A meta-reinforcement learning model migration training method and device based on topology transformation

Abstract

The present invention discloses a meta-reinforcement learning model migration training method and device based on topological transformation, which relates to the field of machine learning technology; acquiring power data of multiple regions and constructing a regional data set; comparing the data features of the regional data set with the model data set of a preset training model, and determining the update strategy of the preset training model according to the data features; migrating the data in the regional data set to the model data set, and performing an update operation according to the update strategy to obtain a target model. By collecting power data from multiple regions to construct a regional data set, and comparing the data features with the model data set of the preset training model to determine the model update strategy, the migration training method is effectively applied to model updating. This method not only accelerates the training process and improves model performance, but also reduces computing resource requirements, thereby achieving an efficient and economical model update method and enhancing the adaptability and accuracy of the model in power fault location.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of machine learning, and particularly relates to a meta-reinforcement learning model migration training method and device based on topology transformation. BACKGROUND

[0002] The meta-reinforcement learning model migration training method has a significant role in the current data-driven and intelligent background. By learning and migrating knowledge from multiple tasks, the model can quickly adapt to a new environment, optimize the decision-making process, and efficiently solve problems and strategy optimization in scientific research, industry, medical treatment and finance. This technology provides strong technical support for the development and application of the model.

[0003] Patent CN115800272A discloses a power grid fault analysis method, system, terminal and medium based on topology identification. After constructing a power grid topology model, the weight of abnormal electrical quantity is analyzed to predict a risk value, and associated topology nodes are extracted and their associated risk values are calculated. The cumulative risk value is obtained by superimposing the associated risk value and the predicted risk value. The node with the largest difference between the cumulative risk value and the threshold value is selected as the predicted fault node, and the selection is continuously performed until the cumulative risk value is lower than the threshold value. The model constructed in the prior art is often specific to a particular region, and the model does not have adaptability after being established, resulting in the need to rebuild the corresponding model for each scene, which increases the consumption of resources. SUMMARY

[0004] The purpose of the present application is to solve the problem that the model does not have adaptability after being established, resulting in the need to rebuild the corresponding model for each scene, and to propose a meta-reinforcement learning model migration training method and device based on topology transformation.

[0005] In the first aspect of the present application, a meta-reinforcement learning model migration training method based on topology transformation is proposed, which comprises:

[0006] Obtaining power data of multiple regions to construct a regional data set; the regional data set includes a training data set, a validation data set and a test data set; the power data includes power grid topology structure and power data of various devices;

[0007] Comparing the data features of the regional data set and the model data set of the preset training model, and determining the update strategy of the preset training model according to the data features;

[0008] Migrating the data in the regional data set to the model data set, and performing an update operation according to the update strategy to obtain a target model.

[0009] Optionally, comparing the data features of the regional data set and the model data set of the preset training model comprises:

[0010] extracting key features of the regional dataset and the model dataset; the key features include key node features in a power grid topology and power data features of devices;

[0011] calculating an absolute difference value of the regional dataset and the preset training model key features, and determining that the similarity of the regional dataset and the preset training model is high if the absolute difference value is less than a preset threshold value;

[0012] comparing a data amount of the regional dataset with a data amount threshold value, and determining that the data amount of the regional dataset is small if the data amount of the regional dataset is less than the data amount threshold value.

[0013] Optionally, the updating strategy of the preset training model according to the data features comprises:

[0014] if the data amount of the regional dataset is small and the similarity is high, determining that the updating strategy of the preset training model is a first updating mode;

[0015] freezing model parameters of convolution layers and pooling layers of the preset training model, unfreezing model parameters of full connection layers and classification layers of the preset training model, inputting data of the regional dataset into the preset training model for model training, and updating parameters of the full connection layers and the classification layers to obtain a target model.

[0016] Optionally, the updating strategy of the preset training model according to the data features comprises:

[0017] if the data amount of the regional dataset is small and the similarity is low, determining that the updating strategy of the preset training model is a second updating mode;

[0018] unfreezing model parameters of full connection layers of the preset model, inputting the regional dataset into the preset training model for model training, and updating parameters of the full connection layers to obtain a target model.

[0019] Optionally, the updating strategy of the preset training model according to the data features comprises:

[0020] if the data amount of the regional dataset is large and the similarity is high, determining that the updating strategy of the preset training model is a third updating mode;

[0021] unfreezing model parameters of each layer of the preset training model in a step-by-step incremental mode, inputting data of the regional dataset into the preset training model for incremental training, until parameters of each layer of the preset training model are updated to obtain a target model.

[0022] Optionally, determining the update strategy of the preset training model according to the data characteristics comprises:

[0023] If the data amount of the regional data set is large and the similarity is low, the update strategy of the preset training model is determined as the fourth update mode.

[0024] The parameters of the preset training model are initialized, the data in the regional data set is added to the model data set of the preset training model for model training, and the model parameters of the preset training model are updated to obtain a target model.

[0025] In the second aspect of the embodiment of the present application, a meta-reinforcement learning model migration training device based on topology transformation is provided, which comprises: the device comprises a data set construction module, a similarity comparison module and a model update module:

[0026] The data set construction module is configured to obtain power data of multiple regions and construct a regional data set; the regional data set comprises a training data set, a verification data set and a test data set; the power data comprises power data of a power grid topology structure and various devices;

[0027] The similarity comparison module is configured to compare data characteristics of the regional data set and a model data set of the preset training model, and determine an update strategy of the preset training model according to the data characteristics.

[0028] The model update module is configured to migrate data in the regional data set to the model data set, and perform an update operation according to the update strategy to obtain a target model.

[0029] The present application has the following advantages:

[0030] The present application provides a meta-reinforcement learning model migration training method based on topology transformation, which obtains power data of multiple regions and constructs a regional data set; compares data characteristics of the regional data set and a model data set of a preset training model, and determines an update strategy of the preset training model according to the data characteristics; migrates data in the regional data set to the model data set, and performs an update operation according to the update strategy to obtain a target model. The method effectively applies the migration training method to model updating by collecting power data of multiple regions to construct a regional data set, comparing data characteristics of the regional data set and a model data set of a preset training model, and determining a model update strategy. The method not only accelerates the training process, improves the model performance, but also reduces the demand for computing resources, thereby realizing an efficient and economical model updating method, and enhancing the adaptability and accuracy of the model in power fault positioning. BRIEF DESCRIPTION OF DRAWINGS

[0031] The present application will be further described below with reference to the accompanying drawings.

[0032] Figure 1 A flowchart of a meta-reinforcement learning model migration training method based on topology transformation is provided for the embodiments of the present application.

[0033] Figure 2 Another structural schematic diagram of a meta-reinforcement learning model migration training device based on topology transformation is provided for the embodiments of the present application. DETAILED DESCRIPTION

[0034] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments. The term "and / or" