A distributed energy power generation power prediction method and system based on federated learning

Abstract

A distributed energy power generation prediction method and system based on federated learning, the method is based on the federated learning framework, the global model of the power generation prediction of each distributed energy node in the distributed energy network is deployed, and the corresponding local model of the power generation prediction in each node cell is deployed, the global model parameters and the local model parameters are initialized, the first round of iteration is carried out, the local model parameters are updated according to the historical data of the distributed energy node, the updated local model parameters are aggregated, the global model parameters are updated according to the aggregation result, whether the global model parameters converge is judged, if not, the next round of iteration is carried out, if yes, the power generation prediction global model at this time is used to predict the power generation of the distributed energy node in a certain future time. The global model and the local model of the power generation prediction are respectively deployed, the local model parameters of the data cell with different privacy levels are aggregated, the data privacy safety of different cells is considered while the prediction effect of the global model is ensured.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of power systems, and particularly relates to a distributed energy power generation power prediction method and system based on federated learning. BACKGROUND

[0002] In recent years, distributed energy has become an important part of the power system. Distributed energy can provide considerable power supply by using natural energy, reduce carbon emissions of power grid generation, and help the power grid to transform to low carbon and green. The power generation power information of the distributed energy has important reference value for the planning, scheduling and maintenance of the power grid. By predicting the power generation power of each distributed energy node in advance, the manager can better formulate the power grid scheduling plan. Therefore, the prediction of the power generation power of the distributed energy has important significance.

[0003] With the development of artificial intelligence, using a deep neural network model for time series prediction has become the main means for predicting the power generation power of distributed energy. However, such a model has the defect that its training process usually needs to process a large amount of data, has a high requirement for computing power, and there is a problem of data privacy and security when collecting data. SUMMARY

[0004] The purpose of the present application is to provide a distributed energy power generation power prediction method and system based on federated learning to solve the above problems in the prior art.

[0005] To achieve the above purpose, the technical solutions of the present application are as follows:

[0006] In a first aspect, the present application provides a distributed energy power generation power prediction method based on federated learning, comprising:

[0007] S1, based on a federated learning framework, deploying a power generation power prediction global model of each distributed energy node in a distributed energy network and a corresponding power generation power prediction local model in each distributed energy node, and initializing the global model parameters and the local model parameters;

[0008] S2, performing a first round of iteration, updating the local model parameters of each power generation power prediction local model according to the historical data of the distributed energy node;

[0009] S3, aggregating the updated local model parameters of each power generation power prediction local model, and updating the global model parameters of the power generation power prediction global model according to the aggregation result;

[0010] S4, judging whether the global model parameters converge, if not, returning to step S2 for next round iteration; if yes, using the power generation at this time to predict the global model, and predicting the power generation of the distributed energy nodes within a certain time in the future.

[0011] The distributed energy node cells include low data privacy level cells and high data privacy level cells.

[0012] In the S2, for the low data privacy level cells, the update formula of the model parameters is:

[0013]

[0014] In the above formula, k represents the low data privacy level cell, w t+1，k is the model parameter of the kth cell in the t+1th iteration, η k is the parameter update step length of the kth cell, D k is the data sample amount in the kth cell, is the gradient of the model loss function of the kth cell, d n，k is the nth data sample of the kth cell;

[0015] For the high data privacy level cells, a random noise matrix is generated in each cell to disturb the update of the model parameters, and the update formula is:

[0016]

[0017] In the above formula, j represents the high data privacy level cell, is the model parameter of the jth cell after adding disturbance in the t+1th iteration, η j is the parameter update step length of the kth cell, D j is the data sample amount in the jth cell, is the gradient of the model loss function of the jth cell, d m，j is the mth data sample of the jth cell, w t，j is the model parameter of the jth cell in the tth iteration, n j is a random noise matrix.

