledge base. These cases are analyzed periodically using machine learning methods: for example, optimizing feature thresholds for early warning algorithms; evaluating the actual effectiveness of different recovery strategies and adjusting weight coefficients in the optimization model; and even learning new anomaly patterns. Updated algorithms and parameters can be securely injected online into the corresponding modules of the edge and cloud layers.

[0042] The system has evolved from a statically designed control system into an intelligent agent capable of autonomously learning from historical data and continuously optimizing its early warning accuracy and decision-making effectiveness, thus achieving a dynamic improvement in system resilience.

[0043] Compared with the prior art, the beneficial effects of the present invention are:

[0044] 1. Through the "cloud-edge-device" collaborative architecture and real-time digital twin model, the system can quickly respond to power supply anomalies, realize rapid fault location, isolation and recovery, and overcome the lag problem of traditional methods in dynamic response.

[0045] 2. By integrating multi-source information of electrical and non-electrical quantities, it enables panoramic perception of equipment thermal status, insulation status, and environmental safety, supports early warning and preventive control, and significantly reduces the probability of failure.

[0046] 3. Based on dynamic topology and multi-agent collaboration mechanism, it supports cross-regional fault collaborative isolation and global optimal recovery path planning, shortens the power outage time of core loads, and improves the system power supply resilience.

[0047] 4. By using software-defined flexible switches, the zero-crossing interruption of fault current and the pre-synchronous seamless switching between power supplies are achieved, reducing electrical shocks and operational overvoltages, and improving equipment lifespan and power supply quality.

[0048] 5. The system continuously optimizes the early warning model, strategy weights, and anomaly recognition capabilities through machine learning, achieving intelligent regulation that learns from experience and evolves during operation, thereby gradually improving the overall resilience of the system. Attached Figure Description

[0049] Figure 1 This is a diagram of the overall system architecture of the present invention;

[0050] Figure 2 This is a schematic diagram of the network topology and device deployment of the present invention;

[0051] Figure 3 This is a structural diagram of the Software Defined Flexible Switch (SDFSS) of this invention;

[0052] Figure 4 This is a compact layout diagram of the abnormal collaborative processing and recovery process of the present invention;

[0053] Figure 5 This is a diagram illustrating the self-learning and resilient evolution mechanism of this invention. Detailed Implementation

[0054] 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 some embodiments of the present invention, and not all embodiments. 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.

[0055] like Figures 1 to 5 As shown, the present invention provides a multi-circuit power supply anomaly coordinated control system for oil storage stations, comprising:

[0056] I. System Hardware Deployment and Network Connection

[0057] 1. On-site equipment deployment:

[0058] Electrical quantity acquisition points: Synchronous phasor measurement units or high-precision integrated protection and control devices with synchronous sampling functions are installed in the incoming lines, outgoing lines, bus couplers, and sectionalizing switchgear of 10kV and above voltage levels. These devices are directly connected to the secondary circuits of current transformers (CTs) and voltage transformers (PTs).

[0059] Non-electrical quantity acquisition points: Wireless temperature sensors are installed at key heat-generating locations such as high-voltage cable joints, switchgear busbar connections, and dry-type transformer windings; ultra-high frequency partial discharge sensors are deployed inside high-voltage cabinets; and combustible gas concentration detectors are installed in cable trenches in power distribution rooms and near process equipment areas. These sensors are connected via built-in or external IoT gateway modules.

[0060] Core execution mechanism deployment: Install software-defined flexible switches in the following key locations:

[0061] (a) Connection point for two incoming power supplies (in lieu of or in parallel with a conventional bus tie circuit breaker).

[0062] (b) In a dual power supply switch cabinet (replacing the automatic transfer switch) for supplying power to Class I critical loads (such as control center UPS, critical pumps).

[0063] This switch employs a fusion structure of a solid-state circuit breaker and a voltage source converter, based on fully controlled power electronic devices (such as SiC MOSFETs or IGBTs). Its primary terminals are connected in series with the main circuit, and the control unit is connected to the control network via optical fiber. Under normal conditions, it operates in a low-loss conduction mode, equivalent to a conductor; in case of a fault, it acts as a solid-state circuit breaker for rapid turn-off; during power switching, it functions as a converter, actively adjusting the amplitude, phase, and frequency of the output voltage to achieve pre-synchronization with the target power supply and impact-free grid connection.

[0064] 2. Edge layer device deployment:

[0065] Based on the division of power distribution areas, edge intelligent control cabinets are installed in the power distribution rooms of each area. The core of the cabinet is an industrial-grade edge computing server, equipped with multi-port communication modules and synchronous clock receiving modules (such as Beidou / GPS+IEEE 1588PTP).

