Mechanism data driven power system optimization operation method based on short-circuit current constraint

Abstract

The invention relates to the technical field of electrical information, and discloses a short-circuit current constraint-based mechanism data-driven power system optimization operation method, which comprises the following steps of: constructing a mechanism model, and obtaining a data set of a mapping relationship formed by a combined current limiting measure and a short-circuit current level through the mechanism model and expert experience; constructing a multi-layer perceptron model, and training the multi-layer perceptron model through the data set to form a data-driven short-circuit current constraint which can be used for optimization; the data driving short-circuit current constraint, a preset mechanism driving power grid operation constraint and an N-1 security constraint are integrated to form a constraint condition, a target function with the minimum total operation cost as the target is set, a mechanism-data combined driving optimization operation model is constructed, the total operation cost comprises the unit output cost in the normal operation state, and the unit output cost comprises the unit output cost in the normal operation state. The network topology adjustment cost comprises line hot standby and bus segmentation; and solving the optimized operation model to obtain a short-circuit current limiting scheme. According to the invention, the current limiting effect is better.

Description

Technical Field

[0001] This invention relates to the field of electrical information technology, and specifically to a mechanism-driven method for optimizing the operation of a power system based on short-circuit current constraints. Background Technology

[0002] With the development of power systems, the increase in load, and the growing complexity of power grid topology, the problem of excessive short-circuit current has significantly increased. While high short-circuit current levels are beneficial for mitigating external shocks, they also increase the risk of equipment damage and circuit breaker failure. Therefore, it is urgent to address how to effectively reduce the impact of current-limiting measures on power grid dispatching and operation while ensuring safety.

[0003] Essentially, short-circuit current can be limited by adjusting the system topology. Specific implementation methods can be categorized into three types: line hot standby, busbar sectionalization, and unit hot standby. Line hot standby is primarily used for system optimization and further applied to short-circuit current limiting by adjusting the system network topology. It is also used to alleviate system congestion and eliminate voltage overruns. However, disconnecting too many lines to meet short-circuit current constraints can negatively impact system safety to some extent. Busbar sectionalization is considered as adjusting the system topology by disconnecting circuit breakers, which can increase static load margin, system reliability, and operational flexibility. Furthermore, in actual power grid operation, not all units are in operation; a certain number are in standby mode. Therefore, unit hot standby is another key factor affecting the magnitude and distribution of short-circuit current. These system topology adjustment measures are not only flexible and convenient to implement, effectively reducing short-circuit current and making full use of existing power grid resources, but also do not require additional equipment. They can avoid the high installation costs and additional equipment maintenance costs associated with configuring current limiting equipment. However, because they disrupt the original grid structure, they may reduce the reliability of the system. In addition, long-term equipment shutdowns can lead to asset waste.

[0004] In summary, current research on integrated network topology optimization and current limiting based on line hot standby, bus segmentation, and unit hot standby still faces several issues requiring further investigation: First, current research mainly focuses on single current limiting methods, often targeting specific current limiting measures for particular substations exceeding safety limits. This leads to suboptimal and limited current limiting schemes, and a unified integrated network topology optimization method considering multiple methods has not yet been established. Second, existing research has failed to establish a direct relationship between dynamic network topology adjustments and short-circuit currents, thus lacking a unified scheduling method for various current limiting measures. Finally, existing research is insufficient in addressing the impact of network topology adjustments on system security. If the N-1 security requirement of the system can be taken into account, and the operation of line hot standby, bus segmentation, and unit hot standby can be combined in actual power grid operation and scheduling, system security will be improved, the current limiting effect will be more significant, and the current limiting schemes will be more diversified. Summary of the Invention

[0005] To address the lack of a unified approach in existing research for scheduling various current-limiting methods such as line hot reserve, bus segmentation, and unit hot reserve, this study employs a data-driven method to establish constraint expressions for dynamic network topology adjustment and short-circuit current, encompassing these three methods. This leads to the construction of a dynamic network topology optimization scheduling method under short-circuit current constraints. This method comprehensively considers the impact of unit commissioning on the system's short-circuit current and introduces N-1 safety constraints to ensure system safety. The goal is to effectively limit the system's short-circuit current by comprehensively scheduling these three network topology adjustment methods. This invention aims to provide a mechanism-driven data-driven power system optimization operation method based on short-circuit current constraint learning, offering a solution for power system operation under short-circuit current safety requirements. This method not only covers various network topology adjustment methods such as line hot reserve, bus segmentation, and unit hot reserve, but also provides an optimal comprehensive network topology current-limiting scheme while ensuring N-1 safety constraints. The technical solution is as follows:

[0006] This invention is achieved through the following technical solution:

[0007] A mechanism-driven power system optimization operation method based on short-circuit current constraints includes:

[0008] A mechanism model based on line hot standby, bus sectioning and unit hot standby is constructed. A dataset mapping the combined current limiting measures and short-circuit current levels is obtained through the mechanism model and expert experience.

[0009] A multilayer perceptron model is constructed, and the multilayer perceptron model is trained with the dataset to obtain the mapping relationship between combined current limiting measures and short-circuit current level, and finally a data-driven short-circuit current constraint that can be used for optimization is formed.

[0010] The data-driven short-circuit current constraint is integrated with the preset mechanism-driven grid operation constraint and N-1 safety constraint to form a constraint condition. The objective function is set to minimize the total operating cost, and an optimized operation model driven by mechanism and data is constructed. The total operating cost includes the unit output cost under normal operating conditions, as well as the cost of line hot standby and bus segment network topology adjustment.

[0011] Input IEEE standard system example data, solve the optimized operation model, and obtain a short-circuit current limiting scheme that includes network topology adjustment measures such as line hot standby, bus denominator and unit hot standby.

[0012] As an optimization, a mechanism model based on line hot reserve, busbar segmentation, and unit hot reserve is constructed. The specific process of obtaining a dataset that maps the combined current limiting measures to the short-circuit current level through the mechanism model and expert experience is as follows:

[0013] A mechanism model based on line hot standby, busbar segmentation, and unit hot standby is constructed. The mechanism model is used to reflect the impact of topology changes on short-circuit current.

[0014] Drawing on expert experience and concentric relaxation theory, a set of topology adjustment combination current limiting measures, including line hot standby, bus segmentation and unit hot standby, is generated stably and adaptively.

[0015] Based on the aforementioned mechanism model, simulation calculations are performed on each combination measure in the topology adjustment combination current limiting measure set to obtain the corresponding system substation short-circuit current data, forming a training dataset for the mapping relationship between combination current limiting measures and short-circuit current levels.