[0018] The S3 includes:

[0019] S31, for the low data privacy level cells, the local model parameters of each cell are aggregated by using the following formula:

[0020]

[0021] In the above formula, K is the total number of the low data privacy level cells, w t+1，KD is the model parameter aggregated by all low-data-privacy-level cells after the t+1th iteration, K D is the data sample size of all low-data-privacy-level cells, k w is the data sample size in the kth cell, t+1，k is the model parameter of the kth cell in the t+1th iteration;

[0022] For high-data-privacy-level cells, the local model parameters of each cell are aggregated by using the following formula:

[0023]

[0024] In the above formula, J is the total number of high-data-privacy-level cells, w t+1，J D is the model parameter aggregated by all high-data-privacy-level cells after the t+1th iteration, J D is the data sample size of all high-data-privacy-level cells, j w is the data sample size in the jth cell, is the model parameter of the jth cell after adding disturbance in the t+1th iteration;

[0025] S32, according to the aggregation results of the local model parameters of the low-data-privacy-level cells and the high-data-privacy-level cells, the global model parameters of the power generation prediction global model are updated by using the following formula:

[0026]

[0027] In the above formula, G represents all cells divided by the distributed energy nodes, w t+1，G D is the model parameter of all distributed energy node cells after the t+1th iteration.

[0028] In a second aspect, the present application provides a distributed energy power generation prediction system based on federated learning, which comprises a model deployment module, a local model parameter updating module, a global model parameter updating module, and an iteration judgment module.

[0029] The model deployment module is used to deploy the power generation prediction global model of each distributed energy node in the distributed energy network and the corresponding power generation prediction local model in each distributed energy node cell based on the federated learning framework, and to initialize the global model parameters and the local model parameters;

[0030] The local model parameter updating module is used to perform the first iteration, and update the local model parameters of each power generation prediction local model according to the historical data of the distributed energy nodes;

[0031] The global model parameter updating module is configured to aggregate the local model parameters of the local power generation prediction models after being updated, and update the global model parameters of the global power generation prediction model according to the aggregation result;

[0032] The iteration judging module is configured to judge whether the global model parameters converge or not, if not, return to the local model parameter updating module for next round of iteration, and if yes, use the global power generation prediction model at this time to predict the power generation of the distributed energy nodes within a certain time in the future.

[0033] The local model parameter updating module comprises a low-data-privacy-level cell model parameter updating unit and a high-data-privacy-level cell model parameter updating unit.

[0034] The low-data-privacy-level cell model parameter updating unit is configured to update the model parameters of the low-data-privacy-level cell by using the following formula:

[0035]

[0036] In the above formula, k represents the low-data-privacy-level cell, w t+1，k is the model parameter of the kth cell in the t+1th iteration, η k is the parameter update step length of the kth cell, D k is the data sample amount in the kth cell, is the gradient of the model loss function of the kth cell, d n，k is the nth data sample of the kth cell;

[0037] The high-data-privacy-level cell model parameter updating unit is configured to generate a random noise matrix in each cell, and update the model parameters of the high-data-privacy-level cell by using the following formula:

[0038]

[0039] In the above formula, j represents the high-data-privacy-level cell, is the model parameter of the jth cell after adding disturbance in the t+1th iteration, η j is the parameter update step length of the kth cell, D j is the data sample amount in the jth cell, is the gradient of the model loss function of the jth cell, d m，j is the mth data sample of the jth cell, w t，j is the model parameter of the jth cell in the tth iteration, n j is the random noise matrix.

[0040] The global model parameter updating module comprises a low-data-privacy-level cell parameter aggregation unit, a high-data-privacy-level cell parameter aggregation unit and a global model parameter updating unit.

