Off-line takeover and fallback method of virtual power plant and computer device

Abstract

The application discloses an off-line takeover and switching-back method of a virtual power plant and computer equipment, and relates to the field of power plant control. The method comprises the following steps: acquiring edge gateway data, and calculating an autonomous signal index according to the edge gateway data; when the autonomous signal index meets at least one preset trigger condition and meets a hysteresis rule, switching the edge gateway to a network-disconnection autonomous mode, wherein the preset trigger condition comprises one or more of an instruction time limit timeout, a link time delay over-limit, an ACK closed loop anomaly and an ACK sequence anomaly; during the network-disconnection autonomous mode, generating a local control instruction based on a safety constraint; and the application has the advantage of stable takeover under weak network / lost connection conditions.

Description

Technical Field

[0001] This invention relates to the field of collaborative control technology for virtual power plants and integrated energy systems in industrial parks, and particularly to an offline take-off and cut-off method and computer equipment for a virtual power plant. Background Technology

[0002] Existing virtual power plants in industrial parks typically adopt an architecture of "unified cloud-based optimization scheduling + edge gateway forwarding and execution". That is, the cloud calculates and issues power / current / policy commands based on predictions and constraints, while the edge is responsible for protocol adaptation, data collection and command issuance. In addition, to deal with communication anomalies, some solutions will configure simple "network outage protection strategies" on the edge side (such as charging and discharging under fixed power / fixed SOC threshold, current limiting of charging piles, etc.) or perform retransmission and timeout processing in the cloud.

[0003] However, in actual operation, the above-mentioned existing technologies have the following drawbacks: (1) The triggering conditions are crude: most of them only use "whether the link is broken" or "heartbeat timeout" as the switching conditions, without comprehensively considering the timeliness of cloud instructions, link delay, execution confirmation (ACK) closed loop status, etc., which makes it easy to switch frequently or misjudge under weak network jitter.

[0004] (2) The autonomous strategy lacks equipment priority and safety constraint closed loop: the common grid disconnection backup strategy is too static and it is difficult to simultaneously take into account multiple constraints such as energy storage power boundary, charging pile power supply continuity, inverter / PCS protection status, etc., which poses the risk of overstepping the boundary or power supply interruption.

[0005] (3) Impact caused by “hard switch back to cloud” after recovery: If cloud scheduling is directly restored after cloud recovery, the inconsistent cloud-edge status (SOC, charging queue, protection status, current power, etc.) may cause command jumps, over-limit ramping, protection malfunctions or load impacts.

[0006] (4) Lack of a complete state machine: Existing solutions often only describe the "cloud-edge collaboration" framework, lacking a reproducible and auditable closed-loop process and state transition conditions of "takeover trigger - autonomous operation - consistency verification - smooth back-switch".

[0007] Therefore, there is still a lack of a feasible technical solution that can stably take over under weak network / disconnection conditions and safely and smoothly switch back after recovery. Summary of the Invention

[0008] The problem to be solved by this invention is to provide an offline take-over and cut-back method for a virtual power plant, which can achieve stable take-over under weak network / disconnection conditions, while ensuring smooth cut-back and safe operation during the restoration of cloud control.

[0009] To address the aforementioned technical problems, this invention provides an offline takeover and switchback method for a virtual power plant, comprising: acquiring edge gateway data and calculating an autonomous signal index based on the edge gateway data; when the autonomous signal index satisfies at least one preset trigger condition and meets a hysteresis rule, switching the edge gateway to a network-disconnected autonomous mode, wherein the preset trigger condition includes one or more of the following: instruction timeout, link delay exceeding limits, ACK closed-loop anomaly, and ACK sequence anomaly; during the network-disconnected autonomous mode, generating local control instructions based on security constraints.

[0010] As an improvement to the above scheme, the instruction timeout means that the timestamp or lifespan of the cloud instruction has expired; and / or the link delay exceeds the limit means that the mean and variance of the link round-trip delay exceed the threshold within the sliding window; and / or the ACK loop closure anomaly is based on the ACK completion rate being lower than the completion rate threshold or the ACK delay being higher than the delay threshold; and / or the ACK sequence anomaly is that the received cloud instruction sequence number is not continuous.

[0011] As an improvement to the above scheme, the safety constraints include energy storage power boundary trimming, continuous power supply guarantee for charging queues, gradual power limiting when station-level constraints are insufficient, and secondary safety strategies when risks escalate.

[0012] As an improvement to the above solution, it also includes: When the cloud recovery candidate conditions are met, the execution of cloud commands is paused and a consistency determination is performed. Based on the consistency determination result, it is determined whether the edge gateway should switch back. If the edge gateway is determined to switch back, control is returned to the cloud according to the smooth switchback strategy. If the edge gateway is determined not to switch back, the network outage autonomous mode is maintained and abnormal status information is reported.

[0013] As an improvement to the above scheme, the cloud recovery candidate conditions include: cloud communication recovery, valid command timeliness, normal link latency, normal ACK closed loop, and normal command sequence; and / or the consistency judgment verification conditions include: state of charge consistency verification, verifying whether the SOC value of the energy storage system stored in the cloud is consistent with the real-time SOC value collected by the edge gateway; charging queue consistency verification, verifying whether the charging pile session and status list recorded in the cloud is consistent with the actual charging queue status of the edge gateway; and protection status consistency verification, verifying whether the protection flag bit of the controlled device recorded in the cloud is consistent with the real-time protection status collected by the edge gateway.

[0014] As an improvement to the above scheme, the formula for calculating the ACK completion rate is as follows:

[0015] in, Indicates the ACK completion rate. This represents the number of valid execution confirmations actually received by the edge gateway from the controlled devices within the same time window. This represents the total number of control commands received by the edge gateway from the cloud within the specified evaluation time window; The formula for calculating the ACK delay is as follows:

[0016] in, Indicates the confirmation delay for a single instruction. This indicates the local time at which the edge gateway receives confirmation of the execution of the corresponding instruction from the controlled device. This indicates the local time when the edge gateway sends a control command to the controlled device.

[0017] As an improvement to the above scheme, the step of energy storage power boundary trimming includes: The energy storage power boundary trimming is calculated using the following formula:

[0018] in, This represents the total station-level power demand of the energy storage system calculated by the local control strategy of the edge gateway under autonomous mode without network access. This represents the lower boundary of the safe power based on the current state of charge of the energy storage and the real-time protection status of the equipment. This represents the upper boundary of the safe power based on the current state of charge of the energy storage and the real-time protection status of the equipment. Indicates the state of charge. Indicates the real-time protection status of the equipment. This indicates the power value of the command executed by the energy storage system.

[0019] As an improvement to the above scheme, the smooth back-cut strategy includes a power ramping constraint strategy, specifically:

[0020] in, Indicates the first The actual output power of the controlled device during the cut-back process Indicates the first The current actual power of the controlled device. Indicates the first The cloud-based target power of the controlled device. Indicates the first The maximum descent ramp rate of each controlled device. Indicates the first The minimum ramp rate for each controlled device.

[0021] As an improvement to the above solution, the method for ensuring continuous power supply of the charging queue includes: Priority calculations are performed on several charging facilities, and power supply is guaranteed based on the calculation results; The priority calculation steps include:

[0022] in, The sequence is represented as Priority rating of charging facilities The sequence is represented as The basic preset priority of charging facilities The sequence is represented as The importance of continuous power supply for charging facilities, The sequence is represented as The charging facilities correspond to the current state of charge of the equipment. The sequence is represented as The charging cutoff pressure of charging facilities The weighting coefficient represents the basic preset priority. Weighting coefficients representing the importance of continuous power supply This represents the weighting coefficient of the device's current state of charge. This represents the weighting coefficient for the cutoff charging pressure.

[0023] Accordingly, the present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the offline take-off and cut-back method for the virtual power plant described in any of the above claims.

[0024] Implementing this invention has the following beneficial effects: This invention acquires edge gateway data and calculates multi-dimensional autonomous signal indicators. It comprehensively considers command timeliness, link latency, ACK closed loop, and sequence consistency to determine the status. By introducing hysteresis rules, it can accurately identify control link failures caused by network fluctuations or cloud service anomalies. This avoids the misjudgment and frequent mode switching problems that are prone to occur in traditional single heartbeat detection in weak network environments. As a result, it achieves stable and reliable offline takeover triggering, laying the foundation for subsequent secure autonomy. During the autonomous mode without network access, this method mandates that all local control commands be processed through a closed-loop safety constraint based on real-time device status. Specifically, dynamic boundary trimming of energy storage power ensures that charging and discharging commands never exceed the safe range allowed by the current state of charge and protection status of the energy storage. Flexible scheduling of charging load and gradual power limiting strategies can optimize and protect high-priority loads based on multi-factor scoring when power is insufficient. The secondary safety strategy provides degradation protection control for risk scenarios such as equipment failure. This series of measures together constitute a multi-layered adaptive security protection system, enabling the edge gateway to not only maintain continuous operation after disconnecting from the cloud, but also ensure intrinsic safety and controllability during operation. This fundamentally overcomes the shortcomings of traditional static protection strategies that are difficult to balance dynamic constraints and business continuity.

[0025] Furthermore, after detecting that the candidate conditions for cloud recovery are met, the cloud command is not executed immediately. Instead, it first enters a pre-synchronization state and suspends cloud scheduling. At the same time, it performs cloud-edge consistency verification, including state of charge, charging queue, and device protection status. This design establishes a safety buffer between cloud recovery and final command execution. Only after the verification is completely successful and the cloud-edge state is confirmed to be consistent will control be returned according to the smooth back-off strategy. If the verification fails, it remains in autonomous mode and reports an anomaly. This process completely solves the risks of power command jumps, electrical shocks, and device protection malfunctions caused by direct hard switching due to state asynchrony after cloud recovery in existing technologies. The power ramping constraint used in the smooth back-off strategy limits the device power change rate within its physical ramping capability through mathematical formulas, ensuring a smooth and uninterrupted transition from edge autonomous power to cloud target power. This achieves safe, smooth, and fully automatic recovery from network outage autonomy to cloud following. Attached Figure Description

[0026] Figure 1 This is a flowchart of the first embodiment of the offline take-off and cut-back method for the virtual power plant of the present invention; Figure 2 This is a flowchart of the second embodiment of the offline takeover and cut-off method for the virtual power plant of the present invention. Detailed Implementation

[0027] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings. It is hereby declared that the directional terms such as up, down, left, right, front, back, inside, and outside used in this text are based solely on the accompanying drawings and are not intended to specifically limit the invention.

[0028] like Figure 1 As shown, Figure 1 The flowchart of the first embodiment of the offline take-off and cut-back method for the virtual power plant of the present invention is shown, which includes: S101. Obtain edge gateway data and calculate autonomous signal indicators based on the edge gateway data; Specifically, it continuously monitors the communication link and command interaction status between the cloud control center and the edge gateway, and collects edge gateway data; It should be noted that the edge gateway data includes, but is not limited to, received cloud instructions and their timestamps, serial numbers, round-trip time (RTT) for communication with the cloud, and the time and content of issuing instructions to controlled devices (such as energy storage converters PCS and charging piles) and receiving execution confirmations (ACKs) returned by them.

[0029] Furthermore, based on the edge gateway data, a set of autonomous signaling indicators is calculated to evaluate the reliability of the cloud control link. These autonomous signaling indicators form the basis for subsequent decisions and specifically include: (1) Command timeliness index: obtained by judging whether the timestamp or preset time to live (TTL) carried by the cloud command has expired.

[0030] (2) Link quality metrics: Calculate the mean and variance of the link round-trip time (RTT) within the sliding time window.

[0031] (3) ACK closed-loop metrics: including ACK completion rate and ACK delay, to determine whether the instruction was successfully executed. The formula for calculating the ACK completion rate is as follows:

[0032] in: Indicates the ACK completion rate; This indicates the number of valid execution confirmations actually received by the edge gateway from the controlled devices (such as energy storage and charging piles) within the same time window; This represents the total number of control commands received by the edge gateway from the cloud within the specified evaluation time window; The formula for calculating the ACK delay is as follows:

[0033] in: Indicates the confirmation delay for a single instruction; This indicates the local time when the edge gateway receives confirmation of the execution of the corresponding instruction from the controlled device. This indicates the local time when the edge gateway sends a control command to the controlled device.

[0034] (4) Command sequence index: Check whether the received cloud command sequence number is continuous and without jumps.

[0035] S102. When the autonomous signal indicator meets at least one preset trigger condition and satisfies the hysteresis rule, the edge gateway is switched to the network disconnection autonomous mode. Specifically, the present invention compares the calculated autonomous signal index with a preset threshold; the preset triggering condition includes at least one of the following: If a command expires, the system compares the timestamp or lifespan of the command in the cloud with a preset time interval to determine if the command has expired. Link latency exceeds the limit. The system determines whether a timeout has occurred based on whether the mean and variance of the link round-trip latency exceed the threshold within the sliding window. ACK loop failure is determined based on whether the instruction was successfully executed, with the ACK completion rate being lower than the completion rate threshold and / or the ACK delay being higher than the delay threshold. An abnormal ACK sequence is detected by determining whether the sequence is abnormal based on the discontinuous sequence numbers of the received cloud instructions.

[0036] It should be noted that a hysteresis rule is introduced to prevent frequent mode switching due to momentary network fluctuations. That is, the trigger threshold for switching from cloud control to network-disconnected autonomous mode is different from the recovery threshold for switching back to cloud control from network-disconnected autonomous mode, and the threshold for restoring to cloud control is usually more stringent.

[0037] For example, the ACK completion rate threshold for triggering the autonomous mode during network outages might be set at 90%, while the ACK completion rate threshold for determining cloud recovery and allowing consideration of a rollback needs to reach 95%. Mode switching will only be executed when the metric continues to deteriorate and exceeds the trigger threshold, and the hysteresis logic is met.

[0038] S103. During the network outage autonomous mode, generate local control commands based on security constraints.

[0039] In the autonomous mode of network outage, the edge gateway calculates the station-level power demand based on the local preset strategy (such as scheduling based on local renewable energy forecasts and load demand), but all instructions must be verified and processed through the local security constraint closed loop before being issued. Specifically, the safety constraints include energy storage power boundary trimming, continuous power supply guarantee for charging queues, gradual power limiting when station-level constraints are insufficient, and secondary safety strategies when risks escalate.

[0040] It should be noted that the executable power of an energy storage system is not a fixed value, but rather changes dynamically with its state of charge (SOC) and the real-time protection status of the equipment (such as over-temperature and over-current indicators). Therefore, the calculation of the executable power of the energy storage system must take into account the state of charge and the real-time protection status of the equipment, as detailed below:

[0041] in: This represents the total station-level power demand of the energy storage system calculated by the local control strategy of the edge gateway under autonomous state without network access. This represents the lower boundary of the safe power based on the current state of charge of the energy storage and the real-time protection status of the equipment. This represents the upper boundary of the safe power based on the current state of charge of the energy storage and the real-time protection status of the equipment. Indicates the state of charge; Indicates the real-time protection status of the equipment; This indicates the power value of the command executed by the energy storage system.

[0042] Furthermore, when the available power within the station is insufficient, orderly power limiting management of multiple charging piles is required. By prioritizing charging equipment, critical loads are ensured while optimizing user experience and maximizing business continuity. The specific priority scoring formula is as follows:

[0043] in: The sequence is represented as Priority rating of charging facilities; The sequence is represented as The basic preset priority of charging facilities; The sequence is represented as The importance of continuous power supply for charging facilities; The sequence is represented as The charging facility corresponds to the current state of charge of the equipment; The sequence is represented as The cutoff charging pressure of the charging facilities; The weighting coefficient represents the basic preset priority. Weighting coefficients indicating the importance of continuous power supply; Indicates the weighting coefficient of the device's current state of charge; This represents the weighting coefficient for the cutoff charging pressure.

[0044] More preferably, in the autonomous mode without network access, when the calculated total local power demand exceeds the current available total power limit of the site, a "site-level constraint insufficiency" state is triggered. At this time, the present invention does not brute-force proportional or random reduction of all loads, but implements a priority-based, flexible, and gradual power limiting strategy, as follows: The system will prioritize reducing the charging power of the session with the lowest score (for example, by lowering its power setting by one step). Each time it is adjusted, the system will recalculate whether the total power of the entire station has met the constraints. If it still does not meet the constraints, the system will continue to limit the power of the next session with a lower score, and so on, until the total power demand is reduced to within the safe limit.

[0045] This "gradual power throttling" process ensures that high-priority and high-urgency charging services can be guaranteed to the greatest extent when load must be reduced, achieving optimal allocation of limited power resources and demonstrating the intelligence and business friendliness of autonomous control.

[0046] Furthermore, when system operational risks escalate, such as when a controlled device (e.g., an energy storage converter) triggers a critical protection signal (the PRT flag indicates over-temperature, over-current, insulation fault, etc.), or when system stability indicators (e.g., bus voltage frequency deviation) further deteriorate, a more stringent secondary safety strategy will be activated. This strategy includes, but is not limited to: Dynamic boundary emergency tightening: Safety power boundary used in large tightening and For example, when the equipment overheats, the maximum permissible charge and discharge power can be reduced to 50% or even lower of the rated value to force the equipment to operate at a reduced rate and prevent the fault from escalating.

[0047] Control mode degradation: In extreme cases (such as multiple devices alarming simultaneously, or the risk not being eliminated after the first-level policy adjustment), the edge gateway will ignore the predetermined power scheduling target and instead execute a series of preset, most conservative safety instructions, such as immediately switching the energy storage system to standby or shutdown state, and supplying power to the highest priority loads (such as security and lighting) according to preset, extremely low guaranteed power levels, so as to ensure the absolute priority of personal and equipment safety.

[0048] During operation, the edge gateway continuously monitors and calculates multi-dimensional autonomous signal indicators such as command timeliness, link latency, ACK loop and sequence. When any one or more of these indicators continue to deteriorate and meet the preset trigger conditions and hysteresis rules, the system determines that the cloud control has failed, then freezes the cloud commands, records a status snapshot, and automatically switches to the network-off autonomous mode. In the autonomous mode, all commands generated by local control strategies, such as energy storage charging and discharging power and charging pile scheduling commands, must be processed by safety constraints such as dynamic power boundary trimming based on real-time SOC and protection status and flexible load scheduling based on priority before they can be issued for execution.

[0049] Compared to existing technologies, this workflow offers significant advantages. Existing solutions typically rely solely on network connectivity or simple heartbeats for judgment, which are prone to false triggers or slow responses under complex network jitter. This method, however, significantly improves the accuracy and robustness of state recognition through multi-dimensional index composite judgment and hysteresis mechanisms, avoiding frequent and unnecessary mode switching. Furthermore, existing technologies' outage protection strategies are often relatively simple and static, which may not be able to adapt to dynamic changes in device status and multi-objective optimization needs, posing risks of security breaches or business interruptions. In contrast, the multi-layered security constraint closed loop embedded in this workflow ensures that control commands during autonomous operation not only meet real-time optimization objectives but are also strictly limited by the device's current actual security boundaries and business priority rules. Thus, even without centralized optimization in the cloud, it can still achieve safe, continuous, and efficient local autonomous operation, enhancing the overall resilience of the virtual power plant system.

[0050] like Figure 2 As shown, Figure 2 The flowchart of the second embodiment of the offline take-off and cut-back method for the virtual power plant of the present invention is shown, which includes: S201. Obtain edge gateway data and calculate autonomous signal indicators based on the edge gateway data; S202. When the autonomous signal indicator meets at least one preset trigger condition and satisfies the hysteresis rule, the edge gateway is switched to the network disconnection autonomous mode. S203. During the network outage autonomous mode, generate local control commands based on security constraints; S204. When the cloud recovery candidate conditions are met, pause the execution of cloud commands and perform a consistency determination. Specifically, during the autonomous mode without network access, the edge gateway continuously probes the cloud status. When the following cloud recovery candidate conditions are met simultaneously, the cloud is considered to have stably recovered, and the switchback preparation phase can begin; the cloud recovery candidate conditions include cloud communication recovery, valid command timeliness, normal link latency, normal ACK loop closure, and normal command sequence. Cloud communication recovery: Within a preset continuous period, both the edge gateway and the cloud control center can normally receive the basic keep-alive signals sent periodically to confirm the other party's survival; receiving a heartbeat once or sporadically is not considered "normal" and must meet the "continuous" condition to prove that the link has entered a "stable" state from "unstable".

[0051] The command is valid: The timestamp or time-to-live of the cloud command received by the edge gateway is valid and has not expired, indicating that the command is "fresh" issued by the cloud, rather than an old command remaining in the network.

[0052] Link latency is normal: This means that the latency from when the edge gateway sends a request to when it receives a response from the cloud is within a preset threshold level. This condition applies hysteresis rules, meaning that the recovery standard is stricter than the standard for triggering autonomy.

[0053] ACK loop is normal: ACK loop is normal, which means that both ACK completion rate and ACK latency are normal. ACK completion rate measures the proportion of control commands sent from the cloud that are successfully received and confirmed for execution by field devices within a certain time window. ACK latency measures the time elapsed from when the edge gateway sends the command to the field device to when the device returns an execution confirmation.

[0054] Normal instruction sequence: This means that the latest received instruction sequence number is continuously increasing from the previously received sequence number (for example, the last one was 100, and this one is 101).

[0055] It should be noted that when the preset trigger condition is switched to the autonomous mode without network access or when the cloud recovery candidate condition is switched back to cloud control, if the system clocks of the cloud server and the edge gateway device are not synchronized, resulting in a difference in the current time recorded by the two, the edge receiving time and window statistics can be used to replace the TTL determination, and the edge snapshot can be pulled by the cloud to complete the correction.

[0056] Furthermore, the consistency determination verification conditions include, but are not limited to, the following verifications: State of Charge (SOC) consistency verification compares the SOC value of the energy storage system stored in the cloud with the SOC value collected in real time by the edge gateway. The deviation must be within the allowable error range. This is to prevent the cloud from issuing dangerous commands based on outdated SOC data. In other words, it verifies whether the SOC value of the energy storage system stored in the cloud is consistent with the real-time SOC value collected by the edge gateway. Charging queue consistency verification compares the charging pile session list and session status recorded in the cloud with the actual queue status on the edge gateway. This prevents the cloud from operating on sessions that have ended or not yet started. In other words, it verifies whether the charging pile session and status list recorded in the cloud is consistent with the actual charging queue status on the edge gateway. The protection status consistency verification compares the protection flag bit of the controlled device recorded in the cloud with the real-time protection status collected by the edge gateway; this prevents the cloud from still attempting to issue power commands when the device has actually failed, i.e., it verifies whether the protection flag bit of the controlled device recorded in the cloud is consistent with the real-time protection status collected by the edge gateway.

[0057] S205. Based on the consistency determination result, determine whether the edge gateway should switch back. If the edge gateway is determined to switch back, return control to the cloud according to the smooth switchback strategy. If the edge gateway is determined not to switch back, maintain the network outage autonomous mode and report the abnormal status information.

[0058] Specifically, decisions are made based on the consistency check results: If all verifications pass, it indicates that the cloud and edge states are consistent, and a smooth switchback is initiated, returning control of the edge gateway to the cloud. If any verification fails, the switchback is immediately prohibited, the edge gateway remains in autonomous mode without network access, and reports inconsistent status information to the cloud. Subsequently, the edge gateway can automatically request synchronization, wait for manual or automatic synchronization to resolve the issue, and then re-verify.

[0059] Furthermore, the smooth back-cut strategy includes, but is not limited to, a power ramping constraint strategy and a fusion coefficient increment strategy. Specifically, the power ramping constraint strategy is as follows:

[0060] in: Indicates the first The actual output power of the controlled device during the cut-back process; Indicates the first The current actual power of each controlled device; Indicates the first The cloud-based target power of each controlled device; Indicates the first The maximum descent ramp rate of each controlled device; Indicates the first The minimum ramp rate for each controlled device.