[0066] Connectivity: Data from all field-level intelligent sensors (electrical and non-electrical) within this area is aggregated to the edge intelligent control cabinet of this area via industrial Ethernet or industrial wireless network. Simultaneously, the output commands from this control cabinet directly control conventional circuit breakers and software-defined flexible switches within this area via hardwiring or a high-speed network.

[0067] 3. Deployment of the station control layer / cloud layer:

[0068] Hardware: Deploy a high-performance server cluster in the station control center, or rent secure and reliable industrial cloud resources.

[0069] Network architecture:

[0070] Process Bus: Edge intelligent control cabinets in each area are connected via a high-speed redundant fiber optic ring network, as well as between them and cross-area software-defined flexible switch controllers. This network carries manufacturing message specifications such as GOOSE for transmitting protection coordination and control commands, ensuring deterministic and extremely low transmission latency.

[0071] Station control bus: All edge intelligent control cabinets, station control center servers, and human-machine interaction workstations are interconnected through station control layer Ethernet switches, carrying monitoring information model (such as IEC 61850 MMS) data, used to transmit non-real-time monitoring data, model parameters, and optimization strategies.

[0072] Internal and external network security isolation: The station control layer network is connected to the enterprise management information network or cloud platform through a forward physical isolation device to ensure control security.

[0073] II. Software System Architecture and Data Flow

[0074] 1. Edge intelligent control cabinet software:

[0075] Data acquisition and fusion module: Receives and time-aligns electrical and non-electrical quantity data of this area to form a "data frame" with a unified time stamp.

[0076] Local rapid diagnosis and agent module: Runs lightweight fault detection algorithms and rule / model-based equipment status assessments (such as temperature-based dynamic load capacity assessment). Diagnostic results (normal, warning, fault) and key characteristic data are uploaded to the station control layer in real time.

[0077] Dynamic topology identification module: Based on real-time collected switch position signals and electrical quantities, it automatically generates and updates the real-time electrical connection topology map of the area, and exchanges topology information with adjacent areas to form a global dynamic topology view, providing real-time network structure information for collaborative protection.

[0078] Regional Coordination Protocol Stack: Maintains the "neighbor relationship table" of this region in the network. When an abnormal event is detected or received from a neighbor, the cooperative processing flow is initiated: exchanging synchronization phasor data, performing fault location and nature determination based on dynamic topology and consensus algorithm, and generating regional cooperative control commands (such as joint trip command and flexible switch start command).

[0079] 2. Digital Twin and Collaborative Control Main Station Software (Cloud Layer):

[0080] Model Management Subsystem: Establishes and maintains a digital twin model of the entire station. The model not only includes the power grid topology and equipment parameters, but also integrates thermal and aging models of key equipment. This subsystem drives the simulation model to maintain "loose synchronization" with the physical system by subscribing to key state changes and telemetry data on the station control bus.

[0081] Global Analysis and Optimization Decision Subsystem:

[0082] Under normal circumstances: Periodically perform "N-1" safety analysis, voltage stability analysis, and dynamic setting calculation based on real-time heat capacity, and send the optimized protection setting or operating limit to the edge layer.

[0083] After a fault occurs: The system receives fault isolation results and the current power grid topology reported by the edge layer. Using a digital twin as a sandbox, the decision engine aims to "restore power supply as quickly as possible," invoking path search and scheduling algorithms to generate a detailed recovery plan that includes conventional switching operations and software-defined flexible switching control sequences. After the plan is generated, a full-process simulation is performed in the digital twin to verify its safety and effectiveness.

[0084] Strategy Execution Management and Learning Subsystem:

[0085] It is responsible for breaking down the verified recovery plan into specific, time-sequenced instruction sequences and sending them to the relevant edge control cabinets and flexible switch controllers through a secure channel.

[0086] At the same time, the actual effect feedback after the strategy is implemented is collected and compared with the simulation prediction, and the complete case (abnormal event, control strategy, execution effect) is stored in the knowledge base.

[0087] The knowledge base is analyzed regularly (e.g., weekly or monthly) using machine learning methods to optimize the feature thresholds of the early warning algorithm, adjust the weight coefficients in the recovery strategy model, identify new abnormal patterns, and inject the updated algorithm and parameters online into the corresponding modules of the edge layer and cloud layer to achieve the system's self-learning and resilient evolution.