[0041] The low-data-privacy-level cell parameter aggregation unit is configured to aggregate the local model parameters of the low-data-privacy-level cells according to the following formula:

[0042]

[0043] In the above formula, K represents the total number of low-data-privacy-level cells, w t+1，K represents the model parameters aggregated by all low-data-privacy-level cells after the t+1th iteration, D K represents the data sample amount of all low-data-privacy-level cells, D k represents the data sample amount in the kth cell, w t+1，k represents the model parameters of the kth cell in the t+1th iteration;

[0044] The high-data-privacy-level cell parameter aggregation unit is configured to aggregate the local model parameters of the high-data-privacy-level cells according to the following formula:

[0045]

[0046] In the above formula, J represents the total number of high-data-privacy-level cells, w t+1，J represents the model parameters aggregated by all high-data-privacy-level cells after the t+1th iteration, D J represents the data sample amount of all high-data-privacy-level cells, D j represents the data sample amount in the jth cell, represents the model parameters of the jth cell after adding the disturbance in the t+1th iteration;

[0047] The global model parameter updating unit is configured to update the global model parameters of the power generation prediction global model according to the aggregation results of the local model parameters of the low-data-privacy-level cells and the high-data-privacy-level cells according to the following formula:

[0048]

[0049] In the above formula, G represents all cells divided by the distributed energy nodes, w t+1，G represents the model parameters of all distributed energy node cells after the t+1th iteration.

[0050] In a third aspect, the present application provides a distributed energy power generation prediction device based on federated learning, comprising a processor and a memory.

[0051] The memory is used to store computer program code and transmit the computer program code to the processor;

[0052] The processor is used to execute the preceding federated learning-based distributed energy power generation prediction method according to instructions in the computer program code.

[0053] In a fourth aspect, the present application provides a computer storage medium having a computer program stored thereon.

[0054] The computer program is executed by the processor to implement the steps of the preceding federated learning-based distributed energy power generation prediction method.

[0055] Compared with the prior art, the present application has the following beneficial effects:

[0056] 1. The present application proposes a federated learning-based distributed energy power generation prediction method and system. The method is based on a federated learning framework, deploys a global model for predicting power generation of each distributed energy node in a distributed energy network and a corresponding local model for predicting power generation in each distributed energy node cell, and initializes global model parameters and local model parameters. Then, the first iteration is performed, the local model parameters of each power generation prediction local model are updated according to the historical data of the distributed energy node, the aggregated local model parameters of each power generation prediction local model are updated, the global model parameters of the power generation prediction global model are updated according to the aggregation result, and finally it is determined whether the global model parameters converge. If not, the next iteration is performed, and the local model parameters of each power generation prediction local model are updated again. If the convergence is reached, the power generation of the distributed energy node within a certain time in the future is predicted using the power generation prediction global model at this time. The method is based on the federated learning framework, and the global model and the local model for predicting power generation of each distributed energy node are deployed in the distributed energy network. The global model parameters are updated by aggregating the updated local model parameters, and the characteristic data of each distributed energy node cell is fully utilized to make more accurate power generation prediction.

[0057] 2. The present application proposes a federated learning-based distributed energy power generation prediction method and system. When deploying the corresponding local model for predicting power generation in each distributed energy node cell, the method considers the privacy level requirements of different cell data in the distributed energy generation network, and aggregates the local model parameters of different privacy levels in a targeted manner. The data privacy requirements of different cells are considered while ensuring the prediction effect of the global model, and the data privacy security in different cells is protected. BRIEF DESCRIPTION OF DRAWINGS

[0058] Figure 1 The figure is a flowchart of the method of the present application.

[0059] Figure 2 Structure diagram of the system according to the application.

[0060] Figure 3 Structure diagram of the device according to the application. DETAILED DESCRIPTION

[0061] The application will be further described in detail below with reference to specific embodiments and drawings.

[0062] The application provides a distributed energy power generation power prediction method and system based on federated learning. Based on a federated learning framework, different community data privacy levels in a distributed energy generation network are considered, and a prediction model of distributed energy power generation is deployed. In the model, a low-data-privacy-level community directly uploads model parameters to a central server, and a high-data-privacy-level community uploads perturbed local model parameters to the central server. The central server receives and aggregates local model parameters of all communities to update a global model, thereby ensuring the prediction effect of the global model and taking into account the data privacy requirements of different communities.

[0063] Embodiment 1:

[0064] As shown in Figure 1 A distributed energy power generation power prediction method based on federated learning is performed in the following steps:

[0065] 1. Based on a federated learning framework, a global model for predicting power generation of each distributed energy node in a distributed energy network is deployed, as well as a corresponding local model for predicting power generation in each distributed energy node community, and global model parameters and local model parameters are initialized.

[0066] The global model for predicting power generation of distributed energy is a deep neural network model for sequence prediction. The input of the model is historical power generation data and historical environmental data of each distributed energy node, and the output is power generation of the distributed energy node within a certain time in the future. Global model parameters include hyperparameters such as network layer number and fully connected layer width in the model.

[0067] According to the geographical position and node type of all distributed energy nodes in the distributed energy network, such as photovoltaic nodes, wind turbine nodes and the like, the distributed energy nodes are divided into several cells G, and considering the privacy level requirements of different distributed energy node cell data, the distributed energy node cells are divided into low data privacy level cells and high data privacy level cells, wherein the low data privacy level cells are denoted as k=1,...,K, K is the total number of low data privacy level cells, the high data privacy level cells are denoted as j=1,...,J, J is the total number of high data privacy level cells, and a power generation prediction local model of the distributed energy node is deployed in each distributed energy node cell, initial local model parameters of the power generation prediction local model of each cell are obtained from a server, and the initial local model parameters are the same as the initialized global model parameters.

[0068] 2. A first round of iteration is performed, the local model parameters of each power generation prediction local model are updated according to the historical data of the distributed energy nodes, and the updated local model parameters of each cell are uploaded to the central server;

[0069] For the low data privacy level cells, in the kth cell, the update formula of the model parameters is:

[0070]

[0071] In the above formula, w t+1，k is the model parameter of the kth cell in the t+1th iteration, η k is the parameter update step length of the kth cell, D k is the data sample size in the kth cell, is the gradient of the model loss function of the kth cell, d n，k is the nth data sample of the kth cell;

[0072] For the high data privacy level cells, a random noise matrix n j is generated in each cell j to disturb the update of the model parameters, the noise matrix is subject to a Gaussian distribution, and the update formula is:

[0073]

[0074] In the above formula, j represents the high data privacy level cell, is the model parameter of the jth cell after adding the disturbance in the t+1th iteration, η j is the parameter update step length of the kth cell, D j is the data sample size in the jth cell, is the gradient of the model loss function of the jth cell, d m，j is the mth data sample of the jth cell, w t，jis the model parameter of the jth cell in the tth iteration, n j is a random noise matrix.

[0075] 3. The central server receives the local model parameters from each cell, aggregates the updated local model parameters of each power generation prediction local model, and updates the global model parameters of the power generation prediction global model according to the aggregation result;

[0076] For low-data-privacy-level cells, the local model parameters of each cell are directly uploaded to the central server, and the central server directly aggregates the model parameters of such cells:

[0077]

[0078] In the above formula, K is the total number of low-data-privacy-level cells, w t+1，K is the aggregated model parameter of all low-data-privacy-level cells after the t+1th iteration, D K is the data sample size of all low-data-privacy-level cells, D k is the data sample size in the kth cell, w t+1，k is the model parameter of the kth cell in the t+1th iteration;

[0079] For high-data-privacy-level cells, the local model parameters of each cell after adding disturbance are uploaded to the central server for parameter aggregation:

[0080]

[0081] In the above formula, J is the total number of high-data-privacy-level cells, w t+1，J is the aggregated model parameter of all high-data-privacy-level cells after the t+1th iteration, D J is the data sample size of all high-data-privacy-level cells, D j is the data sample size in the jth cell, is the model parameter of the jth cell after adding disturbance in the t+1th iteration;

[0082] The central server integrates the aggregation results of the local model parameters of low-data-privacy-level cells and high-data-privacy-level cells, and updates the global model parameters of the power generation prediction global model:

[0083]

[0084] In the above formula, G represents all cells divided by the distributed energy node, w t+1，G is the model parameter of all distributed energy node cells after the t+1th iteration.

