A method for generating distributed collaborative renewable energy output scenarios

Abstract

The present invention belongs to the field of microgrid technology, and specifically relates to a distributed collaborative renewable energy output scenario generation method. The present invention establishes a spatial sharing matching network architecture model of the renewable energy output of an alliance microgrid based on each microgrid, collects information and constructs a latent space at the observation point of each microgrid through a microgrid data generation and processing center node, realizes latent space data sharing through a latent space information storage and transmission smart contract publishing node, and calculates the latent space information association matrix between microgrids through a latent space matching smart contract publishing node to calculate and obtain renewable energy output generation data, which can accurately describe the correlation of renewable energy output data in the time domain and the spatial domain.

Description

Technical Field

[0001] The present invention belongs to the technical field of microgrids, and in particular relates to a method for generating a renewable energy output scenario. Background Art

[0002] Microgrid is a new type of power system, which refers to an energy network composed of a variety of renewable energy sources such as wind power and photovoltaic power and other power equipment in a certain topological structure, which can operate independently or be interconnected with traditional energy networks. The output data generation technology of renewable energy sources such as wind power and photovoltaic power can provide data support for the planning of new energy stations and the optimization and scheduling of power systems, especially in dealing with the problem of insufficient data in extreme scenarios or low-probability events. The existing data generation method only considers the probability of generating data correlation in the time series under a single local sample, and the existing artificial intelligence generative algorithm cannot generate renewable energy output data whose time series scale and spatial scale are heterogeneous to the learning sample, resulting in the inability to accurately describe the correlation of renewable energy output data in the time series domain and the spatial domain. Summary of the invention

[0003] The technical problem to be solved by the present invention is to overcome the defect that the renewable energy output data generation method of the microgrid in the prior art cannot accurately describe the correlation of the renewable energy output data in the time domain and the spatial domain, and is difficult to meet the efficient operation requirements of the microgrid, thereby providing a distributed collaborative renewable energy output scenario generation method.

[0004] A method for generating a distributed collaborative renewable energy output scenario, comprising the following steps:

[0005] Step S1: Establish a spatial sharing matching network architecture model of the alliance microgrid renewable energy output based on each microgrid node, including a microgrid data generation and processing center node, a potential spatial information storage and transmission smart contract publishing node, and a potential spatial matching smart contract publishing node;

[0006] Step S2: Based on the spatial sharing matching network architecture model of the microgrid renewable energy output in step S1, the observation point information is collected through the microgrid data generation and processing center node, the observation point information is diffused to the spatial range, and the noise and the characteristic information of the microgrid are superimposed to obtain the final diffusion information;

[0007] Step S3: define the final diffusion information as a latent space, construct a latent space information storage and transmission smart contract publishing node to realize smart contract publishing and data upload and download of renewable energy output latent space information;

[0008] Step S4: Construct a renewable energy output information reverse diffusion generation node and a renewable energy output information reverse diffusion neural network on the node, download the latent space information storage and transmission smart contract publishing node, calculate the latent space information association matrix with other microgrids by the latent space matching smart contract publishing node, and finally calculate the renewable energy output generation data within the microgrid.

[0009] Furthermore, the microgrid data generation and processing center node includes a data generation information collection node, a scene natural information space association and expansion node, a renewable energy output information discrimination node, a renewable energy output information feature extraction node, a renewable energy output information diffusion node and a renewable energy output information reverse diffusion generation node.

[0010] Furthermore, the data generation information acquisition node includes scene natural information and device physical information; the scene natural information is a diffusion-generated information object, and the device physical information is used to assist the diffusion-generated information to be applicable to the local renewable energy model and output control;

[0011] The scene natural information space association and expansion node is embedded with an information space association and expansion algorithm;

[0012] The renewable energy output information discrimination node includes a renewable energy output information discrimination neural network, a reverse transfer parameter training algorithm of the renewable energy output information discrimination neural network, and a forward transfer judgment algorithm of the renewable energy output information discrimination neural network;

[0013] The renewable energy output information feature extraction node includes a renewable energy output information self-retrieval neural network, a renewable energy output information related neural network, a renewable energy output information feature final extraction neural network, a reverse transfer parameter training algorithm for the renewable energy output feature extraction process, and a forward transfer extraction algorithm for the renewable energy output feature extraction process;

[0014] The renewable energy output information diffusion node is embedded with a renewable energy output information algorithm;

[0015] The renewable energy output information reverse diffusion generation node includes a local renewable energy output information reverse diffusion neural network of the microgrid, a reverse transfer parameter training algorithm of the renewable energy output information reverse diffusion neural network, and a forward transfer information reverse diffusion algorithm of the renewable energy output information reverse diffusion neural network;

[0016] The potential spatial information storage and transmission smart contract publishing node includes a local renewable energy output potential spatial information upload smart contract publishing node, a joint renewable energy output potential spatial information download smart publishing node, a potential spatial information association matrix between microgrids, and a microgrid jointly stored renewable energy output potential spatial information;

[0017] The latent space matching smart contract release node embeds a latent space matching algorithm for renewable energy output.

[0018] Furthermore, step S2 includes the following method steps:

[0019] Step S2.1: The data generation information collection node collects microgrid internal information, including scene natural information and equipment physical information;

[0020] Step S2.2: The scene natural information spatial association and expansion node expands the scene natural information in the observation point and the microgrid internal information on it to a spatial distribution representation through the information spatial association and expansion algorithm, and obtains the diffuse wind energy scene natural information and the diffuse solar energy scene natural information on the diffuse point and on it;

[0021] Step S2.3: Extract the characteristic information of the microgrid through the renewable energy output information diffusion node, the renewable energy output information discrimination node and the renewable energy output information feature extraction node, superimpose the characteristic information and noise on the extended information to obtain the final diffusion information.