[0016] As an optimization, the mechanism model is specifically expressed as follows:

[0017] ;

[0018] This indicates the self-impedance of the power plant after adjustment. This represents the initial self-impedance of the power plant. This represents the change in the self-impedance of the power plant after the implementation of current limiting measures.

[0019] As an optimization, the change in the plant's self-impedance after implementing current limiting measures. include , , One or more of them, wherein This refers to the change in the self-impedance of the substation after the line is put into hot standby mode. To measure the change in the self-impedance of the power plant after the busbar is segmented, This refers to the change in the plant's self-impedance after the unit is put into hot standby mode.

[0020] As an optimization, a multilayer perceptron model is constructed, and the multilayer perceptron model is trained using the dataset to obtain the mapping relationship between combined current limiting measures and short-circuit current levels. The specific process of finally forming data-driven short-circuit current constraints that can be used for optimization is as follows:

[0021] The basic model of the multilayer perceptron is designed. The basic model consists of an input layer, several hidden layers and an output layer. The layers are fully connected. The neurons in the hidden layer and the output layer are configured with activation functions to achieve nonlinear mapping, which is used to establish the correlation between combined current limiting measures and short-circuit current level.

[0022] ReLU was selected as the activation function of the hidden layer, and the model's ability to learn complex mapping relationships was enhanced through nonlinear transformation.

[0023] One-hot encoding is performed on the combined rate limiting measures to convert each topology adjustment scheme into an independent binary 0-1 vector, which serves as the input feature of the base model;

[0024] Define a short-circuit current safety distance, which is the difference between the maximum allowable short-circuit current of the equipment and the predicted short-circuit current. It is used to quantitatively describe the short-circuit current safety margin. When the safety distance is ≥0, it indicates that the short-circuit current constraint is met.

[0025] Based on the aforementioned basic model and the safety distance, a linearization technique is used to process the maximum operator within the basic model. Auxiliary variables and Big M method parameters are introduced to transform the nonlinear neural network output into an integrable linearized constraint relationship, thereby obtaining a multilayer perceptron model.

[0026] Using the dataset, the multilayer perceptron model is trained and optimized so that it learns the mapping relationship between combined current limiting measures and short-circuit current levels, ultimately forming a data-driven short-circuit current constraint that can be directly used for power system optimization.

[0027] As an optimization, the safe distance is expressed as:

[0028] ;

[0029] in, Indicates the safe distance from plant i. This represents the short-circuit current value of plant i. This indicates the maximum allowable short-circuit current limit for power plant i.

[0030] As an optimization, the constraint relationship specifically includes:

[0031] ;

[0032] ;

[0033] ;

[0034] ;

[0035] ;

[0036] ;

[0037] ;

[0038] Among them, set Represents all impedances of the power system, including line impedance, bus section impedance, and unit subtransient reactance; Collection Represents all short-circuit current limiting measures in the power system, including line disconnection, bus sectionalizing, and generator start-up and shutdown; variables This represents the input of the (k+1)th hidden layer, and at the same time It also represents the output after the ReLU transformation of the nonlinear activation function of the k-th hidden layer; variable This represents the input of the k-th hidden layer, and at the same time It also represents the output after the ReLU transformation of the nonlinear activation function of the (k-1)th hidden layer; variable The output of the linear mapping of the k-th hidden layer, with parameters... The weights represent the linear mapping of the k-th hidden layer; parameters The bias represents the linear mapping of the k-th hidden layer; and The auxiliary variable introduced for linearization; M is the parameter value of the Big M method.

[0039] As an optimization, the objective function is expressed as:

[0040] ;

[0041] In the formula, set G represents all generating units in the power system; set l represents all lines in the power system that can be used for hot standby; set n represents all substations in the power system that can be segmented by busbars; cost The unit price of coal; power. Represents the output of unit g in its initial state; cost and These represent the unit costs of line hot standby and busbar segmented operation, respectively. These are variables related to the hot standby of line l, used to quantify the relevant information about the hot standby of the line. It is a variable related to the bus segment n, used to quantify the relevant information about the bus segment.

[0042] As an optimization, the mechanism-driven power grid operation constraints and N-1 security constraints are specifically as follows:

[0043] ;

[0044] ;

[0045] ;

[0046] ;

[0047] ;

[0048] ;

[0049] ;

[0050] In the formula, power This represents the output of unit g under accident c; power. Represents the power flow of line l under accident c; power The load representing plant d; collection Represents the total power of all power plants in the system; s(l) and r(l) represent the power at the beginning and end of the line connected to power plant n, respectively; This represents the voltage phase angle of substation n under accident c; and These represent the lower and upper limits of the voltage phase angle of power plant n, respectively. and These represent the lower and upper limits of the power of line l, respectively; and These represent the upper and lower limits of the unit's power output g, respectively; the 0-1 variable w l and w g These represent the start-up and shutdown states of line l and unit g, respectively. l =0 represents the input of line l, w g =0 indicates that unit g is started, w l =1 indicates that line l is disconnected, w g =1 represents unit g being shut down; 0-1 variable This represents the start-stop status of line l under the N-1 safety criterion. This indicates that line l tripped during fault c; otherwise, Phase angle and The ) represent the voltage phase angles at the beginning and end of line l under fault c, respectively; admittance y l The line admittance of line l is represented by M; the parameter value of the large M method is represented by 0-1 variable w. l Represents the line switching status, wl =0 indicates line input, w l =1 indicates the line is disconnected; 0-1 variable This represents the segmentation status of the busbar. The busbar is not operating in sections. Segmented operation of the busbar; assembly This represents the lines within substation i that are eligible for current limiting operations.

[0051] This invention discloses a mechanism-based data-driven power system optimization operation system based on short-circuit current constraints, used to execute the aforementioned mechanism-based data-driven power system optimization operation method based on short-circuit current constraints, comprising:

[0052] The dataset acquisition module is used to construct a mechanism model based on line hot standby, bus sectioning and unit hot standby, and to acquire a dataset that maps the combined current limiting measures to the short-circuit current level through the mechanism model and expert experience.

[0053] The data-driven short-circuit current constraint construction module is used to build a multilayer perceptron model and train the multilayer perceptron model with the dataset to obtain the mapping relationship between combined current limiting measures and short-circuit current level, and finally form a data-driven short-circuit current constraint that can be used for optimization.

[0054] The optimized operation model construction module is used to integrate the data-driven short-circuit current constraint with the preset mechanism-driven grid operation constraint and N-1 safety constraint to form constraint conditions, and set the objective function to minimize the total operating cost to construct the mechanism-data jointly driven optimized operation model. The total operating cost includes the unit output cost under normal operating conditions, as well as the cost of line hot standby and bus segment network topology adjustment.

[0055] The solution module is used to input IEEE standard system example data, solve the optimized operation model, and obtain a short-circuit current limiting scheme that includes network topology adjustment measures such as line hot standby, bus denominator and unit hot standby.

[0056] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0057] 1) This invention integrates various network topology adjustment methods, including line hot standby, busbar segmentation, and unit hot standby, to limit system short-circuit current. It establishes a data-driven relationship between network topology adjustment methods and short-circuit current, providing a unified analysis of the limiting effect of network topology adjustment methods on short-circuit current. Compared to existing technologies that consider a single current-limiting method, this invention conducts data modeling research on the mechanisms of multiple current-limiting methods and combines current-limiting schemes from various network topology adjustment methods, resulting in superior current-limiting effects and greater diversity in current-limiting schemes.

[0058] 2) Considering the N-1 safety constraints of the system, the concept of short-circuit current safety distance is proposed, which effectively quantifies the degree of violation of short-circuit current constraints by different combinations of current limiting measures. The forward propagation formula of the short-circuit current safety distance predictor is processed by the big M method to obtain the short-circuit current constraint that can be integrated with the grid operation constraints in commercial solvers. Attached Figure Description

[0059] The accompanying drawings, which are included to provide a further understanding of embodiments of the invention and form part of this application, do not constitute a limitation thereof. In the drawings:

[0060] Figure 1 The relevant circuit diagrams and simplified wiring diagrams for the modeling method of network topology adjustment for line hot standby are shown in the following: (a) is a simplified wiring diagram of four transmission lines connected to the short-circuit current concerned substation n; (b) is the equivalent circuit diagram of line l in the commissioning state.

[0061] Figure 2 The relevant circuit diagrams and simplified wiring diagrams for the modeling method of network topology adjustment for bus segment operation are as follows: (a) is a simplified wiring diagram of the short-circuit current of the substation bus n that can be segmented in the modeling method of network topology adjustment for bus segment operation; (b) is an equivalent circuit diagram of bus n operating normally without being segmented in the modeling method of network topology adjustment for bus segment operation.

[0062] Figure 3 The diagram shows the impact of unit hot standby on the short-circuit current level of the system. (a) is the wiring diagram of unit g when the unit is in operation, and (b) is the equivalent circuit diagram of unit g after it is shut down.

[0063] Figure 4 This is a schematic diagram of the forward propagation formula inside an MLP.

[0064] Figure 5 Wiring diagram for the IEEE-30 node system.

[0065] Figure 6 This shows the distribution of the short-circuit current in the system in Example 1-5.

[0066] The attached diagram shows the markings and corresponding component names:

[0067] 11-Conductor sheet, 12-Second insulator, 13-First insulator. Detailed Implementation

[0068] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the embodiments and accompanying drawings. The illustrative embodiments and descriptions of the present invention are only used to explain the present invention and are not intended to limit the present invention.

[0069] This invention proposes a mechanism-driven power system optimization operation method based on short-circuit current constraint learning. This method comprehensively considers three different network topology adjustment methods: line hot reserve, bus segmentation, and unit hot reserve. It comprehensively limits the short-circuit current level of the system and ensures the N-1 safety of the system after the comprehensive current limiting is implemented. The method performs data-driven modeling of line hot reserve, bus segmentation, and unit hot reserve. Then, based on the short-circuit current constraint and the system N-1 safety constraint, a long-term optimal scheduling model of the power system is established with the objective function of minimizing the generation cost, line hot reserve cost, and bus segmentation cost. The method provides a comprehensive current limiting scheme for hot reserve, bus segmentation, and unit hot reserve, including unit scheduling plan.

[0070] Furthermore, the optimization operation model for long-term power system optimization established in this invention is a mixed-integer linear programming (MILP) problem, which can be solved by mature commercial solvers. In the MATLAB environment, this invention calls the commercial solver Gurobi for optimization calculations, and obtains a short-circuit current limiting scheme that includes integrated network topology adjustment measures such as line hot reserve, bus segmentation, and unit hot reserve. The results show that the proposed current limiting method can effectively limit current, and also provides unit scheduling plans and current limiting schemes for line hot reserve, bus segmentation, and unit hot reserve.

[0071] This embodiment 1 provides a mechanism-driven power system optimization operation method based on short-circuit current constraints, including:

[0072] Step 1: Construct a mechanism model based on line hot standby, bus sectioning and unit hot standby, and obtain a dataset that maps the combined current limiting measures to the short-circuit current level through the mechanism model and expert experience.

[0073] Step 1 is actually mechanism-guided training data acquisition. Based on the mechanism models of line hot standby, bus sectioning, and unit hot standby, and drawing on expert experience and concentric relaxation theory, a training set is randomly and adaptively acquired to map the combined current limiting measures to short-circuit current constraints. This provides a training dataset for data-driven machine learning. The specific process is as follows:

[0074] Step 1.1: Construct a mechanism model based on line hot standby, bus sectioning and unit hot standby. The mechanism model is used to reflect the impact of topology changes on short-circuit current.

[0075] The mechanism model is a topology adjustment model for current limiting measures. When a three-phase short-circuit fault occurs in the plant of interest, according to the superposition principle, the relationship of the plant of interest i is as follows (1):

[0076] (1)

[0077] In the formula: U0 is the voltage of the substation of interest before the fault i, i f and Z ff Let i be the short-circuit current and self-impedance of the substation of interest. Its substation voltage can be considered as 1.0 pu, meaning the calculation of the three-phase short-circuit current of the substation of interest can be simplified to the following formula:

[0078] (2)

[0079] Depend on Figure 1 (a) and (b) illustrate the wiring methods for network topology adjustment using line hot standby. Line hot standby leads to topology changes in the plant of interest i and the entire system, thereby affecting its self-impedance Z. ff The change is shown in equation (3) below:

[0080] (3)

[0081] In equation (3), Z ff Z0 represents the adjusted self-impedance of the plant i of interest, and Z0 represents the initial self-impedance of the plant i of interest. ΔZ TS The change in self-impedance caused by line disconnection can be obtained by solving the change in the impedance matrix before and after disconnection.

[0082] Depend on Figure 2 (a) and (b) illustrate the wiring methods for network topology adjustment through busbar segmentation. Busbar segmentation leads to topology changes in the plant of interest (i) and the entire system, thereby affecting its self-impedance (Z). ff The change is as follows: As shown:

[0083] (4)

[0084] In equation (4), △Z BS The change in self-impedance of the plant i of interest caused by the busbar segmentation can be obtained by solving the change in the impedance matrix before and after the segmentation.

[0085] Depend on Figure 3 (a) and (b) show the wiring methods for network topology adjustment methods for unit hot standby. Unit hot standby will cause topology changes between the plant of interest i and the entire system, thereby affecting its self-impedance Z. ff The change is shown in equation (5) below:

[0086] (5)

[0087] In equation (5), △Z UC The change in self-impedance caused by the unit's hot standby can be obtained by solving the change in the impedance matrix before and after the standby.

[0088] Based on the wiring method of the network topology adjustment methods for combined current limiting measures of line hot standby, bus sectioning, and unit hot standby, the combined current limiting measures will cause topology changes in the plant of interest i and the entire system, thereby affecting its self-impedance Z. ff The change is shown in equation (6) below:

[0089] (6)

[0090] Mode In the middle, △Z CLM The change in self-impedance caused by the unit's hot standby can be obtained by solving the change in the impedance matrix before and after the standby.

[0091] Step 1.2: Drawing on expert experience and concentric relaxation theory, a set of topology adjustment combination current limiting measures, including line hot standby, bus segmentation, and unit hot standby, is generated in a randomized adaptive manner.

[0092] This process can be broken down into three progressive steps: "expert experience constrains to define the scope → concentric relaxation theory simplifies the space → random adaptive sampling generates the set," forming a set of measures that combines engineering feasibility and comprehensiveness, as detailed below:

[0093] I. Delineate the "feasible domain boundary" of topology adjustment measures based on expert experience. Expert experience is used to filter out topology adjustment schemes that are "physically infeasible" or "engineeringally meaningless," defining a reasonable range for subsequent generation processes. Specifically, this includes:

[0094] 1. Determine the basic operational units for the three types of traffic restriction measures:

[0095] Hot standby lines: According to the dispatching regulations, the "types of lines that can be hot standby" are clearly defined (such as 110kV and above tie lines, excluding important load lines with single-circuit power supply) and the "maximum number of hot standby lines" is limited (such as a maximum of 2 hot standby lines in the same substation, to avoid excessive reduction in power supply reliability).

[0096] Busbar segmentation: Based on the main wiring configuration of the substation, determine the "range of busbars that can be segmented" (e.g., 220kV and above busbars usually have segmentation switches, while 10kV busbars may not), and specify the "minimum interval for segmentation operations" (e.g., the interval between two segmentation operations on the same busbar should be at least 15 minutes, in accordance with on-site operation specifications).

[0097] Unit hot standby: Based on the unit type, clarify the "conditions for hot standby" (e.g., thermal power units can be hot standby, while new energy units are usually not included due to regulation characteristics limitations), and set the "upper limit of hot standby unit capacity" (e.g., 10% of the total system capacity, to avoid affecting power supply capacity).

[0098] 2. Set correlation constraints for the combination of measures:

[0099] Expert experience can identify "strong correlation rules" between measures, such as:

[0100] "If the 220kV busbar of a substation is segmented, at least one of its associated 110kV outgoing lines must be kept on standby" (to avoid overloading of the 110kV line due to the transfer of local short-circuit current after segmentation).

[0101] "If the hot standby capacity of a certain region's generating units exceeds 500MW, then at least two parallel lines must be maintained in operation" (balancing power transmission capacity). These rules are written into the generation algorithm through logical expressions to filter out conflicting combinations.

[0102] II. Simplifying the "Combinatorial Space" of Topology Adjustment Based on Concentric Relaxation Theory: Concentric relaxation theory reduces the complexity of the combinatorial space by "gradually shrinking the constraint boundary," avoiding combinatorial explosion caused by an excessive number of measures. Specific operations are as follows:

[0103] 1. Construct the "basic constraint layer" (outermost concentric constraint):

[0104] Based on the constraint of "not causing system power imbalance", the maximum possible adjustment range of the three types of measures is determined:

[0105] Line hot standby: Relaxation is defined as "single line power flow ≤ 120% of rated capacity" (instead of strict 100%), allowing short-term overload to expand the range of options;

[0106] Busbar segmentation: Relaxation means "the load difference between each busbar segment after segmentation is ≤30%" (instead of strict load balancing), reducing the severity of the segmentation operation;

[0107] Unit hot standby: Relaxed to "system rotational standby ≥5% after hot standby" (instead of the conventional 10%), expanding the pool of units that can be hot standby.

[0108] 2. Construct a "safety constraint layer" (inner concentric constraint):

[0109] On top of the basic constraint layer, a "preliminary short-circuit current screening" constraint is superimposed:

[0110] For all combinations within the basic constraint layer, by simplifying short-circuit current calculations (e.g., using impedance matrix for rapid estimation), combinations with "short-circuit current ≤ 1.2 times the limit" (rather than strictly ≤ the limit) are retained, further narrowing the range. Note: The core of concentric relaxation is "relax first, then tighten," gradually approximating the reasonable solution space through multiple layers of constraints, reducing the generation of invalid combinations.

[0111] Third, random adaptive sampling generates the final set of topology adjustment combined flow limiting measures. Within the above constraints, the measure set is generated through an adaptive mechanism of "random sampling + dynamic feedback" to ensure coverage and representativeness.

[0112] 1. Stratified random sampling:

[0113] For the three types of measures—line hot standby, busbar segmentation, and unit hot standby—the sampling ratio is allocated according to the "importance weight" (e.g., the sampling ratio for line hot standby of key transmission channels is 40%, busbar segmentation is 30%, and unit hot standby is 30%).

[0114] Within each type of measure, "uniform random sampling" is used. For example, 1-3 lines are randomly selected from 10 hot backup lines (the number follows a uniform distribution) to form an initial combination.

[0115] 2. Adaptive feedback adjustment:

[0116] Perform a "diversity assessment" on the generated combinations: calculate the "Hamming distance" (difference) between any two combinations. If the difference between 5 consecutive combinations is less than 20%, trigger a "disturbance mechanism" (such as forcibly replacing a hot standby line) to avoid sample redundancy.

[0117] Combined with short-circuit current sensitivity feedback: For "short-circuit current sensitive measures" found in the sampling (such as a short-circuit current reduction of >10% after a certain line is in hot standby), increase the probability of their occurrence in the combination (such as increasing from 10% to 20%), and strengthen the coverage of key measures.

