Power grid peak shaving and frequency modulation method, system and device combined with virtual power plant

Abstract

The application discloses a power grid peak regulation and frequency modulation method, system and equipment combined with a virtual power plant, relates to the technical field of power grids, and comprises the following steps: obtaining heterogeneous regulation instructions of multiple peak regulation and frequency modulation units of the virtual power plant; performing multi-dimensional regulation intention feature vector extraction, identifying instruction correlation and generating a regulation instruction correlation graph; identifying instruction relationship type labels; fusing peak regulation and frequency modulation parameters of the heterogeneous regulation instructions to generate multiple fused regulation instructions; establishing a matching relationship between the fused regulation instructions and corresponding instruction regulation objects and sending the multiple fused regulation instructions to corresponding target regulation units. The application solves the technical problems that, in the prior art, heterogeneous regulation instructions lack a collaborative identification and relationship analysis mechanism under multi-agent parallel decision, leading to instruction repeated response, mutual cancellation and resource allocation conflict, and achieves the technical effects of improving the consistency of multi-source instruction fusion, the coordination of control behavior and the utilization rate of regulation resources in the peak regulation and frequency modulation process of the virtual power plant.

Description

Technical Field

[0001] This invention relates to the field of power grid related technologies, specifically to a power grid peak shaving and frequency regulation method, system, and equipment that incorporates a virtual power plant. Background Technology

[0002] Virtual power plants integrate heterogeneous resources such as distributed power sources (e.g., wind and solar), energy storage systems, controllable loads, and electric vehicle charging facilities through information communication and intelligent control, participating in ancillary services such as peak shaving and frequency regulation of the power grid as a whole. However, in the existing virtual power plant control architecture, multiple entities such as energy storage units, load units, new energy generation units, and charging units are often equipped with autonomous decision-making intelligent agents. When the grid side experiences regulation needs such as frequency deviation or power imbalance, each unit responds simultaneously and independently generates regulation commands based on local sensing information and local optimization objectives, issuing them to the corresponding execution units. Due to the lack of effective coordination and information sharing in the decision-making process among the agents, there are often complex correlations between the commands generated by different units. For example, multiple agents may issue similar commands for the same physical quantity regulation needs, leading to excessive occupation of regulation resources or repetitive actions by the execution mechanisms; some commands may have overlapping or interchangeable regulation effects, failing to achieve optimal resource allocation; and commands issued by different agents based on their respective optimization objectives may contradict each other in direction, causing the regulation effects to cancel each other out or even worsening the grid operation. Furthermore, centralized scheduling struggles to cope with the dynamic changes and communication delays of large-scale distributed resources, while hierarchical control lacks the ability to deeply explore the semantic and intent relationships between instructions, failing to accurately identify "duplication," "complementarity," "substitution," or "conflict," which seriously affects the response accuracy and resource utilization efficiency of peak shaving and frequency regulation.

[0003] Therefore, in the current related technologies, there is a technical problem that heterogeneous adjustment instructions under multi-agent parallel decision-making lack collaborative recognition and relationship parsing mechanisms, resulting in repeated instruction responses, mutual cancellation, and resource allocation conflicts. Summary of the Invention

[0004] This application provides a grid peak-shaving and frequency regulation method, system, and equipment that integrates a virtual power plant. It solves the technical problem in the prior art where the lack of collaborative identification and relationship resolution mechanism for heterogeneous regulation commands under multi-agent parallel decision-making leads to repeated command responses, mutual cancellation, and resource allocation conflicts. It achieves the technical effect of improving the consistency of multi-source command fusion, the coordination of control behavior, and the utilization rate of regulation resources during the peak-shaving and frequency regulation process of the virtual power plant.

[0005] This application provides a grid peak shaving and frequency regulation method combined with a virtual power plant. The method includes: acquiring multiple heterogeneous regulation commands generated by multiple peak shaving and frequency regulation units of a virtual power plant; extracting multi-dimensional regulation intention feature vectors from the multiple heterogeneous regulation commands; identifying the command association relationships of the multiple heterogeneous regulation commands based on the multi-dimensional regulation intention feature vectors; generating a regulation command association graph based on the command association relationships; identifying command relationship type labels of the multiple heterogeneous regulation commands according to the regulation command association graph, wherein the command relationship type labels include repetitive regulation relationships, complementary regulation relationships, substitution regulation relationships, and conflicting regulation relationships; fusing the peak shaving and frequency regulation parameters corresponding to the multiple heterogeneous regulation commands according to the command relationship type labels to generate multiple fused regulation commands; establishing a matching relationship between the multiple fused regulation commands and their corresponding command regulation objects; and sending the multiple fused regulation commands to the corresponding target regulation units based on the matching relationships.

[0006] This application also provides a power grid peak shaving and frequency regulation system combined with a virtual power plant. The system includes: a heterogeneous regulation command acquisition module, used to acquire multiple heterogeneous regulation commands generated by multiple peak shaving and frequency regulation units of a virtual power plant; a command association relationship identification module, used to extract multi-dimensional regulation intention feature vectors from the multiple heterogeneous regulation commands, identify the command association relationships of the multiple heterogeneous regulation commands based on the multi-dimensional regulation intention feature vectors, and generate a regulation command association diagram based on the command association relationships; a command relationship type label acquisition module, used to identify command relationship type labels of the multiple heterogeneous regulation commands according to the regulation command association diagram, wherein the command relationship type labels include repetitive regulation relationships, complementary regulation relationships, substitution regulation relationships, and conflicting regulation relationships; a fusion regulation command generation module, used to fuse the peak shaving and frequency regulation parameters corresponding to the multiple heterogeneous regulation commands according to the command relationship type labels, and generate multiple fusion regulation commands; and a matching relationship establishment module, used to establish matching relationships between the multiple fusion regulation commands and their corresponding command regulation objects, and send the multiple fusion regulation commands to the corresponding target regulation units based on the matching relationships.

[0007] This application also provides an electronic device, including: a memory for storing executable instructions; and a processor for implementing a grid peak shaving and frequency regulation method combined with a virtual power plant when executing the executable instructions stored in the memory.

