Power station digital twin intelligent management and control system for unattended scene

Abstract

The application relates to the technical field of smart grids, and discloses a power station digital twin intelligent management and control system for unattended scenarios, which comprises the following steps: constructing a time scale alignment sampling set; constructing an electrical mirror data set; calculating a collection time lag, a time lag power drift amount and a twin electrical residual amount; calculating a remote control synchronization credible margin; generating a switch target value; generating a remote control action permission word and an expected execution time; and generating a remote control instruction frame. In the remote switch operation process of the unattended power station, the application comprehensively considers various factors such as collection transmission delay, new energy output fluctuation, deviation between the twin model and the field state, and electrical overrun after action, uniformly represents the credibility of the remote control action, and outputs the action instruction only when the data synchronization, time delay level and action necessity all meet the requirements, so that the misaction generated based on the lag state can be reduced, and the operation safety of the remote control of the unattended power station is improved.

Description

Technical Field

[0001] This application relates to the field of smart grid technology, specifically to a digital twin intelligent management and control system for power plants in unattended scenarios. Background Technology

[0002] As the scale of distributed photovoltaic, energy storage and other new energy power plants expands, the unmanned operation mode is gradually being promoted. Digital twin systems are used to carry the status, topology and operation data of power plant equipment, support remote equipment operation and fault handling, reduce the configuration of on-site operation and maintenance personnel and reduce the operation and maintenance costs of power plants.

[0003] The remote switching operation of existing unattended power stations is usually based on the collected electrical quantities and switch status. After simulation verification in a twin model, remote control commands are generated and sent to the field switching mechanism via a communication link to achieve fault isolation and power restoration.

[0004] In actual operation, due to factors such as data acquisition and transmission delays, asynchronous timing of multiple measurement points, and rapid fluctuations in new energy output, the electrical state in the twin model may deviate from the actual on-site state. Remote control commands generated based on the lagging state may result in situations such as electrical quantities exceeding limits, switch malfunctions, or failures to operate when the action is implemented, affecting the safe operation of the power station. Summary of the Invention

[0005] This application provides a digital twin intelligent control system for power plants in unattended scenarios, which solves the technical problems mentioned in the background.

[0006] This application provides a digital twin intelligent control system for power plants in unattended scenarios, including:

[0007] The sampling time-stamp alignment module encapsulates the electrical operating quantities, actual switch values, measurement timestamps, and arrival timestamps accessed by the power station acquisition terminal into sampling windows and outputs a time-stamp aligned sampling set.

[0008] The twin flip-over simulation module synchronizes the equipment objects, topology connections, and node injection boundaries of the digital twin model of the power station based on the time-scale aligned sampling set. It performs single-switch flip-over simulation for each remotely controlled switch and outputs an electrical mirror dataset.

[0009] The time delay drift residual calculation module calculates the acquisition time delay, time delay power drift, and twin electrical residual within the set of affected nodes and the set of affected branches for each remotely controlled switch based on the time-scale aligned sampling set and the electrical mirror dataset, and outputs the acquisition time delay, time delay power drift, and twin electrical residual.

[0010] The reliability margin calculation module calculates the proposed action over-limit based on the proposed action electrical quantity and allowable limit, and couples the proposed action over-limit with the time-delay power drift and twin electrical residual to output the remote control synchronization reliability margin.

[0011] The switch target decision module forms a unique switch target value based on health indication data, the limit of the proposed action, the remote control synchronization confidence margin, and the actual switch measurement value.

[0012] The remote control license generation module generates a remote control action license word and the expected execution time based on the switch target value, the actual switch value, the remote control synchronization confidence margin, and the acquisition delay in the set of affected nodes.

[0013] The instruction frame generation module generates a remote control instruction frame with a cyclic redundancy check code according to the switch address, switch target value, remote control action permission word, and expected execution time.

[0014] The beneficial effects of this application are as follows: During the remote switching operation of unattended power plants, this application comprehensively considers multiple factors such as data acquisition and transmission delay, fluctuations in renewable energy output, deviations between the twin model and the actual situation, and electrical limit violations after the action. It provides a unified characterization of the reliability of remote control actions, outputting action commands only when data synchronization, latency levels, and the necessity of the action all meet the requirements, thus reducing erroneous actions based on lagging states. The scheme employs a local area calculation method, which reduces the computational pressure of each round of sampling and adapts to the computing power level of edge nodes. Simultaneously, through mechanisms such as state conflict verification, timing interlocking, and phased actions, it avoids problems such as multi-source command conflicts, premature twin state updates, and the use of backup power supplies with faults. This adapts to the operating scenarios of rapidly fluctuating renewable energy power plant output, improving the operational safety of remote control of unattended power plants. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the specific embodiments of this application or the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0016] Figure 1 A schematic diagram of the modules of a digital twin intelligent control system for power plants in unattended scenarios provided in this application embodiment;

[0017] Figure 2 A deployment diagram of a digital twin intelligent control system for power plants in unattended scenarios, provided in an embodiment of this application;

[0018] Figure 3This is an experimental schematic diagram illustrating the remote control permission determination boundary under different switching action scenarios provided in the embodiments of this application. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the protection scope of this application.

[0020] It should be noted that, in the description of this application, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. The terms "first," "second," etc., in this application are used to distinguish similar objects and are not used to describe a specific order or sequence.

[0021] To enable those skilled in the art to better understand the present application, the present application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0022] like Figure 1 As shown, the digital twin intelligent control system for power plants designed for unattended operation includes:

[0023] The sampling time-stamp alignment module encapsulates the electrical operating quantities, actual switch values, measurement timestamps, and arrival timestamps accessed by the power station acquisition terminal into sampling windows and outputs a time-stamp aligned sampling set.

[0024] The twin flip-over simulation module synchronizes the equipment objects, topology connections, and node injection boundaries of the digital twin model of the power station based on the time-scale aligned sampling set. It performs single-switch flip-over simulation for each remotely controlled switch and outputs an electrical mirror dataset.

[0025] The time delay drift residual calculation module calculates the acquisition time delay, time delay power drift, and twin electrical residual within the set of affected nodes and the set of affected branches for each remotely controlled switch based on the time-scale aligned sampling set and the electrical mirror dataset, and outputs the acquisition time delay, time delay power drift, and twin electrical residual.

[0026] The reliability margin calculation module calculates the proposed action over-limit based on the proposed action electrical quantity and allowable limit, and couples the proposed action over-limit with the time-delay power drift and twin electrical residual to output the remote control synchronization reliability margin.

[0027] The switch target decision module forms a unique switch target value based on health indication data, the limit of the proposed action, the remote control synchronization confidence margin, and the actual switch measurement value.

[0028] The remote control license generation module generates a remote control action license word and the expected execution time based on the switch target value, the actual switch value, the remote control synchronization confidence margin, and the acquisition delay in the set of affected nodes.

[0029] The instruction frame generation module generates a remote control instruction frame with a cyclic redundancy check code according to the switch address, switch target value, remote control action permission word, and expected execution time.