[0118] 3. Final screening and deduplication:

[0119] The generated combinations are processed in two steps:

[0120] Eliminate combinations that violate the hard constraints of expert experience (such as the association rule conflicts mentioned above);

[0121] Remove duplicate or equivalent combinations (such as "hot standby line A + segment bus B" and "segment bus B + hot standby line A" are essentially the same, retain one of them), and finally form a "topology adjustment combination current limiting measure set" of moderate size (such as 1000-5000 sets).

[0122] Through the above process, the generated set of measures not only conforms to the engineering reality of power system operation (guaranteed by expert experience), but also efficiently covers key adjustment schemes (concentric relaxation reduces complexity), and has data diversity (random adaptive sampling), providing high-quality "input-output" samples for the MLP model training in step 2.

[0123] Step 1.3: Based on the aforementioned mechanism model, perform simulation calculations on each combination measure in the topology adjustment combination current limiting measure set, obtain the corresponding system substation short-circuit current data, and form a training dataset for the mapping relationship between combination current limiting measures and short-circuit current levels.

[0124] The specific implementation process is as follows:

[0125] The combined measures in the topology adjustment combined current limiting measures are transformed into parameter inputs that can be identified by the mechanism model, where line hot standby corresponds to line switching status parameters, bus segmentation corresponds to bus structure parameters, and unit hot standby corresponds to unit operating status parameters.

[0126] Based on the aforementioned mechanism model, short-circuit current simulation calculations are performed on the parameter configurations corresponding to each combined measure to obtain the short-circuit current values ​​of each plant in the system.

[0127] Each combination of current limiting measures is associated with and stored with the corresponding short-circuit current value to form a training dataset containing the mapping relationship between "combined current limiting measures and short-circuit current levels".

[0128] Step 2: Construct a multilayer perceptron model and train the multilayer perceptron model with the dataset to obtain the mapping relationship between combined current limiting measures and short-circuit current level, and finally form a data-driven short-circuit current constraint that can be used for optimization.

[0129] Step 2 actually involves data-guided short-circuit current constraint learning, using one-hot encoding to process topology changes and improve the training effect of the machine learning model. The specific process of training and optimizing the machine learning model to obtain the mapping from combined current limiting measures to the power system short-circuit current level is as follows:

[0130] Step 2.1: Design the basic model of the multilayer perceptron model. The basic model consists of an input layer, several hidden layers, and an output layer. All layers are fully connected. The neurons in the hidden layers and the output layer are configured with activation functions to achieve nonlinear mapping, which is used to establish the correlation between combined current limiting measures and short-circuit current levels.

[0131] The Multilayer Perceptron (MLP) model consists of an input layer, an output layer, and several hidden layers. The input layer receives feature vectors, the output layer provides prediction results, and the hidden layers transform the input features through nonlinear transformations and feature extraction. Different layers are fully connected, meaning all neurons in each layer are connected to all neurons in the next layer. Furthermore, each neuron in both the hidden and output layers has an activation function, introducing a nonlinear mapping. The mathematical relationship is as follows: As shown:

[0132] (7)

[0133] In equation (7), parameter b represents the bias of each hidden layer, which can ensure that neurons cannot be activated arbitrarily; parameter W represents the weight of each hidden layer, which represents the connection strength between neurons; function f represents the activation function, which plays a nonlinear mapping role and can limit the output amplitude of neurons to a certain range.

[0134] Step 2.2: Select ReLU as the activation function of the hidden layer to enhance the model's ability to learn complex mapping relationships through nonlinear transformation.

[0135] In neural networks (basic models), activation functions are designed to help the network learn complex patterns in data. They add a non-linear operation to all hidden and output layers, making the neural network's output more complex and expressive. This patent selects ReLU as the activation function for the hidden layers of the MLP network, and its mathematical relationship is shown in equation (8):

[0136] (8)

[0137] Step 2.3: Perform one-hot encoding on the combined current limiting measures, converting each topology adjustment scheme into an independent binary 0-1 vector, which serves as the input feature of the basic model.

[0138] Considering the limited generalization ability of MLP to topology changes, one-hot encoding is used to represent the rate limiting measures. The rate limiting measure dataset is represented by one-hot encoding, where each different topology adjustment is encoded as an independent binary 0-1 vector, with a corresponding bit of 1 indicating that the rate limiting measure is used and 0 indicating that it is not used.

[0139] Step 2.4: Define the short-circuit current safety distance. The safety distance is the difference between the maximum allowable short-circuit current of the equipment and the predicted short-circuit current. It is used to quantitatively describe the short-circuit current safety margin. When the safety distance is ≥0, it means that the short-circuit current constraint is met.

[0140] To quantitatively describe the degree to which different current-limiting measures violate short-circuit current safety constraints, a safety distance d based on short-circuit current is proposed.SCC The short-circuit current safety margin of the power plant is described by the following formula (9):

[0141] (9)

[0142] In equation (9), the current The value of the short-circuit current of the plant i of interest is represented by the following equation (10).

[0143] (10)

[0144] Step 2.5: Based on the basic model and the safety distance, the maximum operator inside the basic model is processed using linearization technology. Auxiliary variables and Big M method parameters are introduced to transform the nonlinear neural network output into an integrable linearized constraint relationship, thereby obtaining the multilayer perceptron model.

[0145] This step involves reconstructing the short-circuit current constraint based on MLP.

[0146] Depend on Figure 4 Simplifying the forward propagation formula of the MLP, it can be seen that linearization technology is required to process the maximum operator inside the MLP. The reconstructed data-driven short-circuit current constraint relationship is shown in the following equations (11)-(17):

[0147] (11)

[0148] (12)

[0149] (13)

[0150] (14)

[0151] (15)

[0152] (16)

[0153] (17)

[0154] In equations (11)-(17), where the set Represents all impedances of the power system, including line impedance, bus section impedance, and unit subtransient reactance; Collection Represents all short-circuit current limiting measures in the power system, including line disconnection, bus sectionalizing, and generator start-up and shutdown; variables This represents the input of the (k+1)th hidden layer, and at the same time It also represents the output after the ReLU transformation of the nonlinear activation function of the k-th hidden layer; variable This represents the input of the k-th hidden layer, and at the same time It also represents the output after the ReLU transformation of the nonlinear activation function of the (k-1)th hidden layer; variable The output of the linear mapping of the k-th hidden layer, with parameters... The weights represent the linear mapping of the k-th hidden layer; parameters The bias represents the linear mapping of the k-th hidden layer; and The auxiliary variable introduced for linearization; M is the parameter value of the Big M method, a sufficiently large positive number.