[0008] This application proposes a method, system, and equipment for grid peak shaving and frequency regulation combined with a virtual power plant. The method acquires heterogeneous regulation commands from multiple peak shaving and frequency regulation units within a virtual power plant; extracts multi-dimensional regulation intent feature vectors, identifies command relationships, and generates a regulation command association graph; identifies command relationship type labels; fuses the peak shaving and frequency regulation parameters of the heterogeneous regulation commands to generate multiple fused regulation commands; establishes a matching relationship between the fused regulation commands and their corresponding regulation objects, and sends the multiple fused regulation commands to the corresponding target regulation units. This addresses the technical problem in existing technologies where the lack of collaborative identification and relationship resolution mechanisms for heterogeneous regulation commands under multi-agent parallel decision-making leads to duplicate command responses, mutual cancellation, and resource allocation conflicts. It achieves the technical effect of improving the consistency of multi-source command fusion, the coordination of control behavior, and the utilization rate of regulation resources during the peak shaving and frequency regulation process of a virtual power plant. Attached Figure Description

[0009] To more clearly illustrate the technical solutions of the embodiments of this disclosure, the accompanying drawings of the embodiments of this disclosure will be briefly described below. Flowcharts are used in this application to illustrate the operations performed by the system according to the embodiments of this application. It should be understood that the preceding or following operations are not necessarily performed precisely in sequence. Instead, various steps can be processed in reverse order or simultaneously as needed. Furthermore, other operations can be added to these processes, or one or more steps can be removed from these processes.

[0010] Figure 1 This is a schematic diagram of the power grid peak shaving and frequency regulation method combined with a virtual power plant provided in the embodiments of this application.

[0011] Figure 2 A schematic diagram of a power grid peak shaving and frequency regulation system combined with a virtual power plant, provided in an embodiment of this application.

[0012] Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.

[0013] Explanation of reference numerals in the attached figures: Heterogeneous adjustment instruction acquisition module 10, instruction association relationship identification module 20, instruction relationship type label acquisition module 30, fusion adjustment instruction generation module 40, matching relationship establishment module 50, input device 401, processor 402, memory 403, output device 404. Detailed Implementation

[0014] To further illustrate the technical means and effects adopted by the present invention in order to achieve the intended purpose, the following detailed description is provided in conjunction with the accompanying drawings and preferred embodiments, based on the specific implementation methods, structures, features and effects of the present invention.

[0015] This application provides a grid peak shaving and frequency regulation method that incorporates a virtual power plant, such as... Figure 1 As shown, the method includes: Step S100: Obtain multiple heterogeneous regulation commands generated by multiple peak shaving and frequency regulation units of the virtual power plant.

[0016] Preferably, the centralized coordination and control system of the virtual power plant receives regulation command data independently generated by multiple peak-shaving and frequency-regulating units in real time through a communication network. These units include energy storage regulation units, distributed renewable energy regulation units, flexible load regulation units, electric vehicle charging and discharging units, and industrial load response units. Specifically, the energy storage regulation unit is the local controller of the battery energy storage system, and its regulation command is a charging and discharging power command calculated based on the energy storage SOC state and grid frequency deviation. The distributed renewable energy regulation unit is the local controller of the wind power inverter and photovoltaic inverter, and its regulation command is based on the predicted power and actual power. The deviation is calculated, and the output power reduction or increase command is given; the flexible load adjustment unit is the load controller of the air conditioning group control system and the electric heating system. The adjustment command is the load increase or decrease command given after calculation based on the demand response target issued by the load aggregator; the electric vehicle charging and discharging unit is the charging pile controller or V2G terminal controller. The adjustment command is the charging power or discharging power command given after calculation based on the vehicle access status, battery power and electricity price signal; the industrial load response unit is the electrolytic cell controller, water pump frequency converter or compressor controller. The adjustment command is the short-term power adjustment command given after calculation based on the adjustment margin allowed by the production process.

[0017] Step S200: Extract multi-dimensional adjustment intention feature vectors from the multiple heterogeneous adjustment commands, identify the command association relationship of the multiple heterogeneous adjustment commands based on the multi-dimensional adjustment intention feature vectors, and generate an adjustment command association graph by combining the command association relationship.

[0018] Step S200 further includes the following: the plurality of peak shaving and frequency regulation units include energy storage regulation units, distributed new energy regulation units, flexible load regulation units, electric vehicle charging and discharging units, and industrial load response units; each heterogeneous regulation command includes power regulation amount, regulation direction, response time, duration, target of action, and area of ​​action information; the multidimensional regulation intention feature vector includes the power regulation feature, regulation direction feature, time response feature, regulation target feature, area of ​​influence feature, and resource status constraint feature extracted for each heterogeneous regulation command.

[0019] Preferably, the heterogeneous regulation command issued by each peak-shaving and frequency-regulating unit includes: power regulation amount, i.e., the power value to be increased or decreased by the executing device; regulation direction, i.e., the direction of power change, such as positive regulation or negative regulation; response time, i.e., the time delay required from the time the command is issued to the time the regulation unit actually begins to execute the power change; duration, i.e. the length of time the regulation action is required to be maintained; target of action, i.e. the grid regulation object targeted by the command, including specific voltage level, specific feeder number or specific substation / grid connection point number; and area of ​​action information, i.e. the geographical or electrical area covered by the command, including specific dispatching zone number, transformer area number or node number. Then, the heterogeneous regulation commands are standardized, encoded, and quantified to determine a multi-dimensional regulation intent feature vector, including the power regulation characteristics, regulation direction characteristics, time response characteristics, regulation target characteristics, affected area characteristics, and resource status constraint characteristics corresponding to each heterogeneous regulation command. Specifically, the power regulation amount is normalized according to the total regulation capacity of the virtual power plant to obtain a value between 0 and 1, where positive and negative indicate the direction of power increase or decrease, and the absolute value indicates the proportion of the regulation intensity of the command in the total capacity, thus determining the power regulation characteristics. The regulation direction is converted into discrete values, with positive regulation encoded as 1 and negative regulation encoded as -1. If the command includes bidirectional regulation capability, it is encoded as 0, thus determining the regulation direction characteristics.

[0020] Preferably, the response time is segmented and quantized according to the virtual power plant's preset response time levels (e.g., ultrafast response ≤100ms, fast response ≤1s, medium-speed response ≤10s, slow response >10s), and the corresponding level code values ​​are output, such as 1, 2, 3, and 4; the target type is discretely encoded, for example, energy storage converter is encoded as 001, wind power inverter as 002, photovoltaic inverter as 003, load aggregation controller as 004, electric vehicle charging pile as 005, and industrial equipment controller as 006. Determine the characteristics of the regulation target; discretize the area of ​​effect, mapping different partition numbers to corresponding area codes, such as area A 01, area B 02, area C 03, to determine the characteristics of the affected area; calculate the ratio of the duration to the maximum allowable continuous regulation time of the corresponding unit to obtain a constraint margin value between 0 and 1. The closer the value is to 1, the closer the instruction is to the equipment's capacity limit, to determine the resource state constraint characteristics. The maximum continuous regulation time is preset by the equipment's rated parameters, such as the maximum continuous discharge time of energy storage depending on the SOC and rated power.

[0021] Furthermore, step S200 also includes obtaining a first multidimensional adjustment intention feature vector and a second multidimensional adjustment intention feature vector corresponding to any two heterogeneous adjustment instructions; calculating the comprehensive intention correlation degree between the first multidimensional adjustment intention feature vector and the second multidimensional adjustment intention feature vector, wherein the comprehensive intention correlation degree is a weighted average of the intention correlation degrees of each adjustment intention feature vector; and determining the instruction association relationship between the two heterogeneous adjustment instructions based on the intention correlation degree.

[0022] Preferably, the command correlation refers to the existence of a quantifiable relationship when any two heterogeneous regulation commands exhibit functional correlation in dimensions such as regulation target, area of ​​action, regulation direction, and time response. Specifically, take any two heterogeneous regulation commands, denoted as regulation commands A and B, and extract their corresponding multi-dimensional regulation intention feature vectors. The feature vector of command A is [power regulation feature a1, regulation direction feature a2, time response feature a3, regulation target feature a4, area of ​​influence feature a5, resource state constraint feature a6], and the feature vector of command B is [power regulation feature b1, regulation direction feature b2, time response feature b3, regulation target feature b4, area of ​​influence feature b5, resource state constraint feature b6]. Calculate the correlation degree between regulation commands A and B in each of the six feature dimensions. For each dimension, obtain a correlation degree value between 0 and 1. Calculate the absolute value of the difference between a1 and b1, subtract this difference from 1, and normalize to obtain the power regulation feature correlation degree. The closer the value is to 1, the closer the regulation power amplitude required by the two commands is.

[0023] Preferably, the following steps are performed: First, determine whether a2 and b2 are both positive, both negative, one positive and one negative, or at least 0. If they are the same, assign a value of 1; if they are opposite, assign a value of 0; if they are bidirectional, assign a value of 0.5 to determine the correlation degree of the adjustment direction feature. Second, determine whether a3 and b3 have the same level code. If they are the same, assign a value of 1; if they are different, calculate the ratio of the level difference to the total number of levels and subtract it from 1 to determine the correlation degree of the time response feature. The closer the value is to 1, the closer the response speed level is. Third, determine whether a4 and b4 point to the same type of equipment code. If they are the same, assign a value of 1; if they are different, determine the correlation degree based on the equipment type. The upstream and downstream relationships are assigned a value between 0 and 1 to determine the correlation degree of the adjustment target feature; it is determined whether a5 and b5 belong to the same electrical zone code. If they are the same, they are assigned a value of 1, indicating that they act on the same electrical area. A decrease in the value indicates an increase in electrical distance. If they are different, they are assigned a value between 0 and 1 according to the relationship between adjacent zones to determine the correlation degree of the influence area feature; the absolute value of the difference between a6 and b6 is calculated, and the difference is subtracted from 1 to determine the correlation degree of the resource status constraint feature. The closer the value is to 1, the more consistent the utilization degree of the equipment resource constraints by the two instructions is.

[0024] Preferably, the correlation values ​​of the six dimensions are weighted and averaged to obtain the comprehensive intention correlation value. The comprehensive intention correlation value is calculated as follows: Comprehensive Intention Correlation Value = (w1 × Power Regulation Feature Correlation Value) + (w2 × Regulation Direction Feature Correlation Value) + (w3 × Time Response Feature Correlation Value) + (w4 × Regulation Target Feature Correlation Value) + (w5 × Influence Area Feature Correlation Value) + (w6 × Resource State Constraint Feature Correlation Value). Here, w1 to w6 are preset weight coefficients, pre-set according to the importance of different feature dimensions in the actual operation of the virtual power plant, and the sum of the six weights equals 1. This value is then compared with a preset correlation threshold (e.g., 0.6). If the comprehensive intention correlation value is greater than or equal to the threshold, it is determined that there is an instruction correlation relationship between the two instructions; if the comprehensive intention correlation value is less than the threshold, it is determined that there is no instruction correlation relationship between the two instructions. Then, each heterogeneous regulation instruction is treated as a node, and an edge is established between any two corresponding nodes based on the correlation relationship between them. The edge carries a weight value, i.e., the comprehensive intention correlation value, ultimately constructing a regulation instruction correlation graph.

[0025] Step S300: Identify the instruction relationship type labels of the plurality of heterogeneous adjustment instructions according to the adjustment instruction association diagram. The instruction relationship type labels include repetitive adjustment relationship, complementary adjustment relationship, substitution adjustment relationship and conflicting adjustment relationship.

[0026] Step S300 further includes: inputting the adjustment instruction association graph into a graph association feature learning model; iteratively propagating the node features and edge weights in the adjustment instruction association graph according to the graph association feature learning model to obtain the association latent feature vector corresponding to each heterogeneous adjustment instruction; calculating the association latent feature vector by combining multiple preset rules of the graph association feature learning model to output the probability distribution of the relationship between different heterogeneous adjustment instructions; and determining the instruction relationship type label corresponding to the multiple heterogeneous adjustment instructions according to the relationship probability distribution.

[0027] Preferably, the input data structure of the graph association feature learning model consists of an adjacency matrix and a node feature matrix of the adjustment instruction association graph. The adjacency matrix is ​​generated based on the adjustment instruction association graph, with a dimension of N×N (N being the total number of heterogeneous adjustment instructions). The value in the i-th row and j-th column of the matrix represents the comprehensive intent association degree between instruction i and instruction j; if two instructions are not associated, the value is 0. The node feature matrix is ​​an N×6 matrix, with each row corresponding to the adjustment intent feature vector of one heterogeneous adjustment instruction. The graph association feature learning model employs a multi-layer graph convolutional network structure, including an input layer, multiple hidden layers, and an output layer. Each hidden layer performs graph convolution operations, and its core operation can be represented as: in, The node feature matrix for all instructions in layer l. The matrix obtained by adding self-loops (each node adds a connection to itself) to the adjacency matrix. for The degree matrix, whose diagonal elements are the sum of the number of edges connecting each node. Let be the learnable weight parameter matrix of the l-th layer. This is a non-linear activation function, such as the ReLU function. The graph association feature learning model receives the input adjacency matrix and node feature matrix and performs the first layer of graph convolution. Each node's feature vector is updated to a weighted sum of its own features and the features of all its neighboring nodes. The weights are determined by the edge association degree and the degree matrix. The updated feature vectors are then fed into the second and third layers of graph convolution for further iterative updates.