[0022] Furthermore, in step S2.2: the information space association and expansion algorithm includes the following steps:

[0023] Step S2.2.1: Determine the coordinate axes inside the microgrid space and divide the space;

[0024] Step S2.2.2: Obtain diffusion points in the spatial distribution according to the unit vector, and group all the diffusion points into a diffusion point set;

[0025] Step S2.2.3: Calculate the diffuse wind energy scene natural information and the diffuse solar energy scene natural information at each diffuse point in the diffuse point set;

[0026] Step S2.2.4: construct a three-dimensional space matrix based on the diffuse points and the diffuse wind energy scene natural information and the diffuse solar energy scene natural information thereon.

[0027] Furthermore, step S3 includes the following method steps:

[0028] Step S3.1: Potential spatial information storage and transmission The smart contract publishing node publishes and uploads the smart contract on the alliance chain. The microgrid completes the upload of the microgrid local information to the renewable energy output potential spatial information through the smart contract;

[0029] Step S3.2: constructing a latent spatial information association matrix between microgrids based on the latent spatial matching algorithm of renewable energy output;

[0030] Step S3.3: The smart contract release node for downloading the potential spatial information of the combined renewable energy output calculates the download information based on the potential spatial information association matrix between microgrids and microgrids.

[0031] Furthermore, the step S3.2 includes the following method steps:

[0032] Step S3.2.1: downloading the latent spatial information of the first microgrid from the renewable energy output latent spatial information stored jointly by the microgrids, and randomly sampling the latent spatial information of a second microgrid;

[0033] Step S3.2.2: Calculate the mean of each meta-information of the latent space information of the first microgrid and the second microgrid and construct a mean vector, calculate the covariance between different meta-information in each latent space information and construct a covariance matrix;

[0034] Step S3.2.3: Calculate the correlation value between the first microgrid and the second microgrid's latent spatial information;

[0035] Step S3.2.4: Calculate the correlation coefficient values ​​between the first microgrid and all other microgrids through the latent space matching algorithm of renewable energy output, and construct a latent space information association matrix between microgrids.

[0036] Furthermore, step S4 includes the following method steps:

[0037] Step S4.1: constructing a renewable energy output information reverse diffusion neural network of a renewable energy output information reverse diffusion generation node, and outputting prediction noise according to the latent space;

[0038] Step S4.2: constructing a reverse transfer parameter training algorithm for the reverse diffusion neural network of renewable energy output information, thereby updating the hyperparameter structure parameters of the reverse diffusion neural network of renewable energy output information;

[0039] Step S4.3: constructing a forward information back-diffusion algorithm for the back-diffusion neural network of renewable energy output information; based on the downloaded information and the natural information generation data and interference information calculated from other microgrids according to the predicted noise;

[0040] Step S4.4: adding interference information to the learning samples of the reverse transfer parameter training algorithm of the renewable energy output information discrimination node;

[0041] Step S4.5: Based on the constructed inverse diffusion neural network of renewable energy output information, inverse diffusion calculation is performed through the natural information generation data obtained in step S4.3, and finally the renewable energy output generation data of the microgrid is obtained.

[0042] Furthermore, the step S4.2 includes the following method steps:

[0043] Step S4.2.1: The characteristic information of the renewable energy output information diffusion algorithm of the renewable energy output information diffusion node is combined with the noise, the number of steps and the spatial diffusion information in the diffusion process to form the training set data;

[0044] Step S4.2.2: Based on the training set data, the prediction noise of the step is calculated by the inverse diffusion neural network of renewable energy output information;

[0045] Step S4.2.3: Calculate the loss function by MSE;

[0046] Step S4.2.4: Update the structural parameters of the inverse diffusion neural network of renewable energy output information according to the predicted noise and loss function.

[0047] Furthermore, the step S4.5 includes the following method steps:

[0048] Step S4.5.1: adding the spatial coordinate information of the data generation information collection node of the microgrid to the spatial distribution of the scene natural information generation data generated by the inverse diffusion calculation, and calculating the final collected scene natural information;

[0049] Step S4.5.2: Calculate the wind energy output generation data of the microgrid according to the finally collected scene natural information and the physical information of the equipment in the microgrid and the wind power generation equipment information;

[0050] Step S4.5.2: The solar energy output generation data of the microgrid is calculated based on the finally collected scene natural information, the physical information of the equipment in the microgrid, and the photovoltaic power generation equipment information.

[0051] Beneficial effect: The present invention discloses a distributed collaborative renewable energy output scenario generation method, establishes a spatial sharing matching network architecture model of the renewable energy output of an alliance microgrid based on each microgrid, collects information at the observation point of each microgrid and constructs a latent space through the microgrid data generation and processing center node, realizes latent space data sharing through the latent space information storage and transmission smart contract publishing node, and calculates the latent space information association matrix between microgrids through the latent space matching smart contract publishing node to calculate the renewable energy output generation data, which can accurately describe the correlation of renewable energy output data in the time domain and the spatial domain. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0053] Figure 1 The figure is a schematic block diagram of the method steps of the present invention. DETAILED DESCRIPTION

[0054] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are described in detail below in conjunction with the accompanying drawings. In the following description, many specific details are set forth to facilitate a full understanding of the present application. However, the present application can be implemented in many other ways different from those described herein, and those skilled in the art can make similar improvements without violating the connotation of the present application, so the present application is not limited by the specific embodiments disclosed below.

[0055] This embodiment provides a method for generating a distributed coordinated renewable energy output scenario, including the following steps:

[0056] Step S1: Establish a spatial sharing matching network architecture model for the output of renewable energy from the alliance microgrid based on each microgrid node , including microgrid data generation and processing center nodes , potential space information storage and transmission smart contract publishing node , Potential Space Matching Smart Contract Release Node ; wherein the spatial shared matching network architecture model includes n microgrids, Represents a finite set of central nodes for microgrid data generation and processing.

[0057] Step S2: Based on the spatial sharing matching network architecture model of the microgrid renewable energy output in step S1, the observation point information is collected through the microgrid data generation and processing center node, the observation point information is diffused to the spatial range, and the noise and the characteristic information of the microgrid are superimposed to obtain the final diffusion information; Step S3: Define the final diffusion information as a latent space, and construct a latent space information storage and transmission smart contract publishing node Realize the release of smart contracts and potential spatial information of renewable energy output Upload and download data;

[0058] Step S4: Constructing a renewable energy output information reverse diffusion generation node and a renewable energy output information reverse diffusion neural network on the node , through the potential space information storage and transmission smart contract release node Download the node published by the latent space matching smart contract The potential spatial information association matrix with other microgrids is calculated and finally the renewable energy output generation data is calculated within the microgrid.

[0059] This embodiment performs feature matching through the latent space of renewable energy scene output at multiple observation points to obtain a high-matching database suitable for inverse diffusion training of each microgrid and a low-matching dataset suitable for training the discriminant network of each microgrid, thereby associating the number of samples in multiple regions for one of the regions.

[0060] In the local reverse diffusion process, this embodiment uses the latent spatial information of renewable energy output scenarios from other microgrids. The latent spatial information obtains high-matching data through the latent spatial matching mechanism, and there are differences in spatiotemporal distribution, time scale and accuracy. Therefore, the present invention is based on the transformer model as the neural network structure of the reverse diffusion process and uses the local renewable energy output information (steps, time intervals, spatial scale) as the model training input to extract and denoise the noise, so as to reversely diffuse and obtain the spatial distribution of the natural information of the scene that conforms to the local renewable energy output. Furthermore, the renewable energy output scenario information that finally conforms to the local time series and spatial scale is obtained through the reverse spatial information expansion process and the local equipment physical information.

[0061] This embodiment locally collects natural scene information with a time scale, and obtains the spatial distribution characteristics of the natural scene information after collection through space. The algorithm is expanded on this basis, and the key features are extracted again based on the self-attention mechanism module and the data diffusion process is constructed together with noise. Finally, the latent space is constructed while considering the correlation between the spatiotemporal domain and the temporal domain, thereby effectively considering the correlation problem between the temporal domain and the spatial domain.

[0062] The microgrid data generation and processing center node Including data generation information collection node , scene natural information space association and expansion nodes , Renewable energy output information identification node , Renewable energy output information feature extraction node , Renewable energy output information diffusion node and the reverse diffusion of renewable energy output information to generate nodes .

[0063] The data generation information collection node Including natural scene information and device physical information ; The natural information of the scene Information objects generated for diffusion, device physical information The information generated by the auxiliary diffusion is applicable to local renewable energy models and output control; the scene natural information and device physical information Generate information collection nodes for the data The object of collection.

[0064] The scene natural information space association and expansion node Information Space Association and Expansion Algorithm ;

[0065] The renewable energy output information determination node Including renewable energy output information discrimination neural network , Backward Transfer Parameter Training Algorithm for Renewable Energy Output Information Discrimination Neural Network , Forward transfer discrimination algorithm of renewable energy output information discrimination neural network ;

[0066] The renewable energy output information feature extraction node Including renewable energy output information self-retrieval neural network , Neural Networks Related to Renewable Energy Output Information , Renewable energy output information feature extraction neural network , Backward Transfer Parameter Training Algorithm for Renewable Energy Output Feature Extraction Process , Forward transfer extraction algorithm for renewable energy output feature extraction process ;

[0067] The renewable energy output information diffusion node Embedding renewable energy output information algorithm ;

[0068] The renewable energy output information reverse diffusion generates a node Inverse diffusion neural network for local renewable energy output information including microgrids , Backward Transfer Parameter Training Algorithm for Reverse Diffusion Neural Network of Renewable Energy Output Information , Renewable energy output information reverse diffusion neural network forward transmission information reverse diffusion algorithm ;

[0069] The potential space information storage and transmission smart contract publishing node The potential spatial information including local renewable energy output is uploaded to the smart contract publishing node , combined with renewable energy output potential space information download intelligent release node , the potential spatial information association matrix between microgrids Potential spatial information of renewable energy output stored jointly with microgrids ;

[0070] The potential space matches the smart contract publishing node Latent spatial matching algorithm for embedding renewable energy output .

[0071] The step S2 comprises the following method steps:

[0072] Step S2.1: Data generation information collection node Collect microgrid internal information, including scene natural information and device physical information ;

[0073] Step S2.2: Spatial association of scene natural information and expansion of nodes The observation point And the scene natural information in the microgrid internal information Through information space association and expansion algorithm Expand to the spatial distribution representation and get the diffusion point And the diffuse wind energy scene natural information on it and spread solar scene natural information ;

[0074] Step S2.3: Diffusion of nodes through renewable energy output information , Renewable energy output information identification node and renewable energy output information feature extraction node , extract the characteristic information of microgrid , the feature information and the noise obtained by random sampling Overlay on extended information To obtain the final diffusion information .