[0030] In one alternative implementation, the sampling time-stamp alignment module is configured to perform:

[0031] Step S101: Collect the electrical operating quantities, actual switch values, measurement timestamps and arrival timestamps connected to the power station acquisition terminal. The electrical operating quantities consist of electrical node voltage, branch current and net active power injection of electrical nodes. The actual switch values ​​include open value and closed value.

[0032] Electrical node voltage refers to the power frequency effective voltage of each electrical connection point in the power supply and distribution network of the power station. It is collected in real time by a 0.5-precision voltage transformer deployed at the corresponding node bus with a fixed period of 100 milliseconds. The collected value is then uploaded to the edge aggregation node after analog-to-digital conversion.

[0033] Branch current refers to the power frequency effective value current of each cable segment, overhead line segment or feeder segment in the power supply and distribution network of the power station. It is collected in real time by a 0.5-level precision current transformer deployed in the switch cabinet at the head of the corresponding branch with a fixed period of 100 milliseconds. The collected value is then uploaded to the edge aggregation node after analog-to-digital conversion.

[0034] Net active power injection at electrical nodes refers to the total active power injected into the distribution network by each electrical connection point. It is collected in real time by a 0.2-accuracy power measurement device at the corresponding grid connection point, energy storage grid connection point, or load inlet, with the direction of flow into the distribution network as the positive direction.

[0035] The measured value of the switch refers to the actual open and closed position of the remotely controlled switch on site. It is collected by the normally open and normally closed contacts of the switch body in a polling method that triggers the change of position immediately and adds a 100-millisecond cycle. 0 represents the open state and 1 represents the closed state.

[0036] The measurement timestamp refers to the moment when the acquisition end completes single-point single-round data sampling and analog-to-digital conversion. It is written by the local clock of the acquisition end. The acquisition end and the aggregation node achieve sub-millisecond clock synchronization through the station's PTP protocol. It adopts a unified time reference and does not need to rely on external time sources such as Beidou or GPS.

[0037] The arrival timestamp refers to the moment when the edge aggregation node fully receives and verifies a single-point, single-round sampling data frame. It is written by the local clock of the aggregation node and kept synchronized with the clock of the acquisition end.

[0038] Step S102: Mark the sampling window with the sampling sequence number and record the sampling interval between adjacent sampling windows;

[0039] The sampling sequence number is a unique integer number assigned to each round of full-station synchronous sampling within the same running session, starting from 1 and incrementing sequentially according to the sampling time. The number is incremented by 1 after each round of sampling is completed. After the device restarts, a new session identifier is generated, and the sampling sequence number starts counting again from 1 to avoid duplicate historical data sequence numbers.

[0040] The sampling window refers to a half-open interval centered on the station's baseline sampling time corresponding to the current sampling sequence number. The range is from 50 milliseconds before the baseline time to 50 milliseconds after the baseline time, with the left side closed and the right side open. Data falling on the upper boundary of the interval, i.e., exactly 50 milliseconds after the baseline time, is assigned to the next sampling sequence number.

[0041] The deadline for data reception is 70 milliseconds after the baseline time. Data that has not been received by the deadline is marked as missing data.

[0042] The sampling interval refers to the time difference between the reference times of two adjacent rounds of full-station sampling, with a value of 100 milliseconds. This interval can cover power fluctuation scenarios within seconds, while avoiding excessive computational pressure on edge nodes due to excessively high sampling frequency.

[0043] Step S103: Encapsulate the electrical node voltage, branch current, net active power injection of the electrical node, actual switch value, measurement timestamp, and arrival timestamp into a time-aligned sampling set according to the device's unique address, measurement point type, and sampling sequence number.

[0044] The time-stamped sampling set refers to a standardized dataset formed by sorting and encapsulating the electrical quantities, switch states, and corresponding dual timestamps of all measurement points under the same sampling sequence number according to the unique device address from smallest to largest. After the dataset is generated, it is passed to the subsequent functional modules without any additional data smoothing or interpolation processing.

[0045] Furthermore, when encapsulating the time-stamped sample set, the data is only included in the dataset corresponding to the current sampling number if the cyclic redundancy check of the measurement point data is correct, the measurement timestamp corresponding to the data falls within the current sampling window range, and arrives before the data reception deadline.

[0046] Furthermore, missing measurement points are not considered valid measured data for the current sampling sequence. To maintain continuous display of the twin model, the valid value of the previous sampling period can be used as a substitute value. The substitute value carries a reuse flag and data age. If there are reused values, verification failure values, or timeout values ​​in the target switch influence set, the corresponding remote control action permission word is locked.

[0047] Furthermore, the net active power injection of electrical nodes is uniformly positive when flowing into the distribution network. The grid connection injection value of photovoltaic modules and energy storage devices in the discharge state is positive, while the grid connection injection value of station load nodes and energy storage devices in the charging state is negative. The actual measured value of the switch uniformly adopts the coding rule that 0 represents the open switch and 1 represents the closed switch. All acquisition devices in the entire station adopt the completely consistent coding rule, and there is no situation of reverse coding or multiple sets of codes running in parallel.

[0048] Furthermore, for the voltage transformer at the bus node of the 35 kV step-up substation of a 10 MW mountain photovoltaic power station, the measurement timestamp during the 1256th sampling round was 14:25:32.120 on the same day. The edge aggregation node fully received the data and completed the verification at 14:25:32.142 on the same day. The data measurement timestamp fell within the half-open sampling window centered at 14:25:32.100 on the same day. The data verification bit was correct with no packet loss, and it arrived before the reception deadline. Therefore, the voltage value, the corresponding measurement timestamp and arrival timestamp, and the unique address of the device were written together into the 1256th time-aligned sampling set.

[0049] Understandably, using a fixed-width half-open sampling window combined with dual timestamps for measurement and arrival to time-stamp and encapsulate electrical data from multiple measurement points does not require all measurement points to have completely identical physical sampling times. It only constrains the data's scope through time windows while retaining the independent sampling and reception times of each measurement point. This provides a complete original time basis for subsequent time delay calculations, and only requires on-site network clock synchronization to achieve standardized alignment of multi-source heterogeneous data. This method enables unified and standardized encapsulation of time stamps for multi-source heterogeneous operating data, eliminating the need for external high-precision time synchronization equipment to merge data from different manufacturers and types of measurement points. This provides a traceable input foundation for subsequent twin model synchronization and reliable computation, reducing the configuration requirements and maintenance costs of on-site clock synchronization equipment in unattended power plants.

[0050] In one alternative implementation, the twin flip simulation module is configured to perform:

[0051] Step S201: Map the unique address of the device in the time-stamped aligned sampling set to the three-dimensional device object in the digital twin model of the power station, and uniquely bind the three-dimensional device object to the electrical node, branch or remotely controlled switch.

[0052] A power plant digital twin model is a fully digital three-dimensional mirror model constructed according to the actual electrical main wiring, three-dimensional spatial position of equipment, equipment nameplate parameters and operating limits of the power plant. It is obtained by joint calibration of power plant as-built drawings, equipment factory records and on-site measured values. The model accuracy meets the requirements of millimeter-level spatial positioning and electrical quantity calculation.

[0053] A 3D device object is a digital entity in a twin model that corresponds one-to-one with each physical device on site. Each object is bound to the physical device on site through a unique asset code issued by the manufacturer. Each object corresponds to only one physical device on site.

[0054] Step S202: Generate the current topology based on the measured values ​​of the switches, write the net active power injection of the electrical nodes into the node injection boundary of the current topology, and calculate the twin node voltage and twin branch current.

[0055] The current topology refers to the distribution network structure generated based on the measured values ​​of all switches in the current cycle, reflecting the actual electrical connection relationship of the power station at present. In the topology, there are electrical connections between connected nodes and no electrical paths between disconnected nodes.

[0056] The node injection boundary refers to the fixed boundary setting that uses the net active power injection and reactive power injection of each electrical node in the current round as power boundary conditions input into the power flow calculation program. The boundary values ​​remain unchanged during the calculation process until the current round of calculation converges.

[0057] The electrical calculation employs a three-phase balanced AC power flow algorithm, designating the connection node between the high-voltage side of the substation's main transformer and the external power grid as the balanced node, and the photovoltaic grid-connected point and energy storage grid-connected point as PQ nodes. Reactive power injection is not treated as a newly collected electrical quantity; instead, it is calculated based on the net active power injection of the electrical node corresponding to the current sampling sequence number and the pre-configured power factor. Specifically, the absolute value of the net active power injection of the electrical node is first taken, then the inverse cosine of the pre-configured absolute power factor is taken, and the tangent of the inverse cosine is calculated. Subsequently, the absolute value of the net active power injection of the electrical node, the tangent, and the reactive power direction flag are multiplied to obtain the reactive power injection value. When reactive power is injected into the distribution network, the reactive power direction flag is set to 1; when reactive power is absorbed by the distribution network, the reactive power direction flag is set to -1. Model parameters include line resistance, line reactance, ground susceptance, transformer turns ratio, and tap position. The convergence error of the power flow calculation is set to 10 to the power of -6, and the maximum number of iterations is set to 20.

[0058] Twin node voltage refers to the effective value of steady-state voltage of each electrical node obtained by power flow calculation of the distribution network based on the current topology and node injection boundary conditions;

[0059] Twin branch currents refer to the effective values ​​of steady-state currents at the head of each branch obtained through power flow calculations based on the current topology and node injection boundary conditions.

[0060] Both the measured branch current and the twin branch current use the branch start-up current value. They are compared according to the same measurement terminal, the same three-phase equivalent diameter, the same effective value definition, and the same sampling number. Before calculation, they are uniformly converted to nominal values ​​in the same unit.

[0061] Step S203: Perform single-switch flip simulation for each remote-controlled switch. The single-switch flip simulation is to only change the measured value of the simulated remote-controlled switch, while the other remote-controlled switches keep their respective measured values, and calculate the voltage of the simulated operating node and the current of the simulated operating branch.

[0062] The simulated operating node voltage refers to the effective value of the steady-state voltage of each electrical node obtained by the same power flow calculation method when only the state of the target switch is reversed, while all other switches remain unchanged in the current measured state.

[0063] The simulated operating branch current refers to the effective value of the steady-state current at the head end of each branch obtained by the same power flow calculation method when only the state of the target switch is reversed, while all other switches remain unchanged in the current round of measured state.

[0064] If a power-free island is formed after the switch flips, the nodes in the island are marked as power-depleted nodes and are not calculated as normally energized nodes; if the power flow calculation does not converge, the corresponding simulated action result is invalid and the switch action is blocked.

[0065] Step S204: Determine the set of affected nodes and the set of affected branches based on the topological connectivity components and power flow change branches before and after the switch flips.

[0066] The set of affected nodes is the union of nodes whose connected components change before and after the switch flips, and nodes whose voltage changes by more than 0.5% of the rated voltage.

[0067] The affected branch set is the union of the branches whose connection state changes after the switch flips and the branches whose current change exceeds 1% of the rated current.

[0068] If the set is empty after threshold filtering, at least the nodes at both ends of the target switch and its directly connected branches should be used as the minimum impact set to avoid the denominator being zero during calculation.

[0069] Step S205: Generate health indication data based on temperature data, insulation data, alarm data, protection action data, inspection data, and historical operation data;

[0070] Temperature data refers to the real-time temperature values ​​of key measuring points of each device. It is collected by contact temperature sensors with an accuracy of 1 degree Celsius deployed on the device casing, cable joints, and switch contacts at a fixed cycle of 1 minute. The data is valid for 2 minutes.

[0071] Insulation data refers to the insulation resistance and partial discharge data of each high-voltage equipment. The data is collected by the insulation monitoring device built into the equipment at a fixed cycle of 10 minutes. When the data is abnormal, the change report is triggered immediately. The data is valid for 30 minutes.

[0072] Alarm data refers to various equipment abnormal alarm information generated by the power plant monitoring system, including alarms such as over-temperature, over-voltage, over-current, and communication interruption. The monitoring system reports the alarm immediately in real time when it is generated, and the data is valid until the alarm is reset.

[0073] Protection action data refers to the action event record of relay protection device, including information such as protection action type, action time, and action value. It is reported by the protection device immediately in real time when the action is triggered, and is only effective within a valid time window of 5 minutes after the event has not been reset or after the action.

[0074] Inspection data refers to data such as equipment appearance abnormalities, abnormal hot spots, and discharge traces collected by drones or track inspection robots in the station. Discharge traces are triggered by image recognition results or manual confirmation markings. After the inspection equipment completes a single round of inspection tasks, the data is uploaded to the edge node in batches. The data is valid for 7 days.

[0075] Historical operating data refers to all equipment operating data, alarm records, maintenance records, and defect records stored by the power station from the date of commissioning. These data are uniformly stored in a historical database deployed locally by the power station, and the data retention period is no less than the designed operating years of the power station.

[0076] Health indication data refers to the indication information of the branch that needs to be isolated or the branch to be put into standby, which is obtained by judging the status data of the above-mentioned multi-source equipment through fixed thresholds. Each judgment threshold is taken from the equipment nameplate, manufacturer's technical documents, station operation procedures or pre-configured equipment threshold table.

[0077] Step S206: Output electrical mirror dataset, which consists of twin node voltage, twin branch current, simulated action node voltage, simulated action branch current, affected node set, affected branch set, and health indication data.

[0078] The electrical mirror dataset refers to a standardized dataset that includes the current topology twin electrical quantities, the simulated action electrical quantities of a single switch, the set of influences of each switch, and health indication data. After being encapsulated according to the sampling sequence number, it is passed to subsequent modules.

[0079] Furthermore, when binding 3D equipment objects with electrical nodes, branches, and remote-controlled switches, a one-to-one correspondence is established strictly according to the unique asset code in the equipment ledger. For multi-break switchgear composed of multiple breaks connected in series, all series breaks are equivalent to a remote-controlled switch object according to the equivalent logic of the main electrical wiring, and no separate switch object is created for each independent break.

[0080] Furthermore, during the single-switch flip simulation, except for the current target switch whose state is reversed, the states of all other switches are strictly kept unchanged from the current round's measured values ​​without any manual adjustment. The current topology and the simulated action topology use the exact same power flow calculation method, iterative convergence threshold, and calculation step size, and different precision calculation settings are not used for different topologies.