[0028] Preferably, the node features and edge weights in the regulation instruction association graph are iteratively propagated according to the graph association feature learning model. Each instruction node generates an association latent feature vector corresponding to each heterogeneous regulation instruction. After the first convolution layer, the features of each node are aggregated with the feature information of its directly adjacent instructions. After the second convolution layer, the features of each node are further aggregated with the feature information of instructions with a distance of two hops. After the Lth convolution layer, the features of each node are aggregated with the feature information of all instructions with a distance of L hops. The final association latent feature vector reflects the global context position of the instruction in the entire association graph. Specifically, it includes instructions that are directly related to the regulation instruction and the strength of the association, more distant instructions that are transitively related to the regulation instruction through indirect paths, the position of the instruction in the entire instruction group (whether it is the center or the edge), and the distance distribution of the instruction features and other instruction features in the abstract space.

[0029] Furthermore, step S300 also includes the following preset rules: when two heterogeneous adjustment commands have the same target, the same area of ​​action, and the same adjustment direction, the corresponding command relationship type label is determined to be a repetitive adjustment relationship; when two heterogeneous adjustment commands correspond to different adjustment directions and act together on the same adjustment target, the corresponding command relationship type label is determined to be a complementary adjustment relationship; when two heterogeneous adjustment commands independently achieve the same adjustment target, the corresponding command relationship type label is determined to be a substitute adjustment relationship; when two heterogeneous adjustment commands correspond to opposite adjustment directions, the corresponding command relationship type label is determined to be a conflicting adjustment relationship.

[0030] Preferably, the preset rules include four decision logic conditions, which are embedded as hard constraint rules in the final decision stage of classification calculation. Multiple preset rule data are shown in Table 1. Table 1 Preset Rule Data Table Rule 1 The two instructions have the same target, the same area of ​​effect, and the same direction of adjustment. Repeated adjustment relationship Rule 2 The two commands correspond to different adjustment directions but work together to achieve the same adjustment target. Complementary adjustment relationship Rule 3 The two instructions each achieve the same adjustment target independently. Substitutional Regulation Rule 4 The two commands correspond to opposite adjustment directions. Conflict mediation relationship Preferably, the associated latent feature vectors are calculated using multiple preset rules. Specifically, the associated latent feature vector of each instruction is input into a classification layer, which is a fully connected network. The classification layer outputs a four-dimensional probability vector, where the sum of the four values ​​is 1, representing the probability that the instruction and another instruction belong to one of four relationship types. Simultaneously, four preset rules are used to logically judge the features of the two instructions, such as the target encoding, the area of ​​action encoding, and the adjustment direction encoding. A four-dimensional rule judgment vector is output and normalized to form a four-dimensional probability distribution. The dimension corresponding to a matching rule is marked as 1, and those that do not match are marked as... The predicted probability vector is set to 0; then, the predicted probability vector and the rule determination vector are weighted and fused, where the weight coefficients are used to adjust the proportion of model prediction and rule constraint respectively, such as 0.4 and 0.6; finally, the probability distribution of the relationship between different heterogeneous regulation commands is output, representing the probability of belonging to the preset rule relationship after comprehensive judgment based on model prediction and rule constraint. For example, [0.85, 0.05, 0.07, 0.03] respectively indicate that the pair of commands has an 85% probability of belonging to the repetitive regulation relationship, a 5% probability of belonging to the complementary regulation relationship, a 7% probability of belonging to the substitution regulation relationship, and a 3% probability of belonging to the conflict regulation relationship. Finally, the maximum probability selection strategy is used to determine the command relationship type label, that is, the relationship type corresponding to the dimension with the largest value in the probability distribution is taken as the final command relationship type label of the pair of commands. All command pairs with edge connections in the regulation command association graph are executed one by one to obtain multiple corresponding command relationship type labels. This clarifies the repetitive, complementary, substitution, or conflict relationship between each pair of associated commands, thereby ensuring that the classification results are physically and logically consistent with the basic principles of power grid regulation.

[0031] Step S400: Based on the instruction relationship type label, the peak-shaving and frequency-modulation parameters corresponding to the multiple heterogeneous adjustment instructions are fused to generate multiple fused adjustment instructions.

[0032] Step S400 further includes: dividing the plurality of heterogeneous regulation commands into a plurality of command relationship sets according to the command relationship type tags; extracting the peak-shaving and frequency-modulation parameters corresponding to each heterogeneous regulation command to generate a peak-shaving and frequency-modulation parameter set based on each command relationship set, wherein the peak-shaving and frequency-modulation parameters include regulation power amplitude parameters, regulation direction parameters, response time parameters, duration parameters, regulation target parameters, and regulation constraint parameters; and generating a plurality of fused regulation commands by fusing the peak-shaving and frequency-modulation parameter sets corresponding to different command relationship type tags.

[0033] Preferably, instruction pairs with the same relation type label are grouped into connected subgraphs. This involves using each instruction as a node and the relation type label as an edge category attribute, splitting the regulation instruction association graph into multiple connected subgraphs. Each connected subgraph corresponds to an instruction relation set, and the instructions within each set are interconnected through the same relation type label. These sets include sets of repeating regulation relations, complementary regulation relations, alternative regulation relations, and conflicting regulation relations. Each instruction relation set contains its number, relation type label members, and all regulation instructions and feature vector data. Then, peak-shaving and frequency-modulation parameters corresponding to heterogeneous regulation instructions are extracted from each instruction relation set. These parameters include regulation power amplitude parameters, regulation direction parameters, response time parameters, duration parameters, regulation target parameters, and regulation constraint parameters. For each instruction relation set, the peak-shaving and frequency-modulation parameters are summarized to obtain the peak-shaving and frequency-modulation parameter set for each instruction relation set.

[0034] Furthermore, step S400 also includes: performing superposition calculations on the peak-shaving and frequency-modulation parameter sets corresponding to repetitive regulation relationships to generate a first fusion regulation command; establishing collaborative regulation parameters based on the contribution relationship of each heterogeneous regulation command to the same regulation target for the peak-shaving and frequency-modulation parameter sets corresponding to complementary regulation relationships, and performing collaborative calculations based on the collaborative regulation parameters to generate a second fusion regulation command; performing optimal solution selection based on the regulation evaluation value corresponding to each heterogeneous regulation command for the peak-shaving and frequency-modulation parameter sets corresponding to conflicting regulation relationships to generate a fourth fusion regulation command; and combining the first, second, third, and fourth fusion regulation commands to generate multiple fusion regulation commands.

[0035] Preferably, for the peak-shaving and frequency-modulation parameters within each set of instruction relationships, fusion calculations are performed according to the association type to which the set belongs. Each set outputs a fusion adjustment instruction. Specifically, all instructions within the set of repeated adjustment relationships are superimposed to generate a first fusion adjustment instruction. For example, the power amplitude is algebraically superimposed but does not exceed the minimum power upper limit among all instruction constraints within the set. The consistent adjustment direction within the set is retained, the minimum response time parameter within the set is taken, the maximum duration parameter within the set is taken, the consistent adjustment target within the set is retained, and the intersection of all constraints within the set is taken. For all heterogeneous adjustment instructions within the set of complementary adjustment relationships, collaborative calculations are performed, including establishing collaborative adjustment parameters based on the contribution relationship of each heterogeneous adjustment instruction to the same adjustment target, and generating a second fusion adjustment instruction. For example, the power amplitude is calculated by weighting the different contributions of each instruction to the same adjustment target, retaining multiple adjustment direction information, taking the minimum response time parameter within the set, taking the maximum duration parameter within the set, unifying them into the same adjustment target, and executing them according to the constraints of each unit.

[0036] Preferably, all instructions within the set of alternative regulation relationships are screened for optimal solutions to generate a third fusion regulation instruction. Specifically, based on the normalized values ​​of power amplitude, response time, and duration, and the normalized value of resource consumption cost, the regulation evaluation value of each regulation instruction in the set is calculated. The instruction with the highest regulation evaluation value is selected as the preferred instruction of the set, and the parameters of the preferred instruction are directly output as the fusion regulation instruction. All instructions within the set of conflicting regulation relationships are conflict-resolved to generate a fourth fusion regulation instruction. This includes determining the regulation direction of all instructions in the set, identifying positive and negative instruction groups, calculating the sum of power of the positive group and the sum of power of the negative group, subtracting the smaller value from the larger value, and directing the remaining net power direction to the direction corresponding to the larger value. The response time is taken as the better value of the two groups, the duration is taken as the larger value of the two groups, the regulation target is retained as the original target, and the constraint conditions are taken as the intersection of the two sets of constraints. Finally, the first, second, third, and fourth fusion control commands are summarized to form multiple fusion control commands. Each fusion control command includes at least the command number, fusion type label, power amplitude, control direction, response time, duration, control target, constraint conditions, and the original heterogeneous control command, thereby avoiding the actuator receiving conflicting commands that would lead to ineffective control.

[0037] Step S500: Establish a matching relationship between the multiple fusion adjustment commands and their corresponding adjustment objects, and send the multiple fusion adjustment commands to the corresponding target adjustment units based on the matching relationship.

[0038] Step S500 further includes: obtaining object state parameters corresponding to multiple regulation objects in the virtual power plant; constructing a fusion regulation command feature set based on the regulation target parameters, regulation power amplitude parameters, response time parameters, and duration parameters corresponding to the multiple fusion regulation commands; performing matching calculations based on the fusion regulation command feature set and the object state parameters corresponding to the multiple regulation objects to obtain the regulation fit degree between each fusion regulation command and each regulation object; establishing a matching relationship between the multiple fusion regulation commands and the multiple regulation objects based on the regulation fit degree, wherein the matching relationship is used to determine the target regulation object corresponding to each fusion regulation command.

[0039] Preferably, a status query request is sent to the local controller of each regulated object through a communication network to collect real-time operating status data of all physical devices or device clusters that can participate in regulation in the virtual power plant. The regulated objects include energy storage devices, new energy power generation devices, flexible load devices, electric vehicle charging devices, and industrial load devices. Specifically, the current adjustable power limit is read in real time from the energy management system or local controller of the device; the current operating power is read in real time from the power sensor or smart meter at the grid connection point of the device; the current response capability is calculated by reading from the controller parameter configuration of the device and combining it with the current communication link status; the energy storage device is calculated based on the current SOC; the load device is determined based on the interruptible time limit set by the user; and the new energy device is predicted based on the current weather conditions, thereby determining the current sustainable regulation time. The current health status of the device is read from the device fault alarm system and online monitoring system (e.g., 0 unavailable, 1 available); and the current device occupancy status is read from the task queue status of the device controller (e.g., 0 idle, 1 occupied). Finally, object status parameters corresponding to multiple regulated objects are formed.

[0040] Preferably, feature parameters for matching calculation are extracted from each fusion adjustment command, including adjustment target parameters, adjustment power amplitude parameters, response time parameters, and duration parameters, and a feature set of fusion adjustment commands is constructed. The feature parameters of each fusion adjustment command are compared with the state parameters of each adjustment object one by one to obtain the adjustment fit degree, which represents the degree of suitability of the command to be executed on the object. Specifically, for each fusion adjustment command F and each adjustment object D, the dimensional fit degree is calculated from the target matching degree, power matching degree, response time matching degree, and duration matching degree. Among them, the adjustment target of F is compared with the electrical zone where D is located. If they are completely consistent, a value of 1 is assigned. If they are adjacent areas, a value of 0.5~0.9 is assigned according to the electrical distance. If they are unrelated, a value of 0 is assigned to determine the target matching degree. The ratio of the power amplitude of F to the current adjustable power limit of D is calculated. The value is between 0 and 1. The closer it is to 1, the closer the power demand is to the optimal adjustment point of the equipment and does not exceed the upper limit of the equipment capacity. The power matching degree is determined.

[0041] Preferably, the response time requirement of F is compared with the current response capability of D. If the response time of D is less than or equal to the response time of F, then this dimension is set to 1; if the response time of D is greater than the response time of F, then the value is set to 1. To determine the response time matching degree, the closer the value is to 1, the more likely the device can complete the action within the response time required by the command. Compare the duration requirement of F with the current sustainable adjustment time of D. If the sustainable duration of D is greater than or equal to the duration of F, the value is 1; otherwise, the value is less than the duration of F. The duration matching degree is determined, with a value closer to 1 indicating that the device can meet the continuous adjustment duration required by the instruction. Finally, the four dimensions of adaptability are weighted and summed, with a hard filter condition added: if the "Current Device Health Status" of an object's status parameter is 0 (unavailable) or "Current Device Occupancy Status" is 1 (occupied), then the adaptability of that object is directly assigned to 0. After calculating for each fused adjustment instruction and each adjustment object, an adaptability matrix is ​​generated, containing the adjustment adaptability between each fused adjustment instruction and each adjustment object.