[0075] In step S2.1:

[0076] Scene Natural Information Including natural information of wind energy scenarios , Renewable energy scenarios, natural information , Solar energy scene natural information ;

[0077] Wind energy scene natural information Including wind speed components in all directions at wind energy observation points (represents the wind speed components at the wind energy observation point in the north direction, i.e. the positive direction of the x-axis, the east direction, i.e. the positive direction of the y-axis, and the vertical direction to the top, i.e. the positive direction of the z-axis, in units of ), air density parameters at the collection point (Unit: ) and the relative spatial coordinates of the wind energy collection point (expressed as The unit is );

[0078] Renewable Energy Scenarios Natural Information Including air temperature (in degrees Celsius), air pressure (Unit: ), relative humidity (Unit: %), relative space coordinates (expressed as );

[0079] Solar energy scene natural information Including global radiation intensity (Unit: ), global radiation intensity Including direct solar radiation intensity , scattered radiation intensity , reflected radiation intensity ; The global solar radiation intensity Direct solar radiation intensity , scattered radiation intensity and reflected radiation intensity Superposition calculation, that is , thereby reducing the scattered radiation intensity affected by the atmospheric environment (humidity, temperature, air density, etc.) around the observation point The reflected radiation intensity affected by the buildings and terrain around the observation point Direct solar radiation intensity The separation of photovoltaic energy output data enables it to capture the data differences under different natural attributes in a more detailed manner.

[0080] Device physical information Including wind power equipment information and photovoltaic power generation equipment information ;

[0081] Wind power equipment information Including wind power equipment rated power (unit: kw), power factor of wind power generation equipment , Wind turbine rotor cross-sectional area (Unit: m2 ), horizontal rotation angle information of wind power generation equipment (Expressed as a three-dimensional vector ), wind speed limit information of wind power generation equipment ,in, and They are the upper and lower limits of wind speed for normal operation of wind power generation equipment. The maximum threshold wind speed for wind turbines to operate;

[0082] Photovoltaic power generation equipment information Including effective production capacity area of ​​power generation equipment (Unit: ) and photovoltaic power generation equipment power factor ;

[0083] In step S2.2: Information space association and expansion algorithm The following steps are involved:

[0084] Step S2.2.1: Determine the coordinate axes inside the microgrid space and divide the space;

[0085] Determine the unit vectors on the X, Y and Z axes inside the microgrid space And divide the space in this way, the calculation method is as follows:

[0086] ;

[0087] ;

[0088] ;

[0089] In the above formula, , and are the distances between the two observation points with the farthest spatial distribution on the X-axis, Y-axis and Z-axis among all observation points (in units of ), , and are the spatial scale of the microgrid space ( ), which is used to control the density of the diffusion points in each dimension, and is usually a positive integer.

[0090] Step S2.2.2: Obtain diffusion points in spatial distribution according to unit vectors , and all the diffusion points Composition of diffusion point set ; Where m is the number of the diffusion point.

[0091] According to the unit vector Get the diffusion points in the spatial distribution , expressed as , all diffusion points Composition of diffusion point set .

[0092] Step S2.2.3: Calculate the diffusion point set Each diffusion point Diffusion wind energy scenario natural information on and spread solar scene natural information ;

[0093] Assume that the extended information is calculated under the guarantee that the spatial distribution is continuous and uniform and there is no distortion point. Each diffusion point Diffusion wind energy scenario natural information on and spread solar scene natural information , calculated as follows:

[0094] ;

[0095] ;

[0096] ;

[0097] ;

[0098] In the above formula, Natural information for wind energy scenarios and solar scene natural information An abstract expression of any element of except for the elements in it). , and Diffusion points The components of the diffusion information on the X-axis, Y-axis, and Z-axis are calculated. For diffusion point The final diffusion information is calculated. Function expression satisfies conditional Element information. and Is to satisfy the distance diffusion point on the X-axis component The closest and smaller and larger than the diffusion point respectively X-axis coordinate of the observation point; and Is to satisfy the distance diffusion point on the Y axis component The closest and smaller and larger than the diffusion point respectively Y-axis coordinate of the observation point; and Is to satisfy the distance from the diffusion point on the Z axis component The closest and smaller and larger than the diffusion point respectively Z-axis coordinate of the observation point.

[0099] Step S2.2.4: Based on the diffusion point And the diffuse wind energy scene natural information on it and spread solar scene natural information Construct a three-dimensional space matrix.

[0100] Information space association and expansion algorithm The expanded information passes through all diffusion points Construct a three-dimensional space matrix of diffusion points, each point in the matrix is ​​a diffusion point The diffusion point information on each diffusion point is expressed as .

[0101] In step S2.3: Information space association and expansion algorithm The following steps are involved:

[0102] Step S2.3.1: Renewable energy output information identification node Neural network composition and its back-transfer parameter training algorithm and forward propagation discriminant algorithm ;

[0103] Renewable energy output information determination node Renewable Energy Output Information Discrimination Neural Network It consists of an input layer, two hidden layers and an output layer. Its neural network parameters and structure are as follows:

[0104] ;

[0105] in, and Represents neural network The weight parameters and bias parameters of the first hidden layer in , and They represent the neural network The weight parameters and bias parameters of the second hidden layer in . Representation Matrix The dimension is expressed as (i.e. row a and column b), Representation Matrix In addition, the neural network input step S2.2 obtained diffusion point information , the activation functions of the two hidden layers are Function, expressed as: , the output layer activation function is , expressed as: .