[0155] Step 2.6: Using the dataset, train and optimize the multilayer perceptron model so that the multilayer perceptron model learns the mapping relationship between combined current limiting measures and short-circuit current levels, and finally forms a data-driven short-circuit current constraint that can be directly used for the optimized operation of the power system.

[0156] Data-driven short-circuit current constraint is .

[0157] Step 3: Integrate the data-driven short-circuit current constraint with the preset mechanism-driven grid operation constraint and N-1 safety constraint to form constraint conditions, and set the objective function to minimize the total operating cost to construct an optimized operation model driven by mechanism and data. The total operating cost includes the unit output cost under normal operating conditions, as well as the cost of line hot standby and bus segment network topology adjustment.

[0158] This step is a mechanism-data joint-driven system optimization operation, which integrates the mechanism-driven grid operation constraints and N-1 constraints with the data-driven short-circuit current operation constraints to form an integrated mechanism-data driven power system optimization operation model to support power system optimization operation decisions that take into account short-circuit current constraints.

[0159] The specific process is as follows:

[0160] Step 3.1: With minimizing the total operating cost as the objective function, determine the dynamic network topology optimization operation. The total operating cost includes the unit output cost under normal operating conditions, as well as the cost of line hot standby and bus segmentation network topology adjustment, as shown in the following expression (18):

[0161] (18)

[0162] Mode In this context, set G represents all generating units in the power system; set l represents all lines in the power system that can be used for hot standby; set n represents all substations in the power system that can be segmented by busbars; cost The unit price of coal; power. Represents the output of unit g in its initial state; cost and This represents the unit cost of line hot standby and busbar segmented operation. These are variables related to the hot standby of line l, used to quantify the relevant information about the hot standby of the line. It is a variable related to the bus segment n, used to quantify the relevant information about the bus segment.

[0163] Step 3.2: Determine the constraints of the dynamic network topology optimization scheduling model considering N-1 security constraints, solve the optimization operation model based on dynamic network topology, and output the network topology adjustment scheme. The constraint expressions are shown in equations (19)-(25) below:

[0164] (19)

[0165] (20)

[0166] (twenty one)

[0167] (twenty two)

[0168] (twenty three)

[0169] (twenty four)

[0170] (25)

[0171] In equations (19)-(25), power This represents the output of unit g under accident c; power. Represents the power flow of line l under accident c; power The load representing plant d; collection Represents the total power of all power plants and substations in the system; s(l) and r(l) represent the power at the beginning and end of the line connected to power plant n, respectively; phase angle This represents the voltage phase angle of substation n under accident c; phase angle and These represent the upper and lower limits of power plant n, respectively; power and These represent the upper and lower limits of line l, respectively; power. and These represent the upper and lower limits of unit g, respectively; 0-1 variable w l and w g These represent the start-up and shutdown states of line l and unit g, respectively. l =0 represents the input of line l, wg =0 indicates that unit g is started, w l =1 indicates that line l is disconnected, w g =1 represents unit g being shut down; 0-1 variable This represents the start-stop status of line l under the N-1 safety criterion. This indicates that line l tripped during fault c; otherwise, Phase angle and These represent the voltage phase angles at the beginning and end of line l under fault c, respectively; admittance y l The line admittance of line l is represented by M, which is a sufficiently large positive number; the variable w is 0-1. l Represents the line switching status, w l =0 indicates line input, w l =1 indicates the line is disconnected; 0-1 variable This represents the segmentation status of the busbar. The busbar is not operating in sections. Segmented operation of the busbar; assembly This represents the lines within substation i that are eligible for current limiting operations.

[0172] Step 4: Input IEEE standard system example data, solve the optimized operation model, and obtain a short-circuit current limiting scheme that includes network topology adjustment measures such as line hot standby, bus denominator and unit hot standby.

[0173] This step involves inputting IEEE standard system example data to optimize and solve the power system optimization operation model for limiting short-circuit current through dynamic network topology optimization, and deriving a short-circuit current limiting scheme that includes a total network topology adjustment method encompassing line hot standby, bus segmentation, and unit hot standby.

[0174] The effects of the present invention will be illustrated below through specific embodiments.

[0175] Example:

[0176] The effectiveness of the proposed model was verified on the IEEE-30 test system. Figure 5 The topology diagram of the simulation system is shown. The modified 30-node system includes 41 lines, 6 generators, and 2 nodes that can be segmented by busbar. The generator nodes are 1, 2, 13, 22, 23, and 27. Nodes 2 and 6 are selected as nodes that can operate in segmented busbar configuration. , and , These are two sub-nodes formed after segmenting a single node busbar.

[0177] To compare and verify the model proposed in this invention, five calculation examples were set up as shown in Table 1.

[0178] Table 1 Examples 1-5 with different settings

[0179] Table 2 shows the unit scheduling plans and integrated flow restriction measures for the five different calculation results, and Table 3 shows the economic costs.

[0180] Table 2 shows the unit scheduling plan and integrated flow restriction measures for examples 1-5.

[0181]

[0182] G n L represents the generating unit located at plant n; m-n This indicates the line located between power station m and n.

[0183] Table 3 Comparison of different costs in Examples 1-5

[0184]

[0185] Figure 6 The distribution of short-circuit current in the system in Examples 1-5 is shown. Figure 6 In the examples 1-5, the short-circuit currents of all nodes of interest in the short-circuit current are highlighted with black boxes.

[0186] The analysis of examples 1-5 is as follows.

[0187] Before any short-circuit current limiting measures were implemented, the short-circuit currents at nodes 1, 2, and 22 within the black box exceeded the limit. It can also be seen that the combined current limiting measures given by the proposed model can effectively control the short-circuit current level within a certain limit, indicating that the model has accuracy in system operation optimization decision considering short-circuit current constraints.

[0188] Example 1 considers only the unit combination under grid operation constraints, with all lines in operation and units 1, 2, 22, and 23 in operation. Therefore, the current short-circuit current level of the entire system is at its highest, with the short-circuit currents at nodes 1, 2, and 22 exceeding the threshold of 12kA. Thus, current limiting measures are required to meet the short-circuit current constraints.

[0189] Compared to Example 1, Example 2, based on short-circuit current constraints, employs current-limiting measures with line hot standby, increasing the total cost by 600,000 yuan. Specific measures include disconnecting line L... 1-2 and L 22-24 Two lines are used to limit the short-circuit current. Line L 1-2 and L 22-24The short-circuit currents at nearby nodes 1, 2, and 22 decreased by 4.08 kA, 3.55 kA, and 1.03 kA, respectively. However, considering the economics of line disconnection, Example 2 had the highest total cost, mainly because disconnecting two important transmission lines significantly increased the cost of topology adjustments. This demonstrates that relying solely on line disconnection to limit short-circuit current can negatively impact the economic operation of the power system.