[0085] 4. Determine whether the global model parameters have converged. If they have not converged, return to step 2 for the next iteration. If they have converged, use the current power generation to predict the global model and forecast the power generation of distributed energy nodes in the future.

[0086] Example 2:

[0087] like Figure 2 As shown, a distributed energy generation power prediction system based on federated learning includes a model deployment module, a local model parameter update module, a global model parameter update module, and an iterative judgment module.

[0088] The model deployment module is used to deploy a global model for predicting the power generation of each distributed energy node in the distributed energy network, as well as a local model for predicting the power generation within each distributed energy node cell, based on a federated learning framework, and to initialize the global model parameters and local model parameters.

[0089] The local model parameter update module is used to perform the first round of iteration, updating the local model parameters of each power generation prediction local model based on the historical data of the distributed energy nodes.

[0090] The global model parameter update module is used to aggregate the updated local model parameters of each power generation prediction local model, and update the global model parameters of the power generation prediction global model according to the aggregation result.

[0091] The iterative judgment module is used to determine whether the global model parameters have converged. If they have not converged, the module returns to the local model parameter update module for the next iteration. If they have converged, the current power generation is used to predict the global model and predict the power generation of the distributed energy nodes in the future within a certain period of time.

[0092] The local model parameter update module includes a low data privacy level cell model parameter update unit and a high data privacy level cell model parameter update unit.

[0093] The low data privacy level cell model parameter update unit is used to update the model parameters of the low data privacy level cell using the following formula:

[0094]

[0095] In the above formula, k represents a low data privacy level cell, and w t+1，k Let η be the model parameters for the k-th cell in the (t+1)th iteration. k For the parameter update step size of the k-th cell, D k Let K be the number of data samples in the k-th cell. Let d be the gradient of the loss function of the model in the k-th cell. n，kis the model parameter of the jth cell after adding disturbance in the tth iteration, and

[0096] The high-data-privacy-level cell model parameter updating unit is configured to generate a random noise matrix in each cell, and update the model parameter of the high-data-privacy-level cell by using the following formula:

[0097]

[0098] In the above formula, j represents the high-data-privacy-level cell, is the model parameter of the jth cell after adding disturbance in the tth iteration, and j is the parameter update step length of the kth cell, and D j is the data sample quantity in the jth cell, is the gradient of the model loss function of the jth cell, and d m，j is the mth data sample in the jth cell, and w t，j is the model parameter of the jth cell in the tth iteration, and n j is a random noise matrix.

[0099] The global model parameter updating module includes a low-data-privacy-level cell parameter aggregation unit, a high-data-privacy-level cell parameter aggregation unit, and a global model parameter updating unit.

[0100] The low-data-privacy-level cell parameter aggregation unit is configured to aggregate the local model parameters of the low-data-privacy-level cells by using the following formula:

[0101]

[0102] In the above formula, K is the total number of low-data-privacy-level cells, w t+1，K is the aggregated model parameter of all low-data-privacy-level cells after the tth iteration, and D K is the data sample quantity of all low-data-privacy-level cells, and D k is the data sample quantity in the kth cell, and w t+1，k is the model parameter of the kth cell in the tth iteration;

[0103] The high-data-privacy-level cell parameter aggregation unit is configured to aggregate the local model parameters of the high-data-privacy-level cells by using the following formula:

[0104]

[0105] In the above formula, J is the total number of high-data-privacy-level cells, w t+1，J is the aggregated model parameter of all high-data-privacy-level cells after the tth iteration, and D JD is the data sample size of all high data privacy level cells j D is the data sample size in the jth cell, is the model parameter of the jth cell after adding disturbance in the t+1th iteration;

[0106] The global model parameter updating unit is configured to update the global model parameter of the power generation prediction global model according to the aggregation result of the local model parameters of the low data privacy level cells and the high data privacy level cells, using the following formula:

[0107]

[0108] In the above formula, G represents all cells divided by the distributed energy node, w t+1，G is the model parameter of all distributed energy node cells after the t+1th iteration.