[0106] Renewable energy output information determination node Backward transfer parameter training algorithm in The process is as follows: First, by collecting scene natural information at each historical moment of the microgrid And calculate its extended information as learning samples; secondly, the training data set is used as input to discriminate the neural network through renewable energy output information Get an output, use the cross entropy loss function as the loss function to calculate the current error and reverse the gradient to modify the neural network The network parameters in , , and The formula is as follows:

[0107] ;

[0108] ;

[0109] ;

[0110] ;

[0111] ;

[0112] In the above formula, is the learning rate (a small positive value is taken in this embodiment); is the learning sample label (in this implementation, the value is 0 or 1); For neural networks The output classification result (in this embodiment, the value is 0 or 1); the classification process is expressed as:

[0113] ;

[0114] Renewable energy output information determination node Forward pass discriminant algorithm in A spread of information As input, through the neural network Output and classify to determine the diffusion information of the input Is it a microgrid? Natural information of the scene.

[0115] Step S2.3.2: Neural network composition of renewable energy output information feature extraction node and its reverse transfer parameter training algorithm and forward transfer feature extraction algorithm;

[0116] Renewable energy output information feature extraction node Self-retrieval neural network of renewable energy output information express , Renewable Energy Output Information Correlation Neural Network Expressed as , Renewable energy output information feature extraction neural network Expressed as .in, In the figure, the total number of elements in the diffusion point set is 13, which comes from each diffusion point. The amount of meta-information stored on ). In addition, the renewable energy output information self-retrieval neural network and renewable energy output information correlation neural network All are based on the extended information of step S2.3 As input, it outputs a Q matrix and a K matrix respectively. The mathematical expression is as follows:

[0117] ;

[0118] ;

[0119] Then the renewable energy output information features are finally extracted by neural network Take the probability distribution after the dot product of Q matrix and K matrix as input and output the final feature extraction information :

[0120] ;

[0121] In the above formula, The function compresses the values ​​of its elements to between 0 and 1 and makes their sum equal to 1.

[0122] Renewable energy output information feature extraction node Backward transfer parameter training algorithm in The process is as follows: First, by collecting scene natural information at each historical moment of the microgrid And calculate its extended information as a learning sample; secondly, through , and Propose feature information ; Again, the proposed feature information As a node for determining renewable energy output information Forward transfer discriminant algorithm Input, get a judgment result (without going through the classification process); finally, the mean square error ( ) as the loss function and reversely calculate the gradient, , and The parameters in are corrected, and the mathematical formula is as follows:

[0123] ;

[0124] ;

[0125] ;

[0126] ;

[0127] ;

[0128] ;

[0129] ;

[0130] Renewable energy output information feature extraction node Forward transfer feature extraction algorithm A spread of information As input, through the neural network , and The output is a dimension with diffusion information Consistent feature information .

[0131] Step S2.3.3: Renewable energy output information diffusion node and its renewable energy output information diffusion algorithm.

[0132] Renewable energy output information diffusion node Renewable energy output information diffusion algorithm The process is as follows:

[0133] First, expand the information As input, and Each piece of metadata is a unit, and normalization is performed on its spatial distribution. The mathematical formula is as follows:

[0134] ;

[0135] in, represents the normalized element data, represents the element data before normalization, Represents all diffusion points Upper element The sum of; secondly, the normalized extended information As a node for extracting renewable energy output information features Forward transfer feature extraction algorithm Input, extract a feature information ; Again, set up a set of auxiliary sequences And from the normal distribution A set of noise is randomly sampled from Make the noise matrix dimension the same as the extended information Consistent; finally, the feature information With noise Overlay on extended information The diffusion data of the first step is formed , the diffusion process is expressed as follows:

[0136] ;

[0137] ;

[0138] The above diffusion process is repeated After that, we get the final diffusion information And define it as the latent space.

[0139] The step S3 comprises the following method steps:

[0140] Step S3.1: Potential space information storage and transmission smart contract publishing node Publish and upload smart contracts on the alliance chain. The microgrid completes the upload of local information of the microgrid to the potential space information of renewable energy output through smart contracts. middle.

[0141] Specifically, the potential space information storage and transmission smart contract release node The potential spatial information of local renewable energy output in the smart contract publishing node is uploaded , publish an upload smart contract with information identification and storage functions on the alliance chain; the microgrid packs the potential space information into the block, and uploads the smart contract to obtain the potential space information in the block , and store it in the potential spatial information of renewable energy output within a certain period to complete the upload of local information of the microgrid; the local diffusion (loading) process and the automatic capture and signature mechanism of smart contracts in the blockchain can ensure the reliability of the information source while ensuring the privacy of the microgrid's own information.

[0142] Step S3.2: Potential spatial matching algorithm based on renewable energy output , construct the potential spatial information association matrix between microgrids , used to determine the download information.

[0143] The step S3.2 comprises the following method steps:

[0144] Step S3.2.1: Potential spatial information of renewable energy output stored in microgrids Download the latent space information of the first microgrid and randomly sample the latent space information of the second microgrid;

[0145] Potential spatial information of renewable energy output from microgrid joint storage Download to Micronet Latent space information ( ) and randomly sample microgrids Latent space information ( ).

[0146] Step S3.2.2: Calculate the mean of each meta-information of the latent space information of the first microgrid and the second microgrid and construct a mean vector, calculate the covariance between different meta-information in each latent space information and construct a covariance matrix;

[0147] Computing Microgrid Latent space information With Microgrid Latent space information The mean of each meta-information is used to construct the microgrid With Microgrid The mean vector of and , and the covariance between different meta-information within each latent space information And build a microgrid based on this With Microgrid The covariance matrix of and , the calculation formula is as follows:

[0148] ;

[0149] ;

[0150] ;

[0151] ;

[0152] ;

[0153] ;

[0154] ;

[0155] ;

[0156] In the above formula, and Microgrid With Microgrid Facial information The mean of and They represent microgrids respectively. With Microgrid Column vector of meta-information of ;

[0157] Step S3.2.3: Calculate the first microgrid and the second microgrid The correlation value of latent space information;

[0158] Calculate the correlation value of two latent spaces , the calculation formula is as follows:

[0159] ;

[0160] In the above formula, Represents the sum of the diagonal elements of a matrix (the trace of the matrix).