[0190] Example 3, based on Example 2, employs a combination of unit hot standby and line hot standby measures to limit short-circuit current. Specific measures include disconnecting line L... 1-2 Shut down unit 22 and start unit 13. Since the electrical distance between node 22 and unit 22 is very close, and considering the economic efficiency of the current limiting measures, line L will not be disconnected. 22-24 The proposed solution is to shut down unit 22 while simultaneously starting unit 13 to maintain load balance. This is because shutting down units with higher subtransient reactance would significantly increase the generation cost of other units with lower subtransient reactance. The short-circuit currents of units 1, 2, and 22 decreased to 9.63 kA, 11.21 kA, and 3.59 kA, respectively. Although Example 3 increased generator costs compared to Example 2, the change in line disconnection strategy reduced topology adjustment costs, resulting in a total cost reduction of 136 thousand yuan compared to Example 2. This demonstrates that the combined approach of unit hot standby and line disconnection is more economical than either approach alone.

[0191] Example 4, based on Example 2, employs a combination of busbar segmentation and line hot standby measures to limit short-circuit current. Specific measures include disconnecting line L... 1-2 Furthermore, busbar segmentation was performed at node 6. Since node 6 is electrically close to node 22, busbar segmentation at node 6 effectively reduces the short-circuit current level near node 22, and eliminates the need to disconnect line L. 22-24 The short-circuit currents of busbars 1, 2, and 22 decreased to 9.61 kA, 11.11 kA, and 11.87 kA, respectively. Compared to Scheme 2, the effective reduction in line disconnection costs leads to a significant reduction in the overall topology adjustment costs. Therefore, without changing generator costs, the total cost of Scheme 4 is 150,000 yuan lower than that of Scheme 2. This demonstrates that combining busbar segmentation with line disconnection is also more economical than either method alone.

[0192] Since examples 2-4 all disconnected line L 1-2 The adjusted system does not satisfy the N-1 constraint. Therefore, based on Example 4, this calculation case considers the N-1 safety constraint and employs a combination of line hot standby, unit hot standby, and bus segmentation measures to limit the short-circuit current. Specifically, this involves disconnecting line L... 22-24The busbar was segmented at node 6, and unit 1 was shut down while unit 13 was started. Due to the shutdown of unit 1, the short-circuit current levels at nodes 1 and 2 decreased significantly, thus eliminating the need to disconnect line L1-2. The short-circuit currents at nodes 1, 2, and 22 decreased by 7.76 kA, 4.42 kA, and 1.01 kA, respectively. Therefore, the line L1-2 was no longer disconnected. 1-2 The proposed solution satisfies the N-1 constraint compared to Example 2-4, but its cost is also the highest due to the consideration of the N-1 constraint.

[0193] Example 2 also discloses a mechanism data-driven power system optimization operation system based on short-circuit current constraints, used to execute the mechanism data-driven power system optimization operation method based on short-circuit current constraints described in Example 1, including:

[0194] The dataset acquisition module is used to construct a mechanism model based on line hot standby, bus sectioning and unit hot standby, and to acquire a dataset that maps the combined current limiting measures to the short-circuit current level through the mechanism model and expert experience.

[0195] The data-driven short-circuit current constraint construction module is used to build a multilayer perceptron model and train the multilayer perceptron model with the dataset to obtain the mapping relationship between combined current limiting measures and short-circuit current level, and finally form a data-driven short-circuit current constraint that can be used for optimization.

[0196] The optimized operation model construction module is used to integrate the data-driven short-circuit current constraint with the preset mechanism-driven grid operation constraint and N-1 safety constraint to form constraint conditions, and set the objective function to minimize the total operating cost to construct the mechanism-data jointly driven optimized operation model. The total operating cost includes the unit output cost under normal operating conditions, as well as the cost of line hot standby and bus segment network topology adjustment.

[0197] The solution module is used to input IEEE standard system example data, solve the optimized operation model, and obtain a short-circuit current limiting scheme that includes network topology adjustment measures such as line hot standby, bus denominator and unit hot standby.

[0198] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for power system optimal operation based on short-circuit current constraint mechanism data driving, characterized in that, The application relates to a method for constructing a short-circuit current constraint based on mechanism and data. The method comprises the following steps: a mechanism model based on line thermal backup, bus sectioning and unit thermal backup is constructed, and a data set of mapping relationship between combined current-limiting measures and short-circuit current levels is obtained through the mechanism model and expert experience; a multilayer perception model is constructed, and the multilayer perception model is trained through the data set to obtain the mapping relationship between the combined current-limiting measures and the short-circuit current levels, so that a data-driven short-circuit current constraint which can be used for optimization is finally formed; the data-driven short-circuit current constraint is integrated with preset mechanism-driven power grid operation constraints and N-1 safety constraints to form a constraint condition, and a target function of minimizing total operation cost is set as a target function, so that an optimization operation model driven by mechanism and data is constructed, and the total operation cost comprises unit output cost in a normal operation state and network topology adjustment cost comprising line thermal backup and bus sectioning; 2. The mechanism data-driven power system optimal operation method based on short-circuit current constraint according to claim 1, characterized in that, IEEE standard system calculation data is inputted, and the optimization operation model is solved to obtain a short-circuit current limiting scheme of total network topology adjustment means comprising line thermal backup, bus sectioning and unit thermal backup. The specific process of constructing a mechanism model based on line thermal backup, bus sectioning and unit thermal backup and obtaining a data set of mapping relationship between combined current-limiting measures and short-circuit current levels through the mechanism model and expert experience is as follows: a mechanism model based on line thermal backup, bus sectioning and unit thermal backup is constructed, and the mechanism model is used for reflecting the influence of topology change on short-circuit current; expert experience and concentric relaxation theory are referred to, and a topology adjustment combined current-limiting measure set comprising line thermal backup, bus sectioning and unit thermal backup is randomly and adaptively generated; 3. The mechanism data-driven power system optimal operation method based on short-circuit current constraint according to claim 1 or 2, characterized in that, based on the mechanism model, simulation calculation is carried out on each combined measure in the topology adjustment combined current-limiting measure set, corresponding system station short-circuit current data is obtained, and a training data set of mapping relationship between combined current-limiting measures and short-circuit current levels is formed. ； Zs,adj represents the adjusted self-impedance of the plant station, Zs,0 represents the initial self-impedance of the plant station, ΔZs represents the change in self-impedance of the plant station after the execution of the flow-limiting measure.