[0081] Furthermore, the network where a certain 35 kV sectionalizing switch is located is a ring network structure, and there is another closed electrical path between the two ends of the switch. Therefore, after performing a switching-off simulation on the switch, the network connectivity component did not change. The calculated voltage changes of the three downstream photovoltaic collector line nodes are 2.1%, 1.7%, and 0.8% of the rated voltage, respectively. The corresponding current changes of the four collector line branches are 3.2%, 2.5%, 1.2%, and 0.3% of the rated current, respectively. Therefore, the three nodes with voltage changes exceeding 0.5% of the rated voltage are included in the set of affected nodes of the switch, and the three branches with current changes exceeding 1% of the rated current are included in the set of affected branches of the switch. Branches with current changes of only 0.3% are not included in the scope of influence.

[0082] Furthermore, when generating health indicator data, if the temperature at the equipment measuring point exceeds the maximum allowable temperature of the equipment by more than 10 degrees Celsius, the insulation resistance is lower than the minimum value required by the regulations, the relay protection action is within the validity period, or the inspection confirms the presence of discharge traces, the corresponding branch is marked as a branch that needs to be isolated, and the boundary switches on both sides of the branch are set to the open state.

[0083] Furthermore, fault isolation and backup power activation are executed in stages. When multiple remotely controllable switches meet the action permission conditions within the same sampling period, the remote control action permission word of only one remotely controllable switch is set to 1, and the remote control action permission words of the remaining remotely controllable switches are set to 0, according to the rule that fault isolation action takes precedence over backup power activation action and similar actions are ordered in ascending order of switch address. After receiving the measured position data of the remotely controllable switch and updating the current topology, the action permission of the remaining candidate switches is recalculated in the next sampling period. The boundary switches on both sides of the fault branch are opened one by one in ascending order of address. After each switch is opened and the measured position is received, the simulation is re-verified based on the new topology. The next boundary switch is opened only after confirming that there are no abnormalities. After all boundary switches of the fault branch are opened and the fault is confirmed to be isolated, the power flow simulation and over-limit verification are re-executed based on the new isolated topology. After the verification is passed, the closing command of the backup power activation switch is generated in the next sampling period. The opening and closing commands are not issued simultaneously in the same command frame.

[0084] Understandably, a single-switch state reversal simulation is performed for each remotely controlled switch, ensuring that the states of other switches are completely consistent with the actual measured values. The influence range is defined by combining connectivity changes and electrical quantity change thresholds. Subsequent reliable calculations are then performed only within this localized area, eliminating the need for a full traversal calculation of all electrical quantities across the entire station. This approach reasonably defines the electrical influence range of each switch action, narrowing the scope of subsequent reliable calculations from the entire station to the localized area related to the switch action. This avoids the steady-state average error of the entire station masking the synchronization deviation in the localized area, while simultaneously reducing the computational load per sampling round, thus adapting to the millisecond-level real-time control requirements of unattended power stations.

[0085] In one alternative implementation, the time-delay drift residual calculation module is configured to perform:

[0086] Step S301: Calculate the acquisition delay based on the measurement timestamp and arrival timestamp;

[0087] Acquisition delay refers to the time difference between the completion of data sampling at a single measurement point and the complete reception and verification by the edge aggregation node. It is calculated by subtracting the measurement timestamp from the arrival timestamp of the corresponding measurement point, and the time unit is uniformly used in the calculation.

[0088] Specifically, the formula for calculating the acquisition delay is as follows:

[0089]

[0090] In the formula, For the first The electrical node at the first The acquisition delay of each sampling sequence number For the first The electrical node at the first The arrival timestamp of each sampling sequence number For the first The electrical node at the first The measurement timestamp of each sampling sequence number.

[0091] Step S302: Calculate the time-delay power drift based on the net active power injection, baseline active power, power stability term, and sampling interval of the electrical nodes of the current and previous samples in the affected node set.

[0092] The reference active power refers to the rated active power value of each electrical node. For a single equipment node, the rated power is taken from the equipment nameplate. For a composite node, the absolute value of the rated active power of the power generation, energy storage and load equipment connected to the node is taken from the sum of the absolute values ​​of the rated active power. It is obtained through the calibration of the equipment nameplate parameters at the factory.

[0093] The power stabilization term is a fixed small value set to avoid calculation overflow when the rated power is zero or at a minimum. It is set to 0.1 kW to ensure that the denominator of the power calculation for all nodes always remains positive and there will be no division by zero error.

[0094] The time-delay power drift refers to the normalized power offset value in the affected area corresponding to the switch, which is determined by the time delay of each measuring point and the power change in adjacent sampling periods. The value is a non-negative number, and the normal calculation result is not subject to upper limit truncation. In the case of missing active power data of the corresponding node in the previous sampling period, abnormal timestamp, or other specified data abnormalities, the time-delay power drift is written as the abnormal set value 1, so that the corresponding remote control synchronization confidence margin is lower than the confidence margin lower limit and the action is blocked.

[0095] Specifically, the formula for calculating the time-delay power drift is as follows:

[0096]

[0097] In the formula, For the first The remote-controlled switch is in the first The time-delay power drift of each sampling sequence number For the first A set of nodes affected by a remotely controlled switch. For the first The number of electrical nodes within the set of nodes affected by a remotely controlled switch. For the first The electrical node at the first Net active power injection of electrical nodes with sampling sequence number For the first The electrical node at the first Net active power injection of electrical nodes with sampling sequence number For the first The electrical node at the first The acquisition delay of each sampling sequence number For the first The baseline active power of each electrical node For the first Power stability terms of each electrical node, The sampling interval is denoted as .

[0098] Step S303: Calculate the twin electrical residual based on the branch current and twin branch current within the affected branch set and the electrical node voltage and twin node voltage within the affected node set.

[0099] Rated current refers to the effective value of the current that each branch can continuously operate under a specified ambient temperature. It is obtained by calibration after correcting the nominal current carrying capacity parameters of the cable or conductor in combination with the on-site laying conditions.

[0100] The current stabilization term is a fixed small value set to avoid calculation overflow when the rated current is zero or at a minimum. It is set to 0.1 amps to ensure that the denominator of the current calculation for all branches always remains positive and there will be no division by zero error.

[0101] Rated voltage refers to the nominal operating voltage effective value of each electrical node, which is obtained through calibration according to the voltage level standard corresponding to the power station.

[0102] The voltage stabilization term is a fixed small value of 0.1 volts set to avoid calculation overflow when the rated voltage is zero or at a minimum value. This ensures that the denominator of the voltage calculation for all nodes always remains positive and will not cause division by zero errors.

[0103] Twin electrical residual refers to the normalized average relative deviation between the measured electrical quantity and the calculated electrical quantity of the twin model within the influence area corresponding to the switch, and the value is a non-negative number;

[0104] Specifically, the formula for calculating the twin electrical residual is as follows:

[0105]

[0106] In the formula, For the first The remote-controlled switch is in the first The twin electrical residual of each sampling sequence number For the first The set of branches affected by a remotely controlled switch For the first The number of branches within a branch set is affected by a remotely controlled switch. For the first The branch road in the Branch current of each sampling number For the first The branch road in the The twin branch current of each sampling sequence number, For the first The rated current of the branch circuit, For the first The current stability term of the branch circuit For the first A set of nodes affected by a remotely controlled switch. For the first The number of electrical nodes within the set of nodes affected by a remotely controlled switch. For the first The electrical node at the first The voltage of the electrical node with each sampling sequence number. For the first The electrical node at the first Twin node voltage of each sampling sequence number For the first The rated voltage of each electrical node, For the first Voltage stability term of each electrical node.

[0107] Step S304: Output the acquisition time delay, time delay power drift, and twin electrical residual.

[0108] Furthermore, when calculating the time-delay power drift, if the current sampling lacks the active power data of the corresponding node from the previous sampling period, or if the acquisition time delay of a certain measuring point is negative (i.e., the receiving time is earlier than the measurement time), the timestamp of that measuring point is marked as abnormal, the time-delay power drift of the corresponding switch is set to 1, and the switch action is blocked during subsequent remote control permission judgment.

[0109] Furthermore, when calculating the twin electrical residual, the arithmetic mean of the relative current deviation of all branches in the affected area and the arithmetic mean of the relative voltage deviation of all nodes in the affected area are calculated separately. The two arithmetic means are added together to obtain the final residual, without using weighted calculation or other nonlinear merging methods.

[0110] Furthermore, the area affected by a certain switch contains two photovoltaic grid-connected nodes with rated powers of 5 MW and 3 MW, respectively. The power stability term is taken as 0.1 kW. The active power difference between the current and previous sampling rounds is 0.2 MW and 0.12 MW, respectively. The acquisition time delays of the corresponding measurement points are 20 ms and 30 ms, respectively. The sampling interval is 100 ms. The time unit is uniformly used in the calculation. The power drift components of the two nodes are 0.008 and 0.012, respectively. After taking the arithmetic mean, the power drift due to time delay of the switch is 0.01, which reflects that the power drift has a small impact on synchronization under the current time delay level.

[0111] Understandably, within the local influence range corresponding to the switch, the time-delay power drift is calculated by combining the independent acquisition delay of each measuring point with the power change rate of adjacent sampling periods. Simultaneously, the average relative deviations of voltage and current between the measured values ​​and the twin-model calculations within the local area are calculated. Both types of deviations are used as the quantitative source of synchronization risk, rather than using the average deviation across the entire station as the judgment criterion. This method can quantify the degree of synchronization offset caused by the coupling of acquisition delay and rapid power fluctuations, identify the electrical deviations between the twin model and the physical power station within the local area of ​​the switch, capture the synchronization mismatch risk in a small area, and avoid the whole-station steady-state calculation masking local anomalies.

[0112] In one alternative implementation, the confidence margin calculation module is configured to perform:

[0113] Step S401: Perform an over-limit comparison on the proposed operating branch current, proposed operating node voltage, upper limit of allowable current, upper limit of allowable voltage, and lower limit of allowable voltage to obtain the proposed operating over-limit amount;

[0114] The upper limit of permissible current refers to the maximum permissible continuous operating current value of each branch under the current operating ambient temperature, which is obtained by combining the nominal current carrying capacity of the cable with the correction factor of the ambient temperature at the site.

[0115] The upper limit of permissible voltage refers to the highest permissible effective value of the operating voltage of each electrical node, which is uniformly taken as 1.1 times the rated voltage and obtained by calibration through the permissible voltage deviation range specified in the power plant operation regulations;

[0116] The lower limit of permissible voltage refers to the minimum permissible effective value of the operating voltage of an electrical node that should remain energized after the planned operation, and is taken as 0.9 times the rated voltage. For nodes that are expected to lose power due to fault isolation operation, they are marked as planned power loss nodes, and the lower limit of permissible voltage for this isolation operation is taken as 0. The voltage drop caused by the planned power loss is not included in the undervoltage over-limit component. The voltage of the planned operation node of the planned power loss node is taken as 0, and the current of the planned operation branch connected to the planned power loss node and no longer supplied with power is taken as 0.

[0117] The simulated action limit exceeds the limit, which is the sum of the maximum relative current limit and the maximum relative voltage limit after the switch is flipped to the target state. The value is a non-negative number.

[0118] Specifically, the calculation formula for the proposed action exceeding the limit is as follows:

[0119]

[0120] In the formula, For the first The remote-controlled switch is in the first The more limited the simulated action for each sampling sequence number, the better. For the first The set of branches affected by a remotely controlled switch For the first When a remotely controlled switch performs a single-switch toggling simulation, the first... The branch road in the The simulated operating branch current of each sampling sequence number, For the first The maximum allowable current for each branch. For the first The current stability term of the branch circuit For the first A set of nodes affected by a remotely controlled switch. For the first When a remotely controlled switch performs a single-switch toggling simulation, the first... The electrical node at the first The voltage of the simulated action node with each sampling sequence number For the first The upper limit of the permissible voltage for each electrical node. For the first The lower limit of the permissible voltage for each electrical node. For the first Voltage stability term of each electrical node, for and The larger value in the range.

[0121] Step S402: Sum and normalize the time-delay power drift, twin electrical residual, and simulated action over-limit to obtain the remote control synchronization confidence margin.

[0122] Remote control synchronization reliability margin refers to the reliability value obtained by adding the time delay power drift, twin electrical residual, and simulated action overrun limit and normalizing it through the reference term. The value ranges from 0 to 1, and the higher the value, the higher the synchronization reliability.

[0123] Specifically, the formula for calculating the remote control synchronization confidence margin is as follows:

[0124]

[0125] In the formula, For the first The remote-controlled switch is in the first Remote synchronization reliability margin for each sampling sequence number For the first The remote-controlled switch is in the first The time-delay power drift of each sampling sequence number For the first The remote-controlled switch is in the first The twin electrical residual of each sampling sequence number For the first The remote-controlled switch is in the first The simulated action of each sampling sequence number exceeds the limit.

[0126] Step S403: Output the simulated motion limit and the remote control synchronization confidence margin.

[0127] Furthermore, when calculating the over-limit quantities, only the portion of current exceeding the upper limit and the portion of voltage exceeding the upper and lower limits are included in the calculation. The component that does not reach the over-limit threshold is set to 0. The maximum value of the current over-limit component and the maximum value of the voltage over-limit component are added together. The average or cumulative value of the over-limit quantities is not used as the result.

[0128] Furthermore, after simulation calculation of a switch closing, the calculated currents of the three branches in the affected area are 0.98 times, 1.05 times, and 0.92 times the upper limit of the allowable current, respectively. The calculated voltages of the four nodes are 1.02 times, 1.12 times, 0.95 times, and 0.88 times the rated voltage, respectively. The allowable voltage range is 0.2 times the rated voltage. The calculated current over-limit component is 0.05, and the voltage over-limit component is 0.10. Therefore, the simulated over-limit amount of the switch is 0.15, reflecting the risk of over-limit after the switch is closed.