[0161] Step S3.2.4: Latent space matching algorithm for renewable energy output Computing the First Microgrid The correlation coefficient values ​​of all other microgrids are constructed to obtain the potential spatial information correlation matrix between microgrids ;

[0162] Latent space matching algorithm for renewable energy output Calculate the microgrid The correlation coefficient values ​​of all other microgrids are used to construct the potential spatial information correlation matrix between microgrids The nth column vector of , expressed as ,in is a finite set of microgrids, and the potential spatial information association matrix between microgrids is constructed .

[0163] Step S3.3: Download the smart contract release node of the potential spatial information of the combined renewable energy output According to the potential spatial information association matrix between microgrids and microgrids Calculate download information ;

[0164] Potential spatial information download smart contract release node for combined renewable energy output Will be based on the download object (based on micro-network as an example) and the potential spatial information association matrix between microgrids Column vector in information Confirm download information , expressed as ,in Represented as latent spatial information with micro-network A finite set of other microgrid latent space information with high similarity is expressed as

[0165] ,

[0166] in Represents the high similarity threshold, and the mathematical formula is as follows:

[0167] ;

[0168] In the above formula, Represents microgrid The high similarity potential space diffusion coefficient of ;akin, Represented as latent spatial information with micro-network The finite set of other microgrid latent space information with lower similarity is expressed as

[0169] ,

[0170] in Represents the low similarity threshold, and the mathematical formula is as follows:

[0171] ;

[0172] In the above formula, Represents microgrid The low similarity potential space diffusion coefficient is .

[0173] The step S4 comprises the following method steps:

[0174] Step S4.1: Construct the reverse diffusion generation node of renewable energy output information Reverse diffusion neural network of renewable energy output information , predicting the noise based on the latent space output ;

[0175] In this embodiment, the prediction noise is jointly output based on the potential space of each step in the local reverse diffusion process of microgrid n, the number of reverse diffusion steps, and the local guidance information of microgrid n (the detailed information of renewable energy output of the microgrid).

[0176] Renewable energy output information reverse diffusion generation node Reverse diffusion neural network of renewable energy output information The structure is a transformer model, including a self-attention layer (single-head self-attention) , two normalization layers And two fully connected layers , expressed as

[0177] .

[0178] Among them, the self-attention layer It consists of Q, K and V neural networks, expressed as . Each neural network contains a The matrix and a b vector are used as parameters of each neural network, so the inverse diffusion neural network of renewable energy output information The structural parameters can be expressed as:

[0179] ,

[0180] ;

[0181] Reverse diffusion neural network of renewable energy output information The structural parameters of , , , 1All the rest The structural parameters of ,all The structural parameters of Therefore, the inverse diffusion neural network of renewable energy output information The input of is the latent space, the denoising step, and the detailed information of renewable energy output of the microgrid, which is expressed as ,in To generate the time interval of the output data (1 minute / set, 15 minutes / set, 1 hour / set, etc.), is the spatial scale; the output is of dimension The prediction noise .

[0182] Step S4.2: Constructing a reverse diffusion neural network for renewable energy output information Backward transfer parameter training algorithm , thereby updating the renewable energy output information inverse diffusion neural network Structural parameters of

[0183] Step S4.3: Constructing a reverse diffusion neural network for renewable energy output information Forward information transfer reverse diffusion algorithm ; Based on download information And calculate the natural information generation data from other microgrid data based on the predicted noise and interference information ;

[0184] Reverse diffusion neural network of renewable energy output information Forward information transfer reverse diffusion algorithm The process is as follows: First, the microgrid in step S3 Download information , reverse diffusion step 、Microgrid The timing interval and spatial scale As a neural network The inverse diffusion step is obtained by The prediction noise at , and then the prediction noise is downloaded from the information Subtract from the original to get the inverse diffusion step The generated data at is as follows:

[0185] ;

[0186] Further, repeat the above formula and finally download the information , other microgrid data with high similarity to renewable energy natural information Used to generate renewable energy output data, and obtain natural information generation data that conforms to the local time series scale and spatial scale ; In addition, download information Other microgrid data with low similarity to renewable energy natural information will eventually get interference information .

[0187] Step S4.4: Backward transfer parameter training algorithm at renewable energy output information identification node Add interference information to the learning samples ;

[0188] Renewable energy output information determination node Backward transfer parameter training algorithm Add the interference information obtained by data expansion in step S4.3 to the learning sample , whose label .

[0189] Step S4.5: Reverse diffusion neural network based on the constructed renewable energy output information , the natural information generated data that conforms to the local temporal and spatial scale is obtained in step S4.3 Perform inverse diffusion calculations and finally obtain the microgrid Renewable energy output generation data.

[0190] The step S4.2 comprises the following method steps:

[0191] Step S4.2.1: The characteristic information of the renewable energy output information diffusion algorithm of the renewable energy output information diffusion node is combined with the noise, the number of steps and the spatial diffusion information in the diffusion process to form the training set data;

[0192] Reverse diffusion neural network of renewable energy output information Backward transfer parameter training algorithm The process is as follows: First, the renewable energy output information is spread to the nodes Renewable energy output information diffusion algorithm in Feature information used at each step in the diffusion process With noise Combination Steps and spatial diffusion information (without adding noise initial information) together constitute a set of data in the training set, expressed as ;

[0193] Step S4.2.2: Based on the training set data, the prediction noise of the step is calculated by the inverse diffusion neural network of renewable energy output information;

[0194] Steps obtained by reverse diffusion neural network calculation of renewable energy output information The prediction noise at , mathematical formula:

[0195] ;

[0196] ;

[0197] The above renewable energy output information reverse diffusion neural network The process unfolds as follows:

[0198] ;

[0199] ;

[0200] ;

[0201] ;

[0202] ;

[0203] ;

[0204] ;

[0205] ;

[0206] ;

[0207] in It means to concatenate tensors a, b, c, ... according to the first dimension.