[0088] III. System Collaboration Workflow Description

[0089] 1. Steady-state monitoring and early warning process:

[0090] Non-electrical quantity sensors (such as temperature sensors) continuously monitor the equipment status. When the temperature of a cable joint continues to rise slowly and approaches the warning threshold, the status assessment module of the edge control cabinet will calculate a dynamic "allowable temperature limit" by combining the real-time load current of that circuit. If the temperature is predicted to exceed the limit, the system will not only issue an alarm, but may also generate a "preventive control suggestion" by controlling the master station. For example, by adjusting the operating mode, part of the load of that circuit can be smoothly transferred to a light-load circuit through the lossless transfer function of a software-defined flexible switch, thereby realizing status-based preventive load scheduling and avoiding overheating faults.

[0091] 2. Fault Occurrence and Rapid Collaborative Response Process:

[0092] Sensing and Information Convergence: When a short circuit occurs on the line, the synchronous phasor measurement units installed at both ends of the line quickly capture the sudden change in current and voltage, and send the sampled value or change characteristics containing high-precision time scale to the corresponding edge control cabinet.

[0093] Regional collaborative diagnosis and decision-making: Two related edge control cabinets exchange information at high speed through the process bus and determine the fault range based on real-time dynamic topology. Utilizing the longitudinal protection principle, a collaborative determination of "intra-zone fault" is made within 5-10ms, and the fault phase is confirmed.

[0094] Collaborative flexible isolation execution: After negotiation, the two edge control cabinets issue trip commands to the switches at both ends of the faulty line. If one or both ends are software-defined flexible switches, they are commanded to disconnect at the next natural zero-crossing point of the detected current, achieving arc-free or low-arc isolation. The fault is confined to a minimum area.

[0095] 3. Global collaborative process for power restoration:

[0096] Status reporting and model synchronization: After fault isolation, the edge control cabinet reports the new switch status and power failure area information to the control master station. The master station's digital twin model immediately updates the topology to the post-fault state.

[0097] Global recovery strategy generation and simulation: The optimization engine of the control master station is activated. It first identifies all loads and their priorities within the power outage area. Then, in the digital twin, it searches for all possible power restoration paths, constrained by the current power grid topology. The optimization calculation considers not only whether power can be restored, but also evaluates the closing impact, voltage stability, and equipment load-bearing capacity under different paths. Ultimately, it generates one or more optimal recovery sequences. For example, the optimal sequence might be: "Step 1: Activate the software-defined flexible switch at location A, synchronizing its output voltage to bus B; Step 2: After synchronization, close the flexible switch to supply power to the primary load X on bus B in a flexible manner; Step 3: After confirming stable operation, close the conventional switch C to restore the remaining loads."

[0098] Strategy Verification, Issuance, and Execution: The strategy undergoes full-dynamic process simulation in a digital twin. After verification that there are no risks, it is converted into a specific list of instructions with strict timing, and issued to the corresponding edge control cabinets and flexible switch controllers. Each execution unit executes the operation precisely according to the timestamps required by the list, and the system enters the recovery process. The master station monitors key telemetry data throughout the execution process and compares it with the simulation expectations to ensure that the recovery process is under control.

[0099] As can be seen from the above description of specific embodiments, the system of the present invention is an organic whole with clear layers, well-defined division of labor, and close collaboration. From bottom-level perception and execution to mid-level rapid collaboration and top-level global optimization, each layer enhances and innovates upon a specific deficiency of existing technologies. These enhancements are seamlessly integrated through standardized interfaces and protocols, together forming an advanced, practical, and evolvable solution capable of addressing the complex power supply security challenges of oil storage facilities.

[0100] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A multi-circuit power supply anomaly coordinated control system for oil storage stations, characterized in that: The architecture adopts a three-layer distributed collaborative architecture of "cloud-edge-device", including: The field sensing and execution layer is deployed on the primary equipment side, including intelligent sensing devices installed in switch cabinets, cable joints, and transformers, and software-defined flexible switches connected in series in critical circuits; The regional collaboration and control layer is deployed in the edge intelligent control cabinets of each power distribution room. It is connected to the field layer devices via industrial Ethernet and is responsible for data fusion and collaborative control in its region. The global optimization and decision-making layer is deployed on the station control center server or cloud platform and is connected to all edge intelligent control cabinets through the station control bus. It is responsible for the simulation of the entire station status and strategy optimization. The three layers are interconnected through a process bus and a station control bus to achieve hierarchical transmission and collaborative interaction of data and commands.