4. The mechanism data-driven power system optimal operation method based on short-circuit current constraint according to claim 3, characterized in that, The amount of change in the self-impedance of the plant station after the current limiting measure is performed comprising , , one or more of, is the amount of change in the self-impedance of the plant station after the line thermal backup is performed, is the amount of change in the self-impedance of the plant station after the bus sectioning is performed, is the amount of change in the self-impedance of the plant station after the unit thermal backup is performed.

5. The mechanism data-driven power system optimal operation method based on short-circuit current constraint according to claim 1, characterized in that, The mechanism model is specifically represented as: a multilayer perception model is constructed, and the multilayer perception model is trained through the data set to obtain the mapping relationship between the combined current-limiting measures and the short-circuit current levels, so that a data-driven short-circuit current constraint which can be used for optimization is finally formed. a basic model of the multilayer perception model is designed, the basic model is composed of an input layer, a plurality of hidden layers and an output layer, full connection is adopted between layers, activation functions are configured to the hidden layers and the output layer to realize nonlinear mapping, and the activation functions are used for establishing the correlation between the combined current-limiting measures and the short-circuit current levels; ReLU is selected as the hidden layer activation function, and the learning ability of the model to complex mapping relationship is enhanced through nonlinear transformation; One-hot coding is carried out on the combined current-limiting measures, each topology adjustment scheme is converted into an independent binary 0-1 vector, and the binary 0-1 vector is used as the input feature of the basic model; a short-circuit current safety distance is defined, the safety distance is the difference between the maximum short-circuit current allowed by equipment and the predicted short-circuit current, and is used for quantitatively describing the short-circuit current safety margin; when the safety distance is greater than or equal to 0, it is indicated that the short-circuit current constraint is met. Based on the base model and the safety distance, a linearization technique is used to process the maximum operator inside the base model, auxiliary variables and large M method parameters are introduced, and the nonlinear neural network output is converted into an integrable linearized constraint relationship, thereby obtaining a multilayer perceptron model; Using the data set, the multilayer perceptron model is trained and optimized, so that the multilayer perceptron model learns the mapping relationship from the combined current limiting measure to the short circuit current level, and finally forms a data-driven short circuit current constraint that can be directly used for power system optimization operation.

6. The mechanism data-driven power system optimal operation method based on short-circuit current constraint according to claim 5, characterized in that, The safety distance is represented as: ； wherein, denotes the safety distance of the plant station i, denotes the short-circuit current value of the plant station i, denotes the maximum short-circuit current limit value allowed by the plant station i.

7. The mechanism data-driven power system optimal operation method based on short-circuit current constraint according to claim 5, characterized in that, The constraint relationship specifically includes: ； ； ； ； ； ； ； where set represents all the impedances of the power system, including line impedance, bus section impedance and unit sub-transient reactance; set represents all the short-circuit current limiting measures of the power system, including line tripping, bus section and unit start-stop; variable represents the input of the k+1th hidden layer, while also represents the output of the kth hidden layer after the nonlinear activation function ReLU changes; variable represents the input of the kth hidden layer, while also represents the output of the k-1th hidden layer after the nonlinear activation function ReLU changes; variable represents the output of the kth hidden layer after linear mapping, parameter represents the weight of the linear mapping of the kth hidden layer; parameter represents the bias of the linear mapping of the kth hidden layer; and auxiliary variable introduced for linearization; M is the parameter value of the large M method.

8. The mechanism data-driven power system optimal operation method based on short-circuit current constraint of claim 1, wherein, The objective function is represented as: ； where set G represents all generators in the power system; set I represents all lines that can be hot standby in the power system; set N represents all substations that can be sectionalized in the power system; cost represents the unit price of coal; power represents the output of generator g in the initial state; Cost and represent the unit cost of line hot standby and bus section operation, respectively, is a variable related to line l hot standby, which is used to quantify the relevant situation of line hot standby, is a variable related to bus section n, which is used to quantify the relevant situation of bus section.

9. The mechanism data-driven power system optimal operation method based on short-circuit current constraint of claim 1, wherein, The mechanism-driven power grid operation constraint and the N-1 safety constraint are specifically: ； ； ； ； ； ； ； where power represents the generation of unit g at contingency c; power represents the flow of line 1 at contingency c; power represents the load of plant station d; set represents the sum of all plant stations of the system; s(l) and r(l) represent the power at the beginning and end of the line connected to the station n, respectively; φn(c) represents the voltage phase angle of the station n under contingency c; and φnmax and φnmin represent the upper and lower limits of the voltage phase angle of the station n, respectively; and plmin and plmax represent the lower and upper limits of the power of the line l, respectively; and pgmax and pgmin represent the upper and lower limits of the power of the generator g, respectively; 0-1 variable w l and w g represent the start and stop state of the line l and the generator g, respectively, represents the line l being put into operation, represents the generator g being started, represents the line l being disconnected, represents the generator g being stopped. 0-1 variable represents the on-off state of line l under N-1 security criterion, represents the on-off state of line l under contingency c, otherwise, ; phase angle and respectively represent the voltage phase angle at the beginning and end of line l under contingency c; admittance y l represents the line admittance of line l; M is the value of the large M method parameter; 0-1 variable representing the status of bus sectioning, bus unsectioned operation, bus sectioned operation; set representing the lines included in station i that can be operated with current limiting.

10. A short circuit current constraint based mechanism data driven power system optimal operation system, configured to perform the short circuit current constraint based mechanism data driven power system optimal operation method according to any one of claims 1-9, characterized in that, Including: The data set acquisition module is used to construct a mechanism model based on line thermal backup, bus sectioning and unit thermal backup, and to obtain a data set that maps the combined current limiting measure to the short circuit current level through the mechanism model and expert experience; The data-driven short circuit current constraint construction module is used to construct a multilayer perceptron model, and train the multilayer perceptron model through the data set to obtain the mapping relationship from the combined current limiting measure to the short circuit current level, and finally form a data-driven short circuit current constraint that can be used for optimization; The optimization operation model construction module is used to integrate the data-driven short circuit current constraint with the preset mechanism-driven power grid operation constraint and N-1 safety constraint to form a constraint condition, and set the minimum total operation cost as the objective function, and construct a mechanism-data jointly driven optimization operation model, wherein the total operation cost includes the unit output cost in the normal operation state, and the network topology adjustment cost including the line thermal backup and bus sectioning; The solving module is used to input the IEEE standard system calculation data, solve the optimization operation model, and obtain a short circuit current limiting scheme including the total network topology adjustment means of line thermal backup, bus sectioning and unit thermal backup.

Images