[0208] Step S4.2.3: Calculate the loss function by MSE;

[0209] pass Calculate the loss function, the mathematical formula is:

[0210] ;

[0211] Step S4.2.4: Update the structural parameters of the inverse diffusion neural network of renewable energy output information according to the predicted noise and loss function.

[0212] Reverse diffusion neural network of renewable energy output information The structural parameters in are updated as follows:

[0213] ;

[0214] In the formula, Represented as an abstract representation of each structural parameter, that is, representing the set ,

[0215] All parameters in .

[0216] The step S4.5 comprises the following method steps:

[0217] Step S4.5.1: adding the spatial coordinate information of the data generation information collection node of the microgrid to the spatial distribution of the scene natural information generation data that meets the local time series scale and spatial scale generated by the inverse diffusion calculation, and calculating the final collected scene natural information;

[0218] The scene natural information generation data that conforms to the local temporal and spatial scale is generated by inverse diffusion calculation Adding microgrids to the spatial distribution of Data generation information collection node The spatial coordinate information is obtained, and the scene natural information at each collection point in the spatial distribution of the newly generated scene natural information is calculated as follows:

[0219] ;

[0220] ;

[0221] ;

[0222] ;

[0223] In the above formula, Natural information for wind energy scenarios and solar scene natural information An abstract expression of any element of except for the elements in it). , and The collection points The components of the diffusion information on the X-axis, Y-axis, and Z-axis are calculated. For collection point The final collected natural information of the scene is calculated.

[0224] Step S4.5.2: Calculate the wind energy output generation data of the microgrid according to the finally collected scene natural information and the physical information of the equipment in the microgrid and the wind power generation equipment information;

[0225] Microgrid Wind power generation data Finally, the natural information of the scene collected With Microgrid Physical information of equipment in the field Wind power equipment information Together we calculated:

[0226] ;

[0227] ;

[0228] ;

[0229] In the above formula, It is the angle between wind direction and blade plane, that is, only the wind speed component perpendicular to the blade plane will produce force.

[0230] Step S4.5.2: The solar energy output generation data of the microgrid is calculated based on the finally collected scene natural information, the physical information of the equipment in the microgrid, and the photovoltaic power generation equipment information.

[0231] Microgrid Solar power generation data Finally, the natural information of the scene collected With Microgrid Physical information of equipment in the field Photovoltaic power generation equipment information Together we calculated:

[0232] .

[0233] The technical features of the above-described embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0234] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be construed as limiting the scope of the patent application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent application shall be subject to the attached claims.

Claims

1. A method for generating a distributed collaborative renewable energy output scenario, characterized in that: The following steps are involved: Step S1: Establish a spatial sharing matching network architecture model of the alliance microgrid renewable energy output based on each microgrid node, including a microgrid data generation and processing center node, a potential spatial information storage and transmission smart contract publishing node, and a potential spatial matching smart contract publishing node; Step S2: Based on the spatial sharing matching network architecture model of the microgrid renewable energy output in step S1, the observation point information is collected through the microgrid data generation and processing center node, the observation point information is diffused to the spatial range, and the noise and the characteristic information of the microgrid are superimposed to obtain the final diffusion information; Step S3: define the final diffusion information as a latent space, construct a latent space information storage and transmission smart contract publishing node to realize smart contract publishing and data upload and download of renewable energy output latent space information; Step S4: construct a renewable energy output information reverse diffusion generation node and a renewable energy output information reverse diffusion neural network on the node, download the latent space information storage and transmission smart contract release node through the latent space matching smart contract release node to calculate the latent space information association matrix with other microgrids and finally calculate the renewable energy output generation data inside the microgrid; Step S4 includes: Step S4.1: constructing a renewable energy output information reverse diffusion neural network of a renewable energy output information reverse diffusion generation node, and outputting prediction noise according to the latent space; Step S4.2: constructing a reverse transfer parameter training algorithm for the reverse diffusion neural network of renewable energy output information, thereby updating the structural parameters of the reverse diffusion neural network of renewable energy output information; Step S4.3: constructing a forward information back-diffusion algorithm for the back-diffusion neural network of renewable energy output information; based on the downloaded information and the natural information generation data and interference information calculated from other microgrids according to the predicted noise; Step S4.4: adding interference information to the learning samples of the reverse transfer parameter training algorithm of the renewable energy output information discrimination node; Step S4.5: Based on the constructed inverse diffusion neural network of renewable energy output information, inverse diffusion calculation is performed through the natural information generation data obtained in step S4.3, and finally the renewable energy output generation data of the microgrid is obtained; Step S4.5 includes: Step S4.5.1: adding the spatial coordinate information of the data generation information collection node of the microgrid to the spatial distribution of the scene natural information generation data generated by the inverse diffusion calculation, and calculating the final collected scene natural information; Step S4.5.2: Calculate the wind energy output generation data of the microgrid according to the finally collected scene natural information and the physical information of the equipment in the microgrid and the wind power generation equipment information; Step S4.5.2: The solar energy output generation data of the microgrid is calculated based on the finally collected scene natural information, the physical information of the equipment in the microgrid, and the photovoltaic power generation equipment information.