[0042] Preferably, the optimal execution object is assigned to each fusion instruction based on the fit matrix, and the instruction is issued. Specifically, the element with the highest fit value is found in the fit matrix. Let this element be located in the i-th row and j-th column, indicating that the fusion instruction Fi and the adjustment object Dj have the highest fit. The adjustment object Dj is determined as the target adjustment object of the fusion instruction Fi, and a matching relationship is established. The i-th row and j-th column are deleted from the fit matrix, that is, the instruction has been assigned and the object has been occupied, avoiding duplicate assignment. The process is repeated in the remaining matrix until all fusion adjustment instructions are assigned to the target adjustment objects. If the maximum fit value in the remaining matrix is ​​lower than a preset lower threshold (e.g., 0.3), the instruction is not assigned and enters the waiting queue. Then, based on the matching relationship, each fusion adjustment instruction is encapsulated into an instruction message according to the communication protocol supported by the corresponding target adjustment object, and sent to the local controller of each target adjustment object through the communication network. After receiving the instruction, the local controller of the target adjustment object returns an acknowledgment signal. If no acknowledgment signal is received within the specified time or an execution rejection signal is received, the instruction is added back to the allocation queue, and the objects that have been confirmed as unexecutable are removed from the current fitness matrix. The matching calculation is then re-executed, and the instruction is allocated to the adjustment object with the second highest fitness.

[0043] In the above text, refer to Figure 1 A grid peak shaving and frequency regulation method incorporating a virtual power plant, according to embodiments of the present invention, is described in detail. Next, reference will be made to... Figure 2 A power grid peak shaving and frequency regulation system incorporating a virtual power plant is described according to an embodiment of the present invention.

[0044] The grid peak-shaving and frequency regulation system combining a virtual power plant according to embodiments of the present invention addresses the technical problems in existing technologies where heterogeneous regulation commands lack collaborative identification and relationship resolution mechanisms under multi-agent parallel decision-making, leading to duplicate command responses, mutual cancellation, and resource allocation conflicts. It achieves the technical effects of improving the consistency of multi-source command fusion, the coordination of control behavior, and the utilization rate of regulation resources during the peak-shaving and frequency regulation process of the virtual power plant. Figure 2 As shown, the power grid peak shaving and frequency regulation system combined with the virtual power plant includes: a heterogeneous regulation command acquisition module 10, a command association relationship identification module 20, a command relationship type tag acquisition module 30, a fusion regulation command generation module 40, and a matching relationship establishment module 50.

[0045] The heterogeneous regulation command acquisition module 10 is used to acquire multiple heterogeneous regulation commands generated by multiple peak-shaving and frequency-regulating units of a virtual power plant; the command association relationship identification module 20 is used to extract multi-dimensional regulation intention feature vectors from the multiple heterogeneous regulation commands, identify the command association relationships of the multiple heterogeneous regulation commands based on the multi-dimensional regulation intention feature vectors, and generate a regulation command association diagram based on the command association relationships; the command relationship type label acquisition module 30 is used to identify the command relationship type labels of the multiple heterogeneous regulation commands according to the regulation command association diagram, wherein the command relationship type labels include repeated regulation relationships, complementary regulation relationships, substitution regulation relationships, and conflicting regulation relationships; the fusion regulation command generation module 40 is used to fuse the peak-shaving and frequency-regulating parameters corresponding to the multiple heterogeneous regulation commands according to the command relationship type labels to generate multiple fusion regulation commands; the matching relationship establishment module 50 is used to establish matching relationships between the multiple fusion regulation commands and the corresponding command regulation objects, and send the multiple fusion regulation commands to the corresponding target regulation units based on the matching relationships.

[0046] The specific configuration of the instruction association identification module 20 will be described in detail below. The instruction association identification module 20 further includes: the plurality of peak shaving and frequency regulation units include energy storage regulation units, distributed new energy regulation units, flexible load regulation units, electric vehicle charging and discharging units, and industrial load response units; each heterogeneous regulation instruction includes power regulation amount, regulation direction, response time, duration, target, and area of ​​effect information; the multi-dimensional regulation intent feature vector includes the extracted power regulation features, regulation direction features, time response features, regulation target features, area of ​​effect features, and resource status constraint features corresponding to each heterogeneous regulation instruction.

[0047] The specific configuration of the instruction association recognition module 20 will be described in detail below. The instruction association recognition module 20 further includes: acquiring a first multidimensional adjustment intention feature vector and a second multidimensional adjustment intention feature vector corresponding to any two heterogeneous adjustment instructions; calculating a comprehensive intention correlation degree between the first multidimensional adjustment intention feature vector and the second multidimensional adjustment intention feature vector, wherein the comprehensive intention correlation degree is a weighted average of the intention correlation degrees of each adjustment intention feature vector; and determining the instruction association relationship between the two heterogeneous adjustment instructions based on the intention correlation degree.

[0048] The specific configuration of the instruction relationship type label acquisition module 30 will be described in detail below. The instruction relationship type label acquisition module 30 further includes: inputting the adjustment instruction association graph into a graph association feature learning model; iteratively propagating the node features and edge weights in the adjustment instruction association graph according to the graph association feature learning model to obtain the association latent feature vector corresponding to each heterogeneous adjustment instruction; calculating the association latent feature vector using multiple preset rules of the graph association feature learning model to output the probability distribution of the relationship between different heterogeneous adjustment instructions; and determining the instruction relationship type label corresponding to the multiple heterogeneous adjustment instructions based on the probability distribution of the relationship.

[0049] The specific configuration of the instruction relationship type label acquisition module 30 will be described in detail below. The instruction relationship type label acquisition module 30 further includes: when two heterogeneous adjustment instructions have the same target, the same area of ​​action, and the same adjustment direction, determining the corresponding instruction relationship type label as a repeating adjustment relationship; when two heterogeneous adjustment instructions correspond to different adjustment directions and act together on the same adjustment target, determining the corresponding instruction relationship type label as a complementary adjustment relationship; when two heterogeneous adjustment instructions independently achieve the same adjustment target, determining the corresponding instruction relationship type label as a substitute adjustment relationship; and when two heterogeneous adjustment instructions correspond to opposite adjustment directions, determining the corresponding instruction relationship type label as a conflicting adjustment relationship.