[0129] Understandably, by summing the non-negative risk quantities of three different dimensions—time-delay power drift, twin electrical residuals, and simulated action exceeding limits—and normalizing them using a fixed value of 1 as the baseline, the three types of risks are uniformly mapped to the interval between 0 and 1. The higher the risk, the lower the remote control synchronization confidence margin. This approach achieves a unified representation of multi-dimensional risks without the need for complex weighting coefficients for the three types of risks. It compresses multi-dimensional synchronization risks into a single 0-to-1 confidence representation parameter, simplifying the input dimensions for remote control permission judgment. Simultaneously, it covers the three main types of risks—time-delay drift, model residuals, and action exceeding limits—and can operate stably for a long time without complex weighting parameter tuning.

[0130] In one alternative implementation, the switch target decision module is configured to perform:

[0131] Step S501: Project the health indication data onto the branch object or equipment object in the digital twin model of the power plant. The health indication data consists of branch isolation indication data and standby input indication data.

[0132] Branch isolation pointer data refers to the pointer information in the health pointer data that marks a faulty branch that needs to be taken out of operation;

[0133] Backup power input data refers to the health input data, which indicates that a certain backup branch needs to be put into operation to restore power supply;

[0134] Step S502: When the branch isolation pointing data points to a certain branch and the boundary switch or isolation switch of the branch is a remotely controllable switch, the candidate target value of the remotely controllable switch is written as the disconnect value.

[0135] Candidate target values ​​refer to the initial open or closed status values ​​assigned to the corresponding switches based on health indication data;

[0136] Step S503: When the backup power supply pointing data points to a certain backup power supply branch and the power supply switch of the backup power supply branch is a remotely controlled switch, the candidate target value of the remotely controlled switch is written as the closing value.

[0137] Step S504: Register the candidate target values ​​using the same sampling sequence number based on the proposed action limit and the remote control synchronization confidence margin;

[0138] Step S505: For remotely controlled switches not pointed to by health indicator data, the switch target value is written as the actual measured value of the switch. When the same remotely controlled switch has both closing and opening targets, its switch target value is written as the actual measured value of the switch.

[0139] The target value of the switch refers to the target state value of the corresponding switch that is finally determined after conflict checking and safety pre-checking.

[0140] Step S506: Output a unique switch target value for each remotely controlled switch.

[0141] Furthermore, when generating candidate target values, the branch isolation pointing data is mapped to the two boundary switches of the corresponding branch, and the candidate target values ​​of both switches are set to the open state. The standby connection pointing data is mapped to the incoming switch of the corresponding standby branch, and the candidate target value of the switch is set to the closed state. For switches not covered by any pointing data, the candidate target value is consistent with the measured value of the current switch.

[0142] Furthermore, during the registration of candidate target values, if the proposed action exceeds the limit by more than 0.1 or the remote control synchronization confidence margin is less than 0.8, the candidate target value will not be registered for the time being. The switch target value will remain consistent with the current measured value, and the judgment will be carried out again after the data of the subsequent sampling period is updated.

[0143] Furthermore, within a certain sampling period, if a sectionalizing switch simultaneously receives a tripping instruction from a cable joint over-temperature alarm and a closing instruction from the automatic backup power supply logic, and the two instructions from different sources act on the same switch at the same time, it is determined to be a target state conflict. No candidate target values ​​are registered, the target value of the switch is written as the currently measured closing state, and a conflict anomaly flag is marked. The decision is made after manual confirmation or data update in the next sampling period.

[0144] Understandably, candidate target values ​​are generated for corresponding switches based on health indication data. During the registration process, a safety pre-verification is performed by combining the proposed action limit exceedance and the remote control synchronization reliability margin. When the same switch has two conflicting candidate targets, opening and closing, the current measured state is maintained, and no state change is performed, ensuring that each switch has only one target state within the same sampling period. This method ensures that each remotely controlled switch has only one target state within the same sampling period, avoiding erroneous opening and closing of switches caused by multi-source control indication conflicts. At the same time, the pre-verification filters out action commands with high limit exceedance risk and low reliability, improving the safety of target state generation.

[0145] In one alternative implementation, the remote license generation module is configured to perform:

[0146] Step S601: Calculate the maximum time delay of the impact within the set of affected nodes;

[0147] The maximum time delay refers to the maximum value of the acquisition time delay of each electrical node measurement point within the set of affected nodes of the target switch, which serves as the timing judgment boundary for the switch control.

[0148] Specifically, the formula for calculating the maximum time delay is as follows:

[0149]

[0150] In the formula, For the first The remote-controlled switch is in the first The sampling sequence number has the greatest impact on the time delay. For the first A set of nodes affected by a remotely controlled switch. For the first The electrical node at the first The acquisition delay of each sampling sequence number.

[0151] Step S602: Generate a remote control action permission word based on the remote control synchronization confidence margin, confidence margin lower limit, maximum impact delay, maximum allowable delay, switch target value, and actual switch measurement value;

[0152] The lower confidence margin refers to the minimum remote control synchronization confidence margin value that allows remote control actions to be performed. It is set to 0.8. At this value, the remote control malfunction rate is at a low level, which can meet the reliability requirements of unattended power stations.

[0153] The maximum allowable time delay refers to the maximum data acquisition time delay value allowed to be executed for remote control actions. The value is 100 milliseconds. When the time delay exceeds this value, the data lag is too large, making it difficult to meet the accuracy requirements of real-time control.

[0154] The remote control action permission word is a binary flag that indicates whether the corresponding switch is allowed to output action commands. A value of 1 means that the output of action commands is allowed, and a value of 0 means that all action commands are blocked.

[0155] Specifically, the formula for calculating the remote control action license word is as follows:

[0156]

[0157] In the formula, For the first The remote-controlled switch is in the first Remote control action permission word for each sampling sequence number, When it is an indicator function and the condition within the curly braces is true, take If not valid, take , For the first The remote-controlled switch is in the first Remote synchronization reliability margin for each sampling sequence number As the lower bound of the confidence margin, For the first The remote-controlled switch is in the first The sampling sequence number has the greatest impact on the time delay. To allow the maximum time delay, For the first The remote-controlled switch is in the first The target value of the switch for each sampling sequence number. For the first The remote-controlled switch is in the first The measured value of the switch for each sampling sequence number.

[0158] Step S603: Calculate the estimated execution time based on the instruction generation time, the maximum impact delay, and the action response time;

[0159] The instruction generation time refers to the precise time marked by the local clock of the edge control node when the current round remote control instruction frame completes all frame assembly logic and is ready to output;

[0160] Action response time refers to the total mechanical action time from when the switch receives a valid action command to when it completes the opening and closing of the contacts and the displacement of the auxiliary contacts. It is calibrated by combining the switch's factory test report with actual transmission test data on site.

[0161] The estimated execution time refers to a conservative control reference time obtained by incorporating data timeliness compensation and mechanical response time. It is used for the timing coordination of actions of field terminals and is not used as a precise calculation value for the actual transmission landing time.

[0162] Specifically, the formula for calculating the expected execution time is as follows:

[0163]

[0164] In the formula, For the first The remote-controlled switch is in the first The expected execution time for each sampling sequence number. For the first The instruction generation time for each sampling sequence number, For the first The remote-controlled switch is in the first The sampling sequence number has the greatest impact on the time delay. For the first The response time of a remotely controlled switch.

[0165] Step S604: Output the remote control action license word and the expected execution time.