[0061] Compared to the first embodiment, this embodiment adds cloud recovery detection, pre-synchronization and consistency verification, and smooth back-off decision execution. When the edge gateway detects that the cloud heartbeat is continuously normal, the command timeliness is valid, and key link quality indicators such as RTT and ACK completion rate have recovered to a more stringent exit threshold during the autonomous period, it determines that the cloud recovery candidate is in place. At this time, the system enters the pre-synchronization state. In this state, the edge gateway receives and parses the cloud command but does not execute it. Instead, it initiates a triple consistency verification of the energy storage SOC, charging queue status, and device protection flag. If all verifications pass, the power of each controlled device is smoothly transitioned from the current value to the target value issued by the cloud at a controlled ramp rate according to the power ramp constraint formula, thereby completing the uninterrupted transfer of control. If any verification fails, back-off is immediately prohibited, the autonomous state is maintained, and an alarm is reported.

[0062] Compared to existing technologies, the core advantage of this work is that it overcomes the traditional weakness of the recovery period. Existing solutions often directly restore cloud scheduling after cloud recovery, ignoring the possibility that the cloud-edge state may have deviated during the network outage, which can easily lead to power surges. This embodiment, through proactive consistency verification in the pre-synchronization state and a safe rollback mechanism when verification fails, further provides a closed-loop path for a safe and smooth return from the autonomous state to centralized cloud control. The combination of these two forms a complete state machine that covers the occurrence of anomalies, continuous response, and recovery to normal, enabling the virtual power plant system to remain reliable, safe, and stable throughout the entire lifecycle of cloud-edge network anomalies, achieving true high availability.

[0063] Accordingly, the present invention also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the offline take-off and cut-back method for the virtual power plant described in any of the above claims.

[0064] In summary, the offline takeover and revert method and computer equipment for virtual power plants provided by this invention construct a complete technical closed loop, from cloud-edge collaborative status monitoring and multi-dimensional intelligent trigger judgment to local autonomous control with multi-layered security constraints, and then to smooth reverting based on strict consistency verification and power ramping. This method improves the accuracy of takeover judgment by introducing hysteresis rules and composite indicators, ensures the safety and business continuity of autonomous operation through dynamic security boundary trimming and flexible load scheduling, and ensures the shock-free and high reliability of the recovery process through pre-synchronization status and active verification mechanisms. Thus, it systematically solves the defects of rough triggering, unsafe autonomy, and shock-induced recovery in existing technologies, significantly enhancing the adaptability and operational resilience of virtual power plants in unstable communication environments such as weak networks and network outages. It is particularly suitable for scenarios with extremely high requirements for power supply continuity and control security, such as park microgrids, photovoltaic-storage-charging integrated power stations, and distributed energy aggregation, and has broad application value.

[0065] The above-disclosed embodiments are merely preferred embodiments of the present invention and should not be construed as limiting the scope of the present invention. Therefore, any equivalent variations made in accordance with the claims of the present invention are still within the scope of the present invention.

Claims

1. A method for offline take-off and cut-off of a virtual power plant, characterized in that, include: Acquire edge gateway data and calculate autonomous signal indicators based on the edge gateway data; When the autonomous signal indicator meets at least one preset trigger condition and satisfies the hysteresis rule, the edge gateway is switched to the network disconnection autonomous mode. The preset trigger conditions include one or more of the following: instruction timeout, link delay exceeding limit, ACK closed loop abnormality, and ACK sequence abnormality. During the network-disconnected autonomous mode, local control commands are generated based on security constraints.

2. The offline take-off and cut-back method for a virtual power plant as described in claim 1, characterized in that, The instruction timeout means that the timestamp or lifespan of the cloud instruction has expired; and / or The link delay exceeding the limit is defined as the mean and variance of the link round-trip delay exceeding a threshold within a sliding window; and / or The ACK closed-loop anomaly is based on an ACK completion rate lower than a completion rate threshold or an ACK delay higher than a delay threshold; and / or The ACK sequence anomaly is caused by the received cloud instruction sequence number being discontinuous.