[0109] Embodiment 3:

[0110] As shown in Figure 3 A distributed energy power generation prediction device based on federated learning, comprising a processor and a memory;

[0111] The memory is configured to store computer program code and transmit the computer program code to the processor;

[0112] The processor is configured to execute the instructions in the computer program code to perform the method of embodiment 1.

[0113] Embodiment 4:

[0114] A computer storage medium having a computer program stored thereon;

[0115] The computer program is executed by the processor to implement the steps of the method of distributed energy power generation prediction based on federated learning.

Claims

1. A distributed energy generation power prediction method based on federated learning, characterized in that, The method includes: S1. Based on the federated learning framework, deploy a global model for predicting the power generation of each distributed energy node in the distributed energy network, as well as a local model for predicting the power generation within each distributed energy node cell, and initialize the global model parameters and local model parameters. The distributed energy node cells include cells with low data privacy level and cells with high data privacy level. S2. Perform the first iteration, updating the local model parameters of each power generation prediction local model based on historical data from distributed energy nodes, including: For cells with low data privacy levels, the update formula for their local model parameters is: In the above formula, k represents a low data privacy level cell, and w t+1，k Let η be the model parameters for the k-th cell in the (t+1)th iteration. k For the parameter update step size of the k-th cell, D k Let K be the number of data samples in the k-th cell. Let d be the gradient of the loss function of the model in the k-th cell. n，k This is the nth data sample from the kth cell; For cells with high data privacy levels, a random noise matrix is ​​generated within each cell to perturb the update of its local model parameters. The update formula is as follows: In the above formula, j represents a high-data-privacy-level cell. The model parameters η in the (t+1)th iteration after adding a perturbation to the j-th cell. j For the parameter update step size of the k-th cell, D j Let J be the number of data samples in the j-th cell. Let d be the gradient of the loss function of the model in the j-th cell. m,j For the m-th data sample in the j-th cell, w t,j Let n be the model parameters for the j-th cell in the t-th iteration. j It is a random noise matrix; S3. Aggregate the updated local model parameters of each power generation prediction local model, and update the global model parameters of the power generation prediction global model based on the aggregation results. S4. Determine whether the global model parameters have converged. If they have not converged, return to step S2 for the next iteration. If they have converged, use the current power generation to predict the global model and predict the power generation of distributed energy nodes in the future within a certain period of time.

2. The distributed energy generation power prediction method based on federated learning according to claim 1, characterized in that, S3 includes: S31. For cells with low data privacy levels, the following formula is used to aggregate the local model parameters of each cell: In the above formula, K represents the total number of low-data-privacy-level cells, and w t+1，K D represents the model parameters aggregated for all low data privacy level cells after the (t+1)th iteration. K D represents the data sample size for all low-data-privacy-level cells. k w represents the number of data samples in the k-th cell. t+1，k These are the model parameters for the k-th cell in the (t+1)th iteration; For cells with high data privacy requirements, the following formula is used to aggregate the local model parameters of each cell: In the above formula, J represents the total number of high-data-privacy-level cells, and w t+1，J D represents the model parameters aggregated for all high data privacy level cells after the (t+1)th iteration. J D represents the data sample size for all high data privacy level cells. j Let J be the number of data samples in the j-th cell. The model parameters in the (t+1)th iteration after adding a perturbation to the j-th cell; S32. Based on the aggregation results of the local model parameters of cells with low data privacy level and cells with high data privacy level, the global model parameters of the global model for power generation prediction are updated using the following formula: In the above formula, G represents all the cells divided by the distributed energy nodes, and w t+1，G These are the model parameters for all distributed energy node communities after the (t+1)th iteration.