[0166] Furthermore, when generating the remote control action license word, three conditions must be met simultaneously: the remote control synchronization confidence margin is not lower than the lower confidence margin limit, the maximum time delay of the impact does not exceed the maximum allowable time delay, and the switch target value is inconsistent with the measured value. If any one of the conditions is not met, the license word value is 0. If all three conditions are met, the license word value is 1. There is no situation where action can be taken if only some conditions are met.

[0167] Furthermore, the calculated reliability margin of a certain switch is 0.85, which meets the requirement of a reliability margin lower limit of not less than 0.8. The maximum acquisition delay of each electrical node measurement point in the affected node set is 42 milliseconds, which meets the requirement of a maximum allowable delay of not more than 100 milliseconds. The current measured value of the switch is in the closed state, and the target value of the switch is in the open state. All three conditions are met at the same time. Therefore, the remote control action permission word of the switch is set to 1, which allows the output of the open action command.

[0168] Understandably, the maximum acquisition delay of each electrical node measurement point within the set of switch-affected nodes is used as the timing judgment boundary for the entire switch control, rather than using the average delay across the entire station. This is combined with a confidence margin threshold and switch state difference conditions to generate a remote control action permission word. An action command is only allowed to be output when all three conditions are simultaneously met; otherwise, a blocking command is output. This method constrains remote control action output from three dimensions: synchronization confidence, delay controllability, and action necessity. Action is only allowed when all safety conditions are met, reducing the risk of erroneous remote control based on lagging states or low-confidence data.

[0169] In one alternative implementation, the instruction frame generation module is configured to execute:

[0170] Step S701: Generate switch instruction records in ascending order of switch addresses, wherein the switch addresses are consistent with the addresses of the field switch control terminals;

[0171] The switch address refers to the unique communication address of the field control terminal corresponding to each remotely controlled switch, which is obtained through parameter calibration in the power station communication point table.

[0172] A switch instruction record refers to a single instruction entry consisting of the address, target status, permission word code, and expected execution time information corresponding to a single switch.

[0173] Step S702: Write the switch address, switch target value, remote control action permission word and expected execution time into the switch command record;

[0174] Step S703: When the remote control action permission word is an enabled value, write the action enable code and the switch target value to the switch command record; when the remote control action permission word is a locked value, write the locked code to the switch command record.

[0175] The action enable code is the allowed action code written into the instruction entry when the remote control action permission word is 1, and it adopts the action execution code specified in the power station remote control protocol.

[0176] The interlock code is a prohibited action code written into the instruction entry when the remote control action permission word is 0. It adopts the no-operation code specified in the power station remote control protocol.

[0177] Step S704: Calculate the cyclic redundancy check code for each switch instruction record and append it to the end of the switch instruction record;

[0178] Cyclic Redundancy Check (CRC) is a 16-bit check code appended to the end of each instruction record to verify whether bit errors have occurred during data transmission.

[0179] The switch address length, target value encoding, license word encoding, expected execution time encoding, byte order, frame header and trailer, CRC polynomial, initial value, XOR value of the result, input / output inversion method, and verification coverage are all read from the communication protocol configuration table corresponding to the field control terminal to ensure that the edge node and the field terminal use the same configuration. In this embodiment, CRC-16 / Modbus parameters are used, with a polynomial of 0x8005, an initial value of 0xFFFF, input / output inversion, and an XOR value of 0x0000. Other CRC-16 configurations that match the field terminal can also be used.

[0180] Step S705: Pack all switch command records into a remote control command frame and output it.

[0181] A remote control command frame refers to a standard message that can be sent through a communication link after all the switch command records are arranged in ascending order of address, and a standard frame header, frame trailer and overall check code are added.

[0182] Furthermore, when assembling remote control command frames, the command records of all switches are arranged in ascending order of switch address. Each record is written with the switch address, target status value, corresponding code of permission word, and expected execution time. Finally, a 16-bit cyclic redundancy check code is calculated and added to each record. The standard frame header and trailer are added together with the overall check field to form a complete sendable message.

[0183] Furthermore, in a certain round of sampling, there are 12 remotely controllable switches. After sorting by action priority and address, only one switch has a remote control action permission word of 1, while the remote control action permission words of the remaining 11 switches are 0. When framing, all instruction records are arranged in order of address from 1 to 12. The switch with permission word 1 is written with the action enable code and the corresponding opening and closing target value, and the 11 switches with permission word 0 are written with the blocking code. All records are supplemented with a check code to form a complete remote control instruction frame, which is sent to the field control terminal through a dedicated communication link. After the instruction is sent, the switch status in the twin model remains unchanged until the next round of sampling receives the actual change data of the switch, and then the status of the corresponding switch in the twin model is updated.

[0184] Understandably, an action enable code or interlock code is written to each switch according to the remote control action permission word. When the permission word is 0, no action pulse is output; only a no-operation message is sent. After the command is output, the switch state in the digital twin model is not updated immediately. The twin model is only updated based on the measured switch values ​​reported in the next round of sampling, avoiding inconsistencies between the virtual and real states caused by premature changes in the virtual state. This method ensures the timing consistency between remote control commands and the physical power station state, avoids the disconnect between the virtual and actual states caused by premature updates of the switch states in the twin model, reduces erroneous action commands generated based on incorrect virtual states within continuous sampling periods, and ensures the control safety of the system during long-term continuous operation.

[0185] It should be noted that this system can be deployed in the edge control nodes of the power plant site. If the acquisition terminals, aggregation nodes, and related network equipment already have PTP functionality, no additional hardware is required. If existing equipment does not support PTP, on-site clock synchronization can be achieved by upgrading the clock synchronization function or configuring PTP-enabled network equipment. It can operate based on existing acquisition devices, communication networks, and the digital twin system. During deployment, the mapping and calibration of equipment asset codes and twin objects are first completed. After configuring various threshold parameters and communication protocol parameters, it can be put into operation. The system continuously acquires on-site operating data at 100-millisecond intervals, completes time-scale alignment, twin simulation, confidence margin calculation, and permission judgment, and then generates remote control command frames, which are sent to the on-site switch control terminals through the existing power plant communication network.

[0186] For example, when a mountain photovoltaic power station experiences rapid cloud shadow movement at midday, its photovoltaic output fluctuates significantly within seconds, and the data acquisition time lag at some measurement points increases. The system calculates that the remote control synchronization reliability margin of the corresponding switch is lower than the set threshold, and automatically blocks the automatic remote control command of the corresponding switch. Remote control access is only restored after the data time lag recovers, the output fluctuations stabilize, and the reliability margin and time lag requirements are met. When an over-temperature anomaly occurs at a cable joint of a power collection line, the system, provided that the data synchronization requirements are met, firstly generates the opening commands of the boundary switches on both sides of the faulty branch in address order. After each switch is opened, it waits for the actual change and re-simulation verification. After the fault is completely isolated, the system completes the over-limit verification based on the new topology and generates the closing command of the standby switch in the next sampling cycle to complete the fault isolation and power supply restoration. After the command is issued, the system waits for the actual measured status of the on-site switches to return before updating the twin model status to avoid inconsistencies between the virtual and real states.

[0187] like Figure 2 As shown, in this embodiment, the external power grid is connected to the substation bus via the main transformer. The bus connects to the photovoltaic grid-connected branch, the energy storage grid-connected branch, the station load branch, and the backup power supply branch. Each branch is equipped with a remotely controllable switch or a switch that can be activated. Voltage, current, power, and switch status are acquired by the corresponding acquisition terminals and sent to the edge control node via the acquisition network. The edge control node is configured with a digital twin model and communication and control services to synchronize equipment objects and electrical topology based on field measurement data and to generate control results for the target switch. The control results are sent to the field control terminal via the control link, and then the field control terminal drives the corresponding switch to operate. After the switch operation is completed, the digital twin model is updated with the measured displacement status obtained from subsequent sampling, thereby maintaining the consistency of the state between the field substation and the digital twin model.

[0188] like Figure 3 As shown in the figure, this diagram illustrates the distribution of 30 independent switching action cases across two dimensions: maximum impact time delay and remote control synchronization reliability margin. The shape of the dots distinguishes between fault isolation and backup activation actions, the size of the dots represents the degree of limit violation in the proposed action, and the color of the dots indicates the final permitted result. Cases with a reliability margin above 0.8 and a time delay not exceeding 100 milliseconds typically meet the action conditions, but still require pre-verification of the proposed action's limit violation. For example, although cases 22 to 24 meet the requirements for data synchronization and time delay, the limit violation exceeds 0.1, therefore they are still blocked. Cases 17 to 21 and case 30 have high reliability margins, but the time delay exceeds 100 milliseconds, so they also cannot execute the action. This diagram shows that remote control permission is not determined by a single indicator, but is constrained by synchronization reliability, data timeliness, post-action operational safety, and state change requirements, thus avoiding risks such as high reliability but expired data, or timely data but limit violation after action. Digital twins utilize power grid models and real-time measurements to conduct scenario simulations, providing a technical foundation for this pre-action safety assessment.

[0189] Although embodiments of this application have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of this application, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A digital twin intelligent control system for power plants in unattended scenarios, characterized in that: include: The sampling time-stamp alignment module encapsulates the electrical operating quantities, actual switch values, measurement timestamps, and arrival timestamps accessed by the power station acquisition terminal into sampling windows and outputs a time-stamp aligned sampling set. The twin flip-over simulation module synchronizes the equipment objects, topology connections, and node injection boundaries of the digital twin model of the power station based on the time-scale aligned sampling set. It performs single-switch flip-over simulation for each remotely controlled switch and outputs an electrical mirror dataset. The time delay drift residual calculation module calculates the acquisition time delay, time delay power drift, and twin electrical residual within the set of affected nodes and the set of affected branches for each remotely controlled switch based on the time-scale aligned sampling set and the electrical mirror dataset, and outputs the acquisition time delay, time delay power drift, and twin electrical residual. The reliability margin calculation module calculates the proposed action over-limit based on the proposed action electrical quantity and allowable limit, and couples the proposed action over-limit with the time-delay power drift and twin electrical residual to output the remote control synchronization reliability margin. The switch target decision module forms a unique switch target value based on health indication data, the limit of the proposed action, the remote control synchronization confidence margin, and the actual switch measurement value. The remote control license generation module generates a remote control action license word and the expected execution time based on the switch target value, the actual switch value, the remote control synchronization confidence margin, and the acquisition delay in the set of affected nodes. The instruction frame generation module generates a remote control instruction frame with a cyclic redundancy check code according to the switch address, switch target value, remote control action permission word, and expected execution time.

2. The digital twin intelligent control system for power plants in unattended scenarios according to claim 1, characterized in that, The sampling timescale alignment module is configured to execute: Collect electrical node voltage, branch current, net active power injection at electrical nodes, measured switch values, measurement timestamp, and arrival timestamp; Time-stamped merging of collected data is performed according to sampling sequence number, sampling window, and sampling interval; The sampling set is encapsulated and aligned with the device's unique address, measurement point type, and sampling sequence number.

3. The digital twin intelligent control system for power plants in unattended scenarios according to claim 1, characterized in that, The twin flip simulation module is configured to execute: Based on the unique address of the equipment, the time-stamped aligned sample set is mapped to the equipment object in the digital twin model of the power plant; The current topology is established based on the measured values ​​of the switches, and the twin node voltage and twin branch current are obtained by using the net active power injection of the electrical nodes as the node injection boundary. Perform a single-switch flipping simulation for each remotely controlled switch to determine the voltage of the node to be activated and the current of the branch to be activated. Based on the topological connectivity components and power flow change branches, generate the set of affected nodes, the set of affected branches, and health pointing data; output the electrical mirror dataset.

4. The digital twin intelligent control system for power plants in unattended scenarios according to claim 1, characterized in that, The time-delay drift residual calculation module is configured to perform: The data acquisition delay is calculated based on the measurement timestamp and the arrival timestamp. The time-delay power drift is calculated based on the net active power injection, baseline active power, power stability term, and sampling interval of the electrical nodes in the current and previous samples within the affected node set. The twin electrical residual is calculated based on the branch currents and twin branch currents within the affected branch set, as well as the electrical node voltages and twin node voltages within the affected node set.

5. The digital twin intelligent control system for power plants in unattended scenarios according to claim 1, characterized in that, The confidence margin calculation module is configured to perform: Perform over-limit comparisons on the proposed operating branch current, proposed operating node voltage, upper limit of allowable current, upper limit of allowable voltage, and lower limit of allowable voltage to obtain the proposed operating over-limit amount; The remote control synchronization reliability margin is obtained by summing and normalizing the time-delay power drift, twin electrical residual, and simulated action over-limit.

6. The digital twin intelligent control system for power plants in unattended scenarios according to claim 1, characterized in that, The switch target decision module is configured to execute: Project health-related data onto branch or equipment objects in the digital twin model of the power plant; Based on the projection results, obtain candidate target values ​​for boundary switches, disconnect switches, or switching switches; Based on the proposed action exceeding the limit and the remote control synchronization confidence margin, candidate target values ​​are registered using the same sampling sequence number; When the same remotely controlled switch has both a closing target and a disconnection target, the switch target value is written as the actual measured value of the switch. Output a unique target value for each remotely controlled switch.

7. The digital twin intelligent control system for power plants in unattended scenarios according to claim 1, characterized in that, The remote control license generation module is configured to execute: Calculate the maximum time delay within the set of affected nodes; The remote control action permission word is generated based on the remote control synchronization confidence margin, confidence margin lower limit, maximum impact delay, maximum allowable delay, switch target value, and actual switch measurement value. Calculate the estimated execution time based on the instruction generation time, the maximum impact delay, and the action response time; output the remote control action license word and the estimated execution time.

8. The digital twin intelligent control system for power plants in unattended scenarios according to claim 1, characterized in that, The instruction frame generation module is configured to execute: Generate switch instruction records in ascending order of switch address; Write the switch address, switch target value, remote control action permission word, and expected execution time into the switch instruction record; Write the action enable code or lockout code to the switch command record according to the remote control action permission word; Calculate the cyclic redundancy check code for the switch command record; Pack the switch command records into a remote control command frame and output it.

Images