3. The offline take-off and cut-back method for a virtual power plant as described in claim 1, characterized in that, The safety constraints include energy storage power boundary trimming, continuous power supply guarantee for charging queues, gradual power limiting when station-level constraints are insufficient, and secondary safety strategies when risks escalate.

4. The offline take-off and cut-back method for a virtual power plant as described in claim 1, characterized in that, Also includes: When the conditions for cloud recovery are met, pause the execution of cloud commands and perform a consistency check. Based on the consistency determination result, determine whether the edge gateway should switch back. When it is determined that the edge gateway is to switch back, control is returned to the cloud according to the smooth switchback strategy. If it is determined that the edge gateway will not switch back, the network outage autonomous mode will still be maintained and abnormal status information will be reported.

5. The offline take-off and cut-back method for a virtual power plant as described in claim 4, characterized in that, The cloud recovery candidate conditions include: cloud communication recovery, valid command timeliness, normal link latency, normal ACK closed loop, and normal command sequence; and / or The consistency determination and verification conditions include: State of Charge (SOC) consistency verification verifies whether the SOC value of the energy storage system stored in the cloud is consistent with the real-time SOC value collected by the edge gateway. Charging queue consistency verification verifies whether the charging pile session and status list recorded in the cloud are consistent with the actual charging queue status on the edge gateway. Protection status consistency verification verifies whether the protection flag bits of the controlled device recorded in the cloud are consistent with the real-time protection status collected by the edge gateway.

6. The offline take-off and cut-back method for a virtual power plant as described in claim 2, characterized in that, The formula for calculating the ACK completion rate is as follows: in, Indicates the ACK completion rate. This represents the number of valid execution confirmations actually received by the edge gateway from the controlled devices within the same time window. This represents the total number of control commands received by the edge gateway from the cloud within the specified evaluation time window; The formula for calculating the ACK delay is as follows: in, Indicates the confirmation delay for a single instruction. This indicates the local time at which the edge gateway receives confirmation of the execution of the corresponding instruction from the controlled device. This indicates the local time when the edge gateway sends a control command to the controlled device.

7. The offline take-off and cut-back method for a virtual power plant as described in claim 3, characterized in that, The steps of energy storage power boundary trimming include: The energy storage power boundary trimming is calculated using the following formula: in, This represents the total station-level power demand of the energy storage system calculated by the local control strategy of the edge gateway under autonomous mode without network access. This represents the lower boundary of the safe power based on the current state of charge of the energy storage and the real-time protection status of the equipment. This represents the upper boundary of the safe power based on the current state of charge of the energy storage and the real-time protection status of the equipment. Indicates the state of charge. Indicates the real-time protection status of the equipment. This indicates the power value of the command executed by the energy storage system.

8. The offline take-off and cut-back method for a virtual power plant as described in claim 4, characterized in that, The smooth back-cut strategy includes a power ramping constraint strategy, specifically: in, Indicates the first The actual output power of the controlled device during the cut-back process Indicates the first The current actual power of the controlled device. Indicates the first The cloud-based target power of the controlled device. Indicates the first The maximum descent ramp rate of the controlled equipment. Indicates the first The minimum ramp rate for each controlled device.

9. The offline take-off and cut-back method for a virtual power plant as described in claim 3, characterized in that, The method for ensuring continuous power supply of the charging queue includes: Priority calculations are performed on several charging facilities, and power supply is guaranteed based on the calculation results; The priority calculation steps include: in, The sequence is represented as Priority rating of charging facilities The sequence is represented as The basic preset priority of charging facilities The sequence is represented as The importance of continuous power supply for charging facilities, The sequence is represented as The charging facilities correspond to the current state of charge of the equipment. The sequence is represented as The charging cutoff pressure of charging facilities The weighting coefficient represents the basic preset priority. Weighting coefficients representing the importance of continuous power supply This represents the weighting coefficient of the device's current state of charge. This represents the weighting coefficient for the cutoff charging pressure.

10. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the offline take-off and cut-back method for the virtual power plant as described in any one of claims 1 to 9.

Images