3. A distributed energy generation power prediction system based on federated learning, characterized in that, The system includes a model deployment module, a local model parameter update module, a global model parameter update module, and an iterative judgment module; The model deployment module is used to deploy a global model for predicting the power generation of each distributed energy node in the distributed energy network, as well as a local model for predicting the power generation within each distributed energy node cell, based on a federated learning framework, and to initialize the global model parameters and local model parameters. The local model parameter update module is used to perform the first round of iteration, and updates the local model parameters of each power generation prediction local model according to the historical data of the distributed energy nodes, including a low data privacy level cell model parameter update unit and a high data privacy level cell model parameter update unit. The low data privacy level cell model parameter update unit is used to update the local model parameters of the low data privacy level cell using the following formula: In the above formula, k represents a low data privacy level cell, and w t+1，k Let η be the model parameters for the k-th cell in the (t+1)th iteration. k For the parameter update step size of the k-th cell, D k Let K be the number of data samples in the k-th cell. Let d be the gradient of the loss function of the model in the k-th cell. n，k This is the nth data sample from the kth cell; The high data privacy level cell model parameter update unit is used to generate a random noise matrix in each cell and update the local model parameters of the high data privacy level cell using the following formula: In the above formula, j represents a high-data-privacy-level cell. The model parameters η in the (t+1)th iteration after adding a perturbation to the j-th cell. j For the parameter update step size of the k-th cell, D j Let J be the number of data samples in the j-th cell. Let d be the gradient of the loss function of the model in the j-th cell. m，j For the m-th data sample in the j-th cell, w t，j Let n be the model parameters for the j-th cell in the t-th iteration. j It is a random noise matrix; The global model parameter update module is used to aggregate the updated local model parameters of each power generation prediction local model, and update the global model parameters of the power generation prediction global model according to the aggregation result. The iterative judgment module is used to determine whether the global model parameters have converged. If they have not converged, the module returns to the local model parameter update module for the next iteration. If they have converged, the current power generation is used to predict the global model and predict the power generation of the distributed energy nodes in the future within a certain period of time.

4. The distributed energy generation power prediction system based on federated learning according to claim 3, characterized in that, The global model parameter update module includes a low data privacy level cell parameter aggregation unit, a high data privacy level cell parameter aggregation unit, and a global model parameter update unit. The low data privacy level cell parameter aggregation unit is used to aggregate the local model parameters of low data privacy level cells using the following formula: In the above formula, K represents the total number of low-data-privacy-level cells, and w t+1，K D represents the model parameters aggregated for all low data privacy level cells after the (t+1)th iteration. K D represents the data sample size for all low-data-privacy-level cells. k w represents the number of data samples in the k-th cell. t+1,k These are the model parameters for the k-th cell in the (t+1)th iteration; The high data privacy level cell parameter aggregation unit is used to aggregate the local model parameters of high data privacy level cells using the following formula: In the above formula, J represents the total number of high-data-privacy-level cells, and w t+1,J D represents the model parameters aggregated for all high data privacy level cells after the (t+1)th iteration. J D represents the data sample size for all high data privacy level cells. j Let J be the number of data samples in the j-th cell. The model parameters in the (t+1)th iteration after adding a perturbation to the j-th cell; The global model parameter update unit is used to update the global model parameters of the power generation prediction global model based on the aggregation results of the local model parameters of low data privacy level cells and high data privacy level cells, using the following formula: In the above formula, G represents all the cells divided by the distributed energy nodes, and w t+1，G These are the model parameters for all distributed energy node communities after the (t+1)th iteration.

5. A distributed energy generation power prediction device based on federated learning, characterized in that, Including the processor and memory; The memory is used to store computer program code and to transmit the computer program code to the processor; The processor is configured to execute, according to instructions in the computer program code, a distributed energy generation power prediction method based on federated learning as described in any one of claims 1-2.

6. A computer storage medium storing a computer program thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the distributed energy generation power prediction method based on federated learning as described in any one of claims 1-2.

Images