Online reconstruction control method and system for secondary equipment of smart substation

Abstract

The application discloses a kind of intelligent substation secondary equipment online reconstruction control method and system, it is related to electric power system automation technical field, including the following steps: the fault information containing digital fingerprint generated by fault secondary equipment is broadcast to adjacent equipment;Make adjacent equipment verify digital fingerprint and carry out lightweight consensus confirmation, and obtain secondary equipment dynamic credit rating based on secondary equipment operation data;According to dynamic credit rating, candidate takeover equipment set is screened from adjacent equipment;Target takeover equipment is determined from candidate takeover equipment set by multi-objective optimization algorithm;Make target takeover equipment execute reconstruction logic.The present application is used to solve the problems of insufficient authenticity of existing secondary equipment fault information, low efficiency of equipment cooperation, lack of intelligence in takeover decision and poor connection in reconstruction process.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power system automation, and more particularly, to a secondary equipment online reconstruction control method and system for a smart substation. BACKGROUND

[0002] With the rapid development of smart grid and digital substation construction, as the core link for realizing protection, monitoring and control in the power system, the operation reliability and intelligent level of secondary equipment are directly related to the safety and stability of the power grid. In recent years, although online monitoring, remote communication and data analysis technologies have been widely used to support the operation and reconstruction of secondary equipment, there are still significant technical bottlenecks in the authenticity guarantee, collaborative consistency confirmation and dynamic takeover decision of fault information, which restricts the further improvement of the intelligent level of the power grid.

[0003] Although the existing secondary equipment reconstruction method for a smart substation can realize fault detection and equipment takeover to a certain extent, it generally has the following problems: the fault information lacks reliable unique identification and tamper-proof mechanism, and the authenticity and traceability are insufficient; the consensus mechanism between adjacent devices is inefficient, relying on centralized or manual confirmation methods, which is difficult to meet the real-time requirements under complex topologies; the selection mechanism of candidate takeover devices relies too much on static thresholds or single indicators, and fails to realize the comprehensive trade-off and dynamic adaptation of multiple factors; the selection and verification of target takeover devices lack systematic optimization and forward-looking inspection, which may lead to deviation of the takeover scheme from the optimal solution, and even cause new risks; the reconstruction process lacks continuity in state loading and logic switching, often resulting in incomplete data or control link interruption, affecting the continuity and stability of power grid operation.

[0004] In view of the above problems, the present application provides an improved scheme. SUMMARY

[0005] In order to overcome the above-mentioned defects of the prior art, the embodiments of the present application provide a secondary equipment online reconstruction control method and system for a smart substation, which introduces digital fingerprints and lightweight consensus mechanisms to guarantee the authenticity and consistency of fault information, and combines multi-objective optimization algorithms, deep reinforcement learning and digital twin verification to dynamically select and verify target takeover devices, finally loads the running state of fault devices to realize seamless function switching, solving the problems of insufficient authenticity of existing secondary equipment fault information, low device collaboration efficiency, lack of intelligence in takeover decision and poor continuity in reconstruction process.

[0006] To achieve the above purpose, the present application provides the following technical scheme:

[0007] The application discloses an online reconstruction control method for secondary equipment of a smart substation, and comprises the following steps: broadcasting fault information containing a digital fingerprint generated by a faulty secondary equipment to adjacent equipment; enabling the adjacent equipment to verify the digital fingerprint and perform lightweight consensus confirmation, and obtaining a dynamic reputation score of the secondary equipment based on secondary equipment operation data; screening a candidate takeover equipment set from the adjacent equipment according to the dynamic reputation score; determining a target takeover equipment from the candidate takeover equipment set through a multi-objective optimization algorithm; and enabling the target takeover equipment to execute reconstruction logic.

[0008] In a preferred embodiment, the broadcasting of the fault information containing the digital fingerprint generated by the faulty secondary equipment to the adjacent equipment specifically comprises: collecting a unique identifier and a real-time fault code of the faulty secondary equipment; combining the unique identifier, the fault code and a current timestamp, and generating the digital fingerprint by using a preset cryptographic hash function; combining the fault code and the digital fingerprint, generating the fault information and broadcasting the fault information.

[0009] In a preferred embodiment, the lightweight consensus confirmation specifically comprises: after receiving the fault information, the adjacent equipment acts as a proposal node; encapsulating the verification result of the digital fingerprint and the fault information into a consensus request, and broadcasting the consensus request to the adjacent equipment; the rest of the adjacent equipment responds to the consensus request, independently verifies the digital fingerprint and feeds back the verification result; when the number of valid verification results received by the proposal node exceeds a first preset threshold, the proposal node determines that the consensus is reached and generates a consensus confirmation signal.