[0050] The specific configuration of the fusion regulation command generation module 40 will be described in detail below. The fusion regulation command generation module 40 further includes: dividing the plurality of heterogeneous regulation commands into multiple command relationship sets according to the command relationship type tags; extracting the peak-shaving and frequency-modulation parameters corresponding to each heterogeneous regulation command to generate a peak-shaving and frequency-modulation parameter set based on each command relationship set, wherein the peak-shaving and frequency-modulation parameters include regulation power amplitude parameters, regulation direction parameters, response time parameters, duration parameters, regulation target parameters, and regulation constraint parameters; and generating multiple fusion regulation commands by fusing the peak-shaving and frequency-modulation parameter sets corresponding to different command relationship type tags.

[0051] The specific configuration of the fusion regulation instruction generation module 40 will be described in detail below. The fusion regulation instruction generation module 40 further includes: performing superposition calculations on the peak-shaving and frequency-modulation parameter sets corresponding to repetitive regulation relationships to generate a first fusion regulation instruction; establishing collaborative regulation parameters based on the contribution relationship of each heterogeneous regulation instruction to the same regulation target for the peak-shaving and frequency-modulation parameter sets corresponding to complementary regulation relationships, and performing collaborative calculations based on the collaborative regulation parameters to generate a second fusion regulation instruction; performing optimal solution selection based on the regulation evaluation value corresponding to each heterogeneous regulation instruction for the peak-shaving and frequency-modulation parameter sets corresponding to alternative regulation relationships to generate a third fusion regulation instruction; resolving conflicts in the peak-shaving and frequency-modulation parameter sets corresponding to conflicting regulation relationships to generate a fourth fusion regulation instruction; and combining the first, second, third, and fourth fusion regulation instructions to generate multiple fusion regulation instructions.

[0052] The specific configuration of the matching relationship establishment module 50 will be described in detail below. The matching relationship establishment module 50 further includes: acquiring object state parameters corresponding to multiple regulation objects in the virtual power plant; constructing a fusion regulation command feature set based on the regulation target parameters, regulation power amplitude parameters, response time parameters, and duration parameters corresponding to the multiple fusion regulation commands; performing matching calculations based on the fusion regulation command feature set and the object state parameters corresponding to the multiple regulation objects to obtain the regulation fit degree between each fusion regulation command and each regulation object; and establishing a matching relationship between the multiple fusion regulation commands and the multiple regulation objects based on the regulation fit degree, wherein the matching relationship is used to determine the target regulation object corresponding to each fusion regulation command.

[0053] The grid peak shaving and frequency regulation system combined with a virtual power plant provided in the embodiments of the present invention can execute the grid peak shaving and frequency regulation method combined with a virtual power plant provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the method.

[0054] Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention, showing a block diagram of an exemplary electronic device suitable for implementing the embodiments of the present invention. Figure 3 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments of the present invention. This electronic device is in the form of a general-purpose computing device, and its components may include, but are not limited to, an input device 401, a processor 402, a memory 403, and an output device 404. The processor 402 may be one or more; the memory 403 may include a computer-readable medium and at least one program product having a set (at least one) of program modules configured to perform the functions of the embodiments of this application.

[0055] The memory 403 shown in this embodiment of the invention can be any combination of one or more computer-readable media. The computer-readable storage media can be, but is not limited to, infrared, semiconductor systems, devices or components, or any combination thereof, used to store software programs, computer-executable programs and modules, such as the program instructions / modules corresponding to the grid peak shaving and frequency regulation method combined with virtual power plants in this embodiment of the invention. The processor 402 executes various functional applications and data processing of the computer device by running the software programs, instructions and modules stored in the memory 403, thereby realizing the above-mentioned grid peak shaving and frequency regulation method combined with virtual power plants.

[0056] Although this application makes various references to certain modules in the system according to the embodiments of this application, any number of different modules can be used and run on user terminals and / or servers. The various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of each functional unit are only for easy distinction between each other and are not used to limit the scope of protection of this invention.

[0057] The specific embodiments described above do not constitute a limitation on the scope of protection of this application. Those skilled in the art should understand that various modifications, combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A grid peak-shaving and frequency regulation method combining virtual power plants, characterized in that, The method includes: Acquire multiple heterogeneous regulation commands generated by multiple peak-shaving and frequency-regulating units in a virtual power plant; Multidimensional adjustment intention feature vectors are extracted from the multiple heterogeneous adjustment commands. The command association relationship of the multiple heterogeneous adjustment commands is identified based on the multidimensional adjustment intention feature vectors. An adjustment command association graph is generated by combining the command association relationship. The instruction relationship type labels of the multiple heterogeneous adjustment instructions are identified according to the adjustment instruction association diagram. The instruction relationship type labels include repetitive adjustment relationship, complementary adjustment relationship, substitution adjustment relationship and conflicting adjustment relationship. Based on the instruction relationship type label, the peak-shaving and frequency-modulation parameters corresponding to the multiple heterogeneous adjustment instructions are fused to generate multiple fused adjustment instructions; Establish a matching relationship between the multiple fusion adjustment commands and their corresponding adjustment objects, and send the multiple fusion adjustment commands to the corresponding target adjustment units based on the matching relationship.

2. The grid peak shaving and frequency regulation method combining a virtual power plant as described in claim 1, characterized in that, The multiple peak-shaving and frequency regulation units include energy storage regulation units, distributed new energy regulation units, flexible load regulation units, electric vehicle charging and discharging units, and industrial load response units; Each heterogeneous regulation command includes information on power regulation amount, regulation direction, response time, duration, target, and area of ​​action; The multidimensional regulation intention feature vector includes the power regulation feature, regulation direction feature, time response feature, regulation target feature, influence area feature, and resource state constraint feature extracted for each heterogeneous regulation command.