2. The multi-circuit power supply anomaly coordinated control system for oil storage stations according to claim 1, characterized in that, The equipment deployment of the on-site perception and execution layer includes: Synchronous phasor measurement units are installed in 10kV and above incoming lines, outgoing lines, bus couplers and sectional switchgear, and directly connected to the secondary circuits of current transformers and voltage transformers. Wireless temperature sensors and ultra-high frequency partial discharge sensors are installed at high-voltage cable joints, switchgear busbar connections, and transformer windings. Install combustible gas concentration detectors in the cable trenches of the power distribution room and near the process equipment area; Install software-defined flexible switches in series at the following locations: (a) The connecting bus section for the two incoming power supplies; (b) In dual power supply switching circuits for Class I particularly important loads, it replaces or is connected in parallel with conventional automatic transfer switches.

3. The multi-circuit power supply anomaly coordinated control system for oil storage stations according to claim 2, characterized in that, The control unit of the software-defined flexible switch is connected to the control network via optical fiber. Its primary terminal is connected in series with the main circuit. Under normal conditions, it is in a low-loss conduction mode. In case of a fault, it acts as a solid-state circuit breaker to perform disconnection at the current zero-crossing point. During power switching, it acts as a converter to perform pre-synchronous grid connection control.

4. The multi-circuit power supply anomaly coordinated control system for oil storage stations according to claim 1, characterized in that, The edge intelligent control cabinets in the regional collaboration and control layer are deployed according to power distribution areas, and each control cabinet integrates: Industrial-grade edge computing servers are used to run local diagnostic and collaborative control algorithms; Multi-port communication module for connecting to all smart sensing devices in this area; Synchronous clock receiving module, used for data time stamp synchronization; The output ports of the control cabinet are connected to circuit breakers and software-defined flexible switches in the area via hardwiring or high-speed networks.

5. The multi-circuit power supply anomaly coordinated control system for oil storage stations according to claim 4, characterized in that, The edge intelligent control cabinets are interconnected through a high-speed redundant fiber optic ring network to form a process bus network, which is used to transmit manufacturing messages such as GOOSE, and realize rapid protection coordination and control command exchange.

6. The multi-circuit power supply anomaly coordinated control system for oil storage stations according to claim 1, characterized in that, The global optimization and decision-making layer includes: The digital twin, built on a high-performance server cluster, is a full-station electromagnetic-electromechanical hybrid simulation model that keeps synchronized with the physical system by subscribing to real-time status data on the station control bus. The collaborative recovery strategy optimizer starts after fault isolation, uses a digital twin as the simulation environment, and combines load priority, switch status and equipment capacity constraints to generate the optimal power supply recovery path. The strategy verification module verifies the security and effectiveness of the recovery plan in the simulation, and executes it after confirming that there are no errors.

7. The multi-circuit power supply anomaly coordinated control system for oil storage stations according to claim 6, characterized in that, The output of the collaborative recovery strategy optimizer is a sequence of operation instructions with timing, including: Synchronous start-up and grid connection commands for software-defined flexible switches; The sequence of opening and closing commands for a conventional circuit breaker; Delay and verification logic between each step.

8. The multi-circuit power supply anomaly coordinated control system for oil storage stations according to claim 1, characterized in that, The system also includes a self-learning and resilient evolution module, deployed in the global optimization layer, including: A case knowledge base is used to archive abnormal events, control strategies, and execution feedback. The machine learning analytics engine regularly extracts features and optimizes models from case data. The online parameter update interface is used to send the optimized algorithm and threshold to edge layer and field layer devices.

9. The multi-circuit power supply anomaly coordinated control system for oil storage stations according to claim 1, characterized in that, The network architecture of the system is divided into: The process bus, using a ring or star fiber optic topology, is used for real-time protection and control. The station control bus uses a dual-network redundant Ethernet for monitoring data and policy transmission; The process bus and the station control bus are securely isolated from each other through a firewall or physical isolation device.

10. The multi-circuit power supply anomaly coordinated control system for oil storage stations according to claim 1, characterized in that, The collaborative workflow of the system includes: During the steady-state operation phase, the edge intelligent control cabinet periodically collects and integrates multi-source data, updates the dynamic topology in real time, and reports it to the global optimization layer; During the abnormality phase, the relevant edge control cabinets quickly complete fault location and isolation decisions based on dynamic topology and collaborative algorithms, and trigger the action of flexible switches or circuit breakers through the process bus. During the recovery phase, the global optimization layer quickly generates a recovery strategy, which is then verified through simulation and deployed to the edge layer for execution. The system monitors the execution process in real time and compares it with the simulation results.

Images