2. A distributed collaborative renewable energy output scenario generation method according to claim 1, characterized in that: The microgrid data generation and processing center nodes include data generation information collection nodes, scene natural information space association and expansion nodes, renewable energy output information discrimination nodes, renewable energy output information feature extraction nodes, renewable energy output information diffusion nodes and renewable energy output information reverse diffusion generation nodes.

3. A distributed collaborative renewable energy output scenario generation method according to claim 2, characterized in that: The data generation information acquisition node includes scene natural information and device physical information; the scene natural information is a diffusion-generated information object, and the device physical information is used to assist the diffusion-generated information to be applicable to the local renewable energy model and output control; The scene natural information space association and expansion node is embedded with an information space association and expansion algorithm; The renewable energy output information discrimination node includes a renewable energy output information discrimination neural network, a reverse transfer parameter training algorithm of the renewable energy output information discrimination neural network, and a forward transfer judgment algorithm of the renewable energy output information discrimination neural network; The renewable energy output information feature extraction node includes a renewable energy output information self-retrieval neural network, a renewable energy output information related neural network, a renewable energy output information feature final extraction neural network, a reverse transfer parameter training algorithm for the renewable energy output feature extraction process, and a forward transfer extraction algorithm for the renewable energy output feature extraction process; The renewable energy output information diffusion node is embedded with a renewable energy output information algorithm; The renewable energy output information reverse diffusion generation node includes a local renewable energy output information reverse diffusion neural network of the microgrid, a reverse transfer parameter training algorithm of the renewable energy output information reverse diffusion neural network, and a forward transfer information reverse diffusion algorithm of the renewable energy output information reverse diffusion neural network; The potential spatial information storage and transmission smart contract publishing node includes a local renewable energy output potential spatial information upload smart contract publishing node, a joint renewable energy output potential spatial information download smart publishing node, a potential spatial information association matrix between microgrids, and a microgrid jointly stored renewable energy output potential spatial information; The latent space matching smart contract release node embeds a latent space matching algorithm for renewable energy output.

4. A distributed collaborative renewable energy output scenario generation method according to claim 3, characterized in that: The step S2 comprises the following method steps: Step S2.1: The data generation information collection node collects microgrid internal information, including scene natural information and equipment physical information; Step S2.2: The scene natural information spatial association and expansion node expands the scene natural information in the observation point and the microgrid internal information on it to a spatial distribution representation through the information spatial association and expansion algorithm, and obtains the diffuse wind energy scene natural information and the diffuse solar energy scene natural information on the diffuse point and on it; Step S2.3: Extract the characteristic information of the microgrid through the renewable energy output information diffusion node, the renewable energy output information discrimination node and the renewable energy output information feature extraction node, superimpose the characteristic information and noise on the extended information to obtain the final diffusion information.

5. A distributed collaborative renewable energy output scenario generation method according to claim 4, characterized in that: In step S2.2: the information space association and expansion algorithm includes the following steps: Step S2.2.1: Determine the coordinate axes inside the microgrid space and divide the space; Step S2.2.2: Obtain diffusion points in the spatial distribution according to the unit vector, and group all the diffusion points into a diffusion point set; Step S2.2.3: Calculate the diffuse wind energy scene natural information and the diffuse solar energy scene natural information at each diffuse point in the diffuse point set; Step S2.2.4: construct a three-dimensional space matrix based on the diffuse points and the diffuse wind energy scene natural information and the diffuse solar energy scene natural information thereon.

6. A distributed collaborative renewable energy output scenario generation method according to claim 3, characterized in that: The step S3 comprises the following method steps: Step S3.1: Potential spatial information storage and transmission The smart contract publishing node publishes and uploads the smart contract on the alliance chain. The microgrid completes the upload of the microgrid local information to the renewable energy output potential spatial information through the smart contract; Step S3.2: constructing a latent spatial information association matrix between microgrids based on the latent spatial matching algorithm of renewable energy output; Step S3.3: The smart contract release node for downloading the potential spatial information of the combined renewable energy output calculates the download information based on the potential spatial information association matrix between microgrids and microgrids.

7. A distributed collaborative renewable energy output scenario generation method according to claim 6, characterized in that: The step S3.2 comprises the following method steps: Step S3.2.1: downloading the latent spatial information of the first microgrid from the renewable energy output latent spatial information stored jointly by the microgrids, and randomly sampling the latent spatial information of a second microgrid; Step S3.2.2: Calculate the mean of each meta-information of the latent space information of the first microgrid and the second microgrid and construct a mean vector, calculate the covariance between different meta-information in each latent space information and construct a covariance matrix; Step S3.2.3: Calculate the correlation value between the first microgrid and the second microgrid's latent spatial information; Step S3.2.4: Calculate the correlation coefficient values ​​between the first microgrid and all other microgrids through the latent space matching algorithm of renewable energy output, and construct a latent space information association matrix between microgrids.

8. The method for generating a distributed collaborative renewable energy output scenario according to claim 1, characterized in that: The step S4.2 comprises the following method steps: Step S4.2.1: The characteristic information of the renewable energy output information diffusion algorithm of the renewable energy output information diffusion node is combined with the noise, the number of steps and the spatial diffusion information in the diffusion process to form the training set data; Step S4.2.2: Based on the training set data, the prediction noise of the step is calculated by the inverse diffusion neural network of renewable energy output information; Step S4.2.3: Calculate the loss function by MSE; Step S4.2.4: Update the structural parameters of the inverse diffusion neural network of renewable energy output information according to the predicted noise and loss function.

Images