3. The grid peak shaving and frequency regulation method combining a virtual power plant as described in claim 1, characterized in that, The method for identifying the instruction association relationship of the multiple heterogeneous adjustment instructions based on the multidimensional adjustment intention feature vector includes: Obtain the first multidimensional adjustment intention feature vector and the second multidimensional adjustment intention feature vector corresponding to any two of the heterogeneous adjustment instructions; Calculate the comprehensive intent correlation degree between the first multidimensional adjustment intent feature vector and the second multidimensional adjustment intent feature vector, wherein the comprehensive intent correlation degree is a weighted average of the intent correlation degrees of each adjustment intent feature vector; The instruction association relationship between the two heterogeneous adjustment instructions is determined based on the intent association degree.

4. The grid peak shaving and frequency regulation method combining a virtual power plant as described in claim 1, characterized in that, The method for identifying the instruction relationship type labels of the multiple heterogeneous adjustment instructions based on the adjustment instruction association diagram includes: The adjustment instruction association graph is input into the graph association feature learning model; Based on the graph association feature learning model, the node features and edge weights in the adjustment instruction association graph are iteratively propagated to obtain the association latent feature vector corresponding to each heterogeneous adjustment instruction; The graph association feature learning model is combined with multiple preset rules to calculate the association latent feature vectors and output the probability distribution of the relationship between different heterogeneous adjustment commands. The instruction relationship type label corresponding to the plurality of heterogeneous adjustment instructions is determined based on the relationship probability distribution.

5. The grid peak shaving and frequency regulation method combining a virtual power plant as described in claim 4, characterized in that, The multiple preset rules include: When two heterogeneous adjustment commands have the same target, the same area of ​​action, and the same adjustment direction, the corresponding command relationship type label is determined to be a repetitive adjustment relationship. When two heterogeneous adjustment commands correspond to different adjustment directions and act together on the same adjustment target, the corresponding command relationship type label is determined to be a complementary adjustment relationship. When two heterogeneous regulation commands independently achieve the same regulation objective, the corresponding command relationship type label is determined to be a substitute regulation relationship; When two heterogeneous adjustment commands correspond to opposite adjustment directions, the corresponding command relationship type label is determined to be a conflict adjustment relationship.

6. The grid peak shaving and frequency regulation method combining a virtual power plant as described in claim 1, characterized in that, The method involves fusing the peak-shaving and frequency-modulation parameters corresponding to the multiple heterogeneous adjustment commands based on the command relationship type label to generate multiple fused adjustment commands, including: Based on the instruction relationship type label, the multiple heterogeneous adjustment instructions are divided into multiple instruction relationship sets; Extract the peak shaving and frequency modulation parameters corresponding to each heterogeneous regulation command, and generate a peak shaving and frequency modulation parameter set based on each command relationship set. The peak shaving and frequency modulation parameters include regulation power amplitude parameters, regulation direction parameters, response time parameters, duration parameters, regulation target parameters, and regulation constraint parameters. By fusing the peak-shaving and frequency-modulation parameter sets corresponding to different instruction relationship type labels, multiple fused adjustment instructions are generated.

7. The grid peak shaving and frequency regulation method combining a virtual power plant as described in claim 6, characterized in that, Multiple fused adjustment commands are generated by fusing the peak-shaving and frequency-modulation parameter sets corresponding to different command relationship type labels. The method includes: The first fusion control command is generated by superimposing and calculating the set of peak-shaving and frequency-modulation parameters corresponding to the repeated adjustment relationship. For the set of peak-shaving and frequency-modulation parameters corresponding to complementary regulation relationships, collaborative regulation parameters are established based on the contribution relationship of each heterogeneous regulation command to the same regulation target. Based on the collaborative regulation parameters, collaborative calculations are performed to generate a second fusion regulation command. For the set of peak shaving and frequency modulation parameters corresponding to the alternative regulation relationship, the optimal solution is selected based on the regulation evaluation value corresponding to each heterogeneous regulation command, and a third fusion regulation command is generated. Conflict resolution is performed on the set of peak shaving and frequency modulation parameters corresponding to the conflict regulation relationship to generate the fourth fusion regulation command; By combining the first, second, third, and fourth fusion adjustment commands, multiple fusion adjustment commands are generated.

8. The grid peak shaving and frequency regulation method combining a virtual power plant as described in claim 1, characterized in that, The method for establishing the matching relationship between the multiple fusion adjustment commands and their corresponding adjustment objects includes: Obtain the object status parameters corresponding to multiple control objects in the virtual power plant; Based on the adjustment target parameters, adjustment power amplitude parameters, response time parameters, and duration parameters corresponding to the multiple fusion adjustment commands, a fusion adjustment command feature set is constructed; Based on the feature set of the fusion adjustment instruction and the object state parameters corresponding to the multiple adjustment objects, the adjustment adaptability between each fusion adjustment instruction and each adjustment object is obtained by matching and calculating. Based on the adjustment adaptability, a matching relationship is established between the multiple fusion adjustment commands and the multiple adjustment objects. The matching relationship is used to determine the target adjustment object corresponding to each fusion adjustment command.

9. A power grid peak-shaving and frequency regulation system combined with a virtual power plant, characterized in that: The system is used to implement the grid peak shaving and frequency regulation method combined with a virtual power plant as described in any one of claims 1 to 8, and the system includes: The heterogeneous regulation command acquisition module is used to acquire multiple heterogeneous regulation commands generated by multiple peak-shaving and frequency-regulating units of a virtual power plant. The instruction association recognition module is used to extract multi-dimensional adjustment intention feature vectors from the multiple heterogeneous adjustment instructions, identify the instruction association relationships of the multiple heterogeneous adjustment instructions based on the multi-dimensional adjustment intention feature vectors, and generate an adjustment instruction association graph by combining the instruction association relationships. The instruction relationship type label acquisition module is used to identify the instruction relationship type labels of the plurality of heterogeneous adjustment instructions according to the adjustment instruction association diagram. The instruction relationship type labels include repetitive adjustment relationship, complementary adjustment relationship, substitution adjustment relationship and conflict adjustment relationship. The fusion adjustment instruction generation module is used to fuse the peak-shaving and frequency-modulation parameters corresponding to the multiple heterogeneous adjustment instructions according to the instruction relationship type label, and generate multiple fusion adjustment instructions. The matching relationship establishment module is used to establish a matching relationship between the multiple fusion adjustment commands and the corresponding command adjustment objects, and send the multiple fusion adjustment commands to the corresponding target adjustment units based on the matching relationship.

10. An electronic device, characterized in that, The electronic device includes: Memory, used to store executable instructions; The processor, when executing executable instructions stored in the memory, implements the grid peak shaving and frequency regulation method incorporating a virtual power plant as described in any one of claims 1-8.

Images