[0010] In a preferred embodiment, the obtaining of the dynamic reputation score of the secondary equipment based on secondary equipment operation data specifically comprises: obtaining historical and real-time operation state data of the secondary equipment, and performing data preprocessing to obtain first data; performing time series weighting on the first data based on an adaptively adjusted sliding time window and a forgetting factor to obtain second data; obtaining statistical features of the second data, and combining device multi-modal features to construct a multi-modal reputation feature vector; inputting the feature vector into a scoring model based on an attention mechanism, calculating and outputting the dynamic reputation score, and the attention mechanism is used for adaptively assigning weights to different features in the multi-modal reputation feature vector.

[0011] In a preferred embodiment, the screening of the candidate takeover equipment set from the adjacent equipment according to the dynamic reputation score specifically comprises: adding equipment with a reputation score higher than a dynamic threshold to an initial candidate takeover equipment set; performing secondary screening on the initial candidate takeover equipment set according to a topological position and a load balancing factor, and outputting the candidate takeover equipment set.

[0012] In a preferred embodiment, the determining of the target takeover device from the candidate takeover device set by the multi-objective optimization algorithm specifically comprises: constructing a multi-objective optimization function, and dynamically generating multi-objective weight coefficients matched with the current power grid state based on the deep reinforcement learning model; inputting the candidate takeover device set as a solution space into a non-dominated sorting algorithm, and taking the multi-objective optimization function and its weight coefficients as evaluation basis, and obtaining a Pareto optimal solution set through iterative solution; sorting the Pareto optimal solution set, and selecting the candidate device corresponding to the highest ranked solution as the preliminary target takeover device; and performing prospective verification on the preliminary target takeover device based on digital twinning, and finally determining the target takeover device after verification.

[0013] In a preferred embodiment, the deep reinforcement learning model comprises: constructing a reinforcement learning model based on an Actor-Critic framework; taking real-time power grid state information as the input state of the model; and mapping the output action of the model as the weight coefficient of the multi-objective optimization function; wherein the model is trained by a reward function based on historical takeover success rate and system stability.

[0014] In a preferred embodiment, the sorting of the Pareto optimal solution set specifically comprises: taking the multi-objective weight coefficients as the evaluation criterion weight of the TOPSIS method; determining the positive ideal value and the negative ideal value of each optimization objective, and calculating the weighted Euclidean distance between each candidate solution in the Pareto optimal solution set and the positive ideal value and the negative ideal value in each optimization objective dimension; calculating the relative closeness of each candidate solution based on the weighted Euclidean distance, and sorting according to the size of the relative closeness to obtain a sorting result.

[0015] In a preferred embodiment, the prospective verification of the preliminary target takeover device based on digital twinning specifically comprises: inputting the topology parameters and operating state data of the target takeover device into a power grid digital twinning simulation platform; importing future load prediction data and new energy output prediction data in the simulation platform to construct a takeover simulation scenario; performing power grid dynamic load flow calculation in the takeover simulation scenario to obtain node voltage, line power and power balance data; and generating a verification conclusion based on the results of the power grid dynamic load flow calculation.

[0016] In a second aspect, the application provides an online reconstruction control system for secondary equipment of a smart substation, comprising: a fingerprint broadcast module configured to broadcast fault information generated by a faulty secondary equipment to neighboring equipment, the fault information containing a digital fingerprint; a verification consensus and scoring module configured to cause the neighboring equipment to verify the digital fingerprint and perform lightweight consensus confirmation, and obtain a dynamic reputation score of the secondary equipment based on operation data of the secondary equipment; a candidate screening module configured to screen a candidate takeover equipment set from the neighboring equipment according to the dynamic reputation score; an optimization selection module configured to determine a target takeover equipment from the candidate takeover equipment set by using a multi-objective optimization algorithm; and a reconstruction execution module configured to cause the target takeover equipment to execute reconstruction logic.

[0017] As can be seen from the above technical solutions, the application verifies the fault information by using a digital fingerprint and a lightweight consensus mechanism, thereby solving the problems of difficulty in guaranteeing the authenticity of fault information and low verification efficiency in the prior art; and by combining a multi-objective optimization algorithm, deep reinforcement learning weight self-adaptation, and digital twin forward-looking verification, the application solves the problems of lack of intelligence in candidate takeover equipment screening and insufficient reliability of takeover schemes. BRIEF DESCRIPTION OF DRAWINGS

[0018] Figure 1 FIG. 1 is a flowchart of a method for online reconstruction control of secondary equipment of a smart substation according to the application.

[0019] Figure 2 FIG. 2 is a structural diagram of an online reconstruction control system for secondary equipment of a smart substation according to the application.

[0020] Figure 3 FIG. 3 is a verification flowchart of a target takeover equipment according to the application. DETAILED DESCRIPTION

[0021] The technical solutions in the embodiments of the application will be described clearly and completely below with reference to the accompanying drawings in the embodiments of the application. Obviously, the described embodiments are only a part of the embodiments of the application, rather than all the embodiments of the application. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of the application.

[0022] Embodiment 1, Figure 1 A method for online reconstruction control of secondary equipment of a smart substation according to the application is given, comprising the following steps:

[0023] S1, broadcast fault information generated by a faulty secondary equipment to neighboring equipment, the fault information containing a digital fingerprint.

[0024] In this embodiment, the broadcasting of the fault information generated by the faulty secondary equipment to the neighboring equipment specifically comprises:

[0025] Collect the unique identifier of the faulty secondary equipment and the real-time fault code;

[0026] Combine the unique identifier, the fault code and the current timestamp, and generate a digital fingerprint using a preset cryptographic hash function;

[0027] Combine the fault code with the digital fingerprint, generate fault information and broadcast it.

[0028] Specifically:

[0029] When the secondary equipment fails, first collect the unique identifier of the equipment and the real-time fault code , and synchronously obtain the current standard timestamp ;

[0030] Combine the above three elements and input them into a preset cryptographic hash function to calculate the digital fingerprint :

[0031]

[0032] Among them:

[0033] represents the string or data splicing operation;

[0034] is a cryptographic hash function, such as SHA-256, used to map the input element combination to a fixed-length digest value;

[0035] is the digital fingerprint, which is used to uniquely identify the fault event.

[0036] By introducing the triplets of equipment identifier, fault code and timestamp, and calculating through the hash function, the digital fingerprint has uniqueness, tamper resistance and time sensitivity, so as to guarantee the authenticity and traceability of the fault information.

[0037] After obtaining the digital fingerprint , combine it with the fault code to generate a complete fault information packet :

[0038]

[0039] Finally, the faulty secondary equipment broadcasts the fault information to the adjacent equipment through the communication link.

[0040] S2, let the adjacent equipment verify the digital fingerprint and perform lightweight consensus confirmation, and obtain the dynamic reputation score of the secondary equipment based on the secondary equipment operation data.

[0041] The neighboring device reacquires the unique identifier, fault code and timestamp in the message after receiving the fault information message and recalculates the digital fingerprint using the hash function used when generating the digital fingerprint:

[0042]

[0043] If , it is determined that the fault information has not been tampered with and the verification is passed; otherwise, the message is discarded.

[0044] After verifying the digital fingerprint, a lightweight consensus confirmation is performed:

[0045] In this embodiment, the lightweight consensus confirmation is specifically:

[0046] The neighboring device receives the fault information and acts as a proposal node;

[0047] The verification result of the digital fingerprint and the fault information are encapsulated as a consensus request and broadcast to the neighboring devices;

[0048] The remaining neighboring devices respond to the consensus request, independently verify the digital fingerprint, and feed back the verification result;

[0049] When the proposal node counts that more than a first preset threshold of valid verification results are received, it determines that the consensus is reached and generates a consensus confirmation signal.

[0050] Specifically:

[0051] When a neighboring device receives the fault information and completes local verification of the digital fingerprint, the device acts as a proposal node, encapsulates the verification result together with the fault information as a consensus request message, and broadcasts it to the remaining neighboring devices;

[0052] The proposal device is the device that first receives the fault information and completes the digital fingerprint verification in the set of neighboring devices, and it undertakes the responsibility of broadcasting the consensus request. In the consensus process of the same fault event, the system usually designates a single proposal device.

[0053] If multiple devices almost simultaneously meet the proposal conditions, a unique proposal device is determined according to a preset rule to avoid repeated broadcasting, such as size comparison of device unique identifiers or time stamp sequence.

[0054] Each neighboring device independently performs the same hash function verification process after receiving the consensus request and obtains its own verification result , wherein:

[0055] indicates that the verification is passed;​​

[0056] indicates that the verification fails.

[0057] Each device then feeds back its verification result to the proposer node;

[0058] After receiving and collecting the verification results of all adjacent devices, the proposer node calculates the total number of valid verifications:

[0059]

[0060] wherein, indicates the set of adjacent devices, indicates the total number of devices.

[0061] The proposer node compares the number of valid verifications with a first preset threshold When the number of valid verifications is greater than the first preset threshold , it is determined that consensus is reached, and a consensus confirmation signal is generated and broadcast to all adjacent devices.

[0062] After completing the lightweight consensus confirmation, the device dynamic reputation score of each candidate takeover device is calculated based on the operation data of the secondary device;

[0063] In this embodiment, the secondary device dynamic reputation score is obtained based on the operation data of the secondary device, specifically:

[0064] The historical and real-time operation state data of the secondary device are obtained and preprocessed to obtain first data;

[0065] The first data is time-weighted based on an adaptively adjusted sliding time window and a forgetting factor to obtain second data;

[0066] The statistical features of the second data are obtained, and a multi-modal reputation feature vector is constructed in combination with the multi-modal features of the device;

[0067] The feature vector is input into a scoring model based on an attention mechanism to calculate and output a dynamic reputation score, wherein the attention mechanism is used to adaptively assign weights to different features in the multi-modal reputation feature vector.

[0068] Specifically:

[0069] First, the historical operation state data and the current operation state data of the secondary device are collected within a preset time period;

[0070] The data includes electrical parameter data, communication link state data, and operation and maintenance log information. The historical operation state data can be called by a monitoring center database, and the current operation state data is collected by an online monitoring device in real time;

[0071] Timestamp alignment and missing value filling are performed on the historical operating state data and the current operating state data to obtain first data, and the timestamp alignment is used to map different modal data to a unified time axis to ensure that the data has a corresponding relationship at the same time;

[0072] The missing value filling can be achieved by interpolation method, previous value retention or prediction model-based method, thereby ensuring the integrity of the time series data;

[0073] Then, the first data is time series weighted based on a sliding time window and a forgetting factor of operating condition adaptive adjustment to obtain second data;

[0074] The weighting calculation formula is:

[0075]

[0076] Among them:

[0077] represents the time The second data obtained by weighting calculation;

[0078] represents the first data at time index ;

[0079] represents the forgetting factor, The smaller, the faster the historical data decays;

[0080] represents a sliding window centered on the current time ;

[0081] represents the length of the window, which is adaptively adjusted according to the operating condition, and when the operating condition is steady, it is , and when the operating condition is disturbed or faulty, it is ;

[0082] represents the time index in the sliding window, i.e. the value point at a certain time in the window; in the weighting process, all in the window are traversed to calculate their corresponding weighted contributions.

[0083] Statistical features are extracted from the second data, and are spliced with a multi-modal feature vector of the device to obtain a multi-modal reputation feature vector, the statistical features including a trend slope, a fluctuation amplitude and a stability index, and the multi-modal feature vector including a device type, an environmental factor, a communication state and the like;

[0084] Further, a reference information representing the global state is generated according to the current operating condition;

[0085] The reference information is compared with each component in the multi-modal credibility feature vector to determine the matching degree of different features with the current operating condition;

[0086] In the comparison process, the higher the correlation, the greater the weight of the multi-modal credibility feature vector, and the weight of the feature component with lower correlation is correspondingly reduced, thereby realizing dynamic sorting of feature importance;

[0087] After completing the weight distribution, all multi-modal credibility feature vectors are weighted and combined to obtain a score result that can comprehensively reflect the credibility status of the device at that moment.

[0088] The score result is the dynamic credibility score, which can be updated in real time with the change of the operating condition, so that the model has both adaptability and strong robustness.

[0089] S3, according to the dynamic credibility score, a candidate takeover device set is selected from the adjacent devices.

[0090] In this embodiment, the candidate takeover device set is selected from the adjacent devices according to the dynamic credibility score, specifically:

[0091] Devices with a credibility score higher than the dynamic threshold are added to the initial candidate takeover device set;

[0092] The initial candidate takeover device set is further screened according to the topological position and load balancing factor, and the candidate takeover device set is output.

[0093] Specifically:

[0094] First, the credibility score results of all adjacent devices are preliminarily screened, and a dynamic threshold is set, the dynamic threshold is based on the average value of all device credibility scores and the standard deviation of the credibility score, when the credibility score of the device is higher than the dynamic threshold, the corresponding device is added to the initial candidate takeover device set;

[0095] After obtaining the initial candidate takeover device set, the candidate set is further screened based on the topological position and load balancing factor;

[0096] The topological position factor is calculated as follows:

[0097]

[0098] Wherein, is the network hop number of the candidate device and the faulty device;

[0099] The load balancing factor calculation formula is as follows:

[0100]

[0101] Wherein, is the candidate device The current load proportion is;

[0102] The load balancing factor and the topological position factor are integrated into a secondary screening index :

[0103]

[0104] Wherein, is an adjustment coefficient, used to balance the importance of topology priority and load balancing;

[0105] Finally, the secondary screening index is higher than the secondary screening threshold The corresponding device joins the candidate takeover device set;

[0106] Output the candidate takeover device set.

[0107] S4, determine the target takeover device from the candidate takeover device set through a multi-objective optimization algorithm, specifically:

[0108] Construct a multi-objective optimization function, and dynamically generate multi-objective weight coefficients matched with the current power grid state based on a deep reinforcement learning model;

[0109] Input the candidate takeover device set as the solution space into the non-dominated sorting algorithm, and take the multi-objective optimization function and its weight coefficients as the evaluation basis, and after iterative solution, obtain the Pareto optimal solution set;

[0110] Sort the Pareto optimal solution set, and select the candidate device corresponding to the highest ranked solution as the preliminary target takeover device;

[0111] Perform prospective verification of the preliminary target takeover device based on digital twinning, and after verification, finally determine the target takeover device.

[0112] Specifically:

[0113] According to the power grid operation demand, define multiple optimization objectives for the candidate takeover device, and construct a multi-objective optimization function according to these optimization objectives ; The optimization objectives include minimum takeover delay, maximum operation reliability, optimal load balancing, and maximum communication timeliness:

[0114] Wherein, indicates the candidate device takeover delay, representing candidate devices operational reliability, representing candidate devices load balancing degree, representing candidate devices communication timeliness, , , , The weights respectively corresponding to the takeover delay, the operational reliability, the load balancing degree and the communication timeliness can be set according to actual needs of the power grid operation scene.

[0115] Under different working conditions, the importance of the optimization objectives is not the same. Therefore, the present application introduces a deep reinforcement learning model to dynamically output a set of multi-objective weight coefficients by perceiving the power grid operation state, so that the optimization objectives can be adaptively adjusted according to the power grid state.

[0116] The deep reinforcement learning model is constructed based on an Actor-Critic framework;

[0117] The input state of the deep reinforcement learning model includes real-time acquired fault attributes, power grid load level and network topology information;

[0118] The output action of the deep reinforcement learning model is used for dynamically generating the multi-objective weight coefficients;

[0119] The deep reinforcement learning model is trained by a reward function, and the value of the reward function is positively correlated with the success rate of historical takeover operations and system stability indicators.

[0120] After obtaining the multi-objective optimization function and the multi-objective weight coefficients, a set of candidate takeover devices is input into a non-dominated sorting algorithm as a solution space;

[0121] In the initial stage, each device in the set of candidate takeover devices is encoded and identified as an individual;

[0122] Based on the constructed multi-objective optimization function, the original evaluation values of the device on multiple objectives such as takeover delay, operational reliability, load balancing degree and communication timeliness are calculated;

[0123] Combined with the multi-objective weight coefficients, the original evaluation values are weighted or normalized to obtain a comprehensive evaluation vector that can reflect the preference of the current working condition;

[0124] Each candidate device and its corresponding comprehensive evaluation vector are taken as an individual to form an initial solution set of the algorithm.

[0125] In the iteration process, the algorithm compares the candidate devices based on the comprehensive evaluation vector and the non-dominated sorting principle, specifically:

[0126] If a device is not lower than another device in all weighted objectives and higher than the device in at least one objective, it is determined to be better in multi-objective performance. At this time, the worse device is marked as a disadvantageous solution and no longer enters the next round of iteration; and the better performing device is retained as a potential candidate solution to continue to participate in evolution;

[0127] Through such step-by-step comparison and screening, the solutions retained after each iteration can gradually approach the Pareto optimal solution set, which have relative advantages in multiple objective dimensions.

[0128] After multiple iterations, the Pareto optimal solution set is finally obtained. The devices in the solution set are compromise optimal solutions based on the multi-objective optimization function calculation and dynamic adjustment of the weight coefficient, representing a reasonable balance between different optimization objectives.

[0129] In the embodiment, the Pareto optimal solution set is sorted, and the specific process includes:

[0130] The multi-objective weight coefficient is used as the evaluation criterion weight of the TOPSIS method;

[0131] The positive ideal value and the negative ideal value of each optimization objective are determined, and the weighted Euclidean distance between each candidate solution in the Pareto optimal solution set and the positive ideal value and the negative ideal value in each optimization objective dimension is calculated;

[0132] The relative closeness of each candidate solution is calculated based on the weighted Euclidean distance, and the sorting result is obtained according to the size of the relative closeness.

[0133] Specifically:

[0134] The multi-objective weight coefficient dynamically generated by the deep reinforcement learning model is used as the weight input of the TOPSIS method, which can ensure that the sorting process is consistent with the current power grid operating condition, and the influence of different objectives in the sorting can be adjusted in real time according to the operating state;

[0135] The first candidate solution in the Pareto optimal solution set , whose evaluation value under the first optimization objective is denoted as . ;

[0136] When the dimension difference of different objectives is large, the objectives can be normalized first to obtain the normalized value ​If the target dimensions are consistent, the original evaluation values can also be used directly.

[0137] On this basis, the following definitions are given:

[0138] Positive ideal value represents the optimal performance under the i-th target, taking the maximum value for benefit-type targets and the minimum value for cost-type targets.

[0139] Positive ideal value represents the worst performance under the i-th target, taking the minimum value for benefit-type targets and the maximum value for cost-type targets.

[0140] Combined with the multi-target weight coefficient , the weighted Euclidean distance between the candidate solution and the ideal solution can be calculated:

[0141]

[0142]

[0143] wherein:

[0144] represents the weighted Euclidean distance between the candidate solution and the positive ideal solution;

[0145] represents the weighted Euclidean distance between the candidate solution and the negative ideal solution;

[0146] represents the weight coefficient of the i-th target. Through the above calculation, the relative difference between each candidate solution and the optimal state and the worst state can be quantified, providing a basis for subsequent closeness calculation and sorting.

[0147] After obtaining the distance from the ideal solution, the relative closeness of the candidate solution

[0148] is further calculated:

[0149]

[0150] wherein the value of ranges from 0 to 1, and the larger the value, the closer the candidate solution is to the ideal solution.

[0151] Finally, the candidate solutions in the Pareto optimal solution set are sorted according to the size of , and the candidate device corresponding to the highest ranked solution is selected as the preliminary target takeover device.

[0152] ​​​In the embodiment, the preliminary target takeover equipment is subjected to digital-twin-based prospective verification, specifically:

[0153] The topology parameters and operating state data of the target takeover equipment are input into a power grid digital twin simulation platform.

[0154] The future load prediction data and new energy output prediction data are imported into the simulation platform to construct a takeover simulation scenario.

[0155] The power grid dynamic load flow calculation is performed in the takeover simulation scenario to obtain node voltage, line power and power balance data.

[0156] The verification conclusion is generated based on the results of the power grid dynamic load flow calculation.

[0157] As shown in Figure 3 , the present application provides a target takeover equipment verification flowchart.

[0158] Specifically:

[0159] The topology parameters and operating state data of the preliminary target takeover equipment are input into a power grid digital twin simulation platform.

[0160] The topology parameters are used to describe the node connection relationship and line attributes of the equipment in the power grid, and the operating state data are used to reflect the real-time operating conditions such as voltage, current and power of the equipment, thereby forming a corresponding virtual model in the simulation platform.

[0161] The power grid digital twin simulation platform is used to construct a model corresponding to the actual power grid operating state in a virtual environment, perform dynamic load flow calculation by importing power grid topology parameters, equipment operating data and prediction data, and output operating constraint comparison results, thereby performing prospective verification on the target takeover equipment.

[0162] The load prediction data for the future period and the new energy output prediction data are imported into the simulation platform; the load prediction data are used to set the power demand curve of each node in the future period, and the new energy output prediction data are used to set the output curve of renewable energy sources such as wind power and photovoltaic power.

[0163] Based on the above data, the simulation platform introduces future operating disturbances into the original power grid digital twin model and embeds the takeover logic of the preliminary target takeover equipment into the simulation process, such as switching control at a specific time or replacing a faulty equipment.

[0164] In this way, the simulation platform can form a takeover simulation scenario containing equipment takeover actions and future load and output changes, so that the power grid operating state can dynamically reflect the possible future situations.

[0165] Under the takeover simulation scene, dynamic power flow calculation is performed to obtain voltage distribution of each node, power transmission of each line, and power balance result of the whole system;

[0166] After obtaining the result of dynamic power flow calculation, it needs to be compared with preset operation constraints to form a verification conclusion. The constraint conditions mainly include node voltage constraint, line power constraint, and system power balance constraint, which are as follows:

[0167] Node voltage constraint:

[0168]

[0169] Line power constraint:

[0170]

[0171] System power balance constraint:

[0172]

[0173] Wherein:

[0174] Vn represents the voltage amplitude of the nth node;

[0175] and respectively represent the lower limit and upper limit of the node voltage;

[0176] Pn represents the apparent power of the nth line;

[0177] Pmax represents the maximum allowable power of the line;

[0178] Pn represents the apparent power of the nth line;

[0179] Pn represents the apparent power of the nth line;

[0180] Pn represents the apparent power of the nth line;

[0181] If the above three conditions are met, a verification passed conclusion is generated; if any condition is not met, a verification failed conclusion is generated;

[0182] S5, let the target takeover device execute reconstruction logic. ​​

[0183] After determining the target takeover device, it is necessary to make it have the same operating conditions as the failed secondary device in order to complete the replacement of subsequent control functions;

[0184] First, the operating state data of the failed secondary device is loaded to the target takeover device, including real-time monitoring data, control instruction cache, and communication interface parameters of the device.

[0185] Through the loading process, the target takeover device can inherit the operating context of the failed device at the data level, thereby avoiding data loss or control logic interruption during the switching process;

[0186] Subsequently, the target takeover device executes the preset reconstruction logic based on the loaded operating state, which includes steps such as restoring protection and control strategies, rebuilding communication links with the upper layer scheduling system, and re-establishing collaborative control relationships with adjacent devices.

[0187] Through this process, the target takeover device can seamlessly replace the failed device in terms of function, quickly reconstruct the control link, and ensure the continuity and stability of power grid operation.

[0188] Embodiment 2, Figure 2 An online reconstruction control system for secondary devices in an intelligent substation is provided, which includes:

[0189] A fingerprint broadcast module is used to broadcast fault information containing digital fingerprints generated by the failed secondary device to adjacent devices.

[0190] A verification consensus and scoring module is used to verify the digital fingerprints and perform lightweight consensus confirmation by the adjacent devices, and to obtain a dynamic reputation score of the secondary device based on the secondary device operating data.

[0191] A candidate screening module is used to screen a set of candidate takeover devices from the adjacent devices according to the dynamic reputation score.

[0192] An optimization selection module is used to determine the target takeover device from the set of candidate takeover devices through a multi-objective optimization algorithm.

[0193] A reconstruction execution module is used to make the target takeover device execute the reconstruction logic.

[0194] The above formulas are dimensionless numerical calculations, and the formulas are obtained by software simulation of a large amount of data to obtain a formula of the nearest real situation, and the preset parameters in the formula are set by a person skilled in the art according to the actual situation.

[0195] The above-described embodiments can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented by software, the above-described embodiments can be implemented in whole or in part in the form of a computer program product.

[0196] Those skilled in the art can realize that the modules and algorithm steps of the examples described in conjunction with the embodiments disclosed herein can be realized by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are realized in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to realize the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0197] In addition, each functional module in each embodiment of the present application can be integrated into one processing module, or each module can exist physically alone, or two or more modules can be integrated into one module.

[0198] The above is merely specific embodiments of the present application, but the protection scope of the present application is not limited thereto, and any person skilled in the art can easily think of changes or replacements within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

[0199] Finally: the above is merely preferred embodiments of the present application, and is not intended to limit the present application, and any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application should be included in the protection scope of the present application.

Claims

1. A method for online reconfiguration control of secondary equipment in a smart substation, characterized in that, The method comprises the following steps: broadcasting fault information containing a digital fingerprint generated by a faulty secondary device to neighboring devices; causing the neighboring devices to verify the digital fingerprint and perform lightweight consensus confirmation, and obtaining a dynamic reputation score of the secondary device based on operation data of the secondary device; the lightweight consensus confirmation specifically comprises: after receiving the fault information, the neighboring devices act as proposal nodes; the verification result of the digital fingerprint is encapsulated into a consensus request together with the fault information, and is broadcast to the neighboring devices; the rest of the neighboring devices respond to the consensus request, independently verify the digital fingerprint, and feed back the verification result; when the number of valid verification results received by the proposal node exceeds a first preset threshold, the proposal node determines that consensus is reached and generates a consensus confirmation signal; selecting a candidate takeover device set from the neighboring devices according to the dynamic reputation score; determining a target takeover device from the candidate takeover device set through a multi-objective optimization algorithm, specifically: constructing a multi-objective optimization function, and dynamically generating multi-objective weight coefficients matched with the current power grid state based on a deep reinforcement learning model; inputting the candidate takeover device set as a solution space into a non-dominated sorting algorithm, and taking the multi-objective optimization function and its weight coefficients as evaluation criteria to obtain a Pareto optimal solution set through iterative solution; sorting the Pareto optimal solution set, and selecting the candidate device corresponding to the highest ranked solution as the preliminary target takeover device; A preliminary target takeover device is subjected to digital-twin-based forward-looking verification, and after verification, the target takeover device is finally determined; a multi-target optimization function wherein, represents a candidate device takeover delay, represents a candidate device operation reliability, represents a candidate device load balancing degree, represents a candidate device communication timeliness, , , , respectively correspond to the weights of the takeover delay, the operation reliability, the load balancing degree, and the communication timeliness; the optimization targets include minimization of the takeover delay, maximization of the operation reliability, optimization of the load balancing degree, and maximization of the communication timeliness; causing the target takeover device to execute reconstruction logic.

2. The intelligent substation secondary equipment online reconfiguration control method of claim 1, wherein, The broadcasting of the fault information containing the digital fingerprint generated by the faulty secondary device to the neighboring devices specifically comprises: collecting a unique identifier and a real-time fault code of the faulty secondary device; combining the unique identifier, the fault code and a current timestamp, and generating the digital fingerprint by using a preset cryptographic hash function; combining the fault code and the digital fingerprint, generating fault information and broadcasting the fault information.

3. The online reconfiguration control method for secondary devices of a smart substation of claim 2, characterized in that, The obtaining of the dynamic reputation score of the secondary device based on operation data of the secondary device specifically comprises: obtaining historical and real-time operation state data of the secondary device, and performing data preprocessing to obtain first data; performing time series weighting on the first data based on an adaptively adjusted sliding time window and a forgetting factor to obtain second data; obtaining statistical features of the second data, and combining device multi-modal features to construct a multi-modal reputation feature vector; inputting the feature vector into a scoring model based on an attention mechanism, calculating and outputting a dynamic reputation score, and the attention mechanism is used to adaptively assign weights to different features in the multi-modal reputation feature vector.

4. The online reconfiguration control method for secondary devices of a smart substation of claim 3, characterized in that, The selecting of the candidate takeover device set from the neighboring devices according to the dynamic reputation score specifically comprises: adding devices with a reputation score higher than a dynamic threshold to an initial candidate takeover device set; performing secondary screening on the initial candidate takeover device set according to a topological position and a load balancing factor, and outputting a candidate takeover device set.

5. The online reconfiguration control method for secondary devices of a smart substation of claim 4, characterized in that, The deep reinforcement learning model comprises: constructing a reinforcement learning model based on an Actor-Critic framework; taking real-time power grid state information as an input state of the model; mapping an output action of the model to a weight coefficient of a multi-objective optimization function; wherein the model is trained by a reward function based on a historical takeover success rate and system stability.

6. The online reconfiguration control method for secondary devices of a smart substation of claim 5, characterized in that, The Pareto optimal solution set is sorted, and the specific process includes the following steps: The multi-objective weight coefficient is used as the evaluation criterion weight of the TOPSIS method; The positive ideal value and the negative ideal value of each optimization target are determined, and the weighted Euclidean distance between each candidate solution in the Pareto optimal solution set and the positive ideal value and the negative ideal value in each optimization target dimension is calculated; The relative closeness of each candidate solution is calculated based on the weighted Euclidean distance, and the sorting result is obtained according to the size of the relative closeness.

7. The online reconfiguration control method for secondary devices of a smart substation of claim 6, characterized in that, The preliminary target takeover device is verified based on digital twinning, specifically: The topology parameters and operating state data of the target takeover device are input into the power grid digital twinning simulation platform; The future load prediction data and new energy output prediction data are imported into the simulation platform to construct a takeover simulation scenario; The power grid dynamic load flow calculation is performed in the takeover simulation scenario to obtain node voltage, line power and power balance data; The verification conclusion is generated based on the results of the power grid dynamic load flow calculation.

8. A system for using the intelligent substation secondary equipment online reconstruction control method according to any one of claims 1-7, comprising: a fingerprint broadcast module for broadcasting the fault information generated by the faulty secondary equipment to the adjacent devices, the fault information containing digital fingerprints; a verification consensus and scoring module for verifying the digital fingerprints by the adjacent devices and performing lightweight consensus confirmation, and obtaining dynamic reputation scores of the secondary devices based on the secondary device operation data; a candidate screening module for screening a candidate takeover device set from the adjacent devices according to the dynamic reputation scores; an optimization selection module for determining a target takeover device from the candidate takeover device set through a multi-objective optimization algorithm; a reconstruction execution module for enabling the target takeover device to execute the reconstruction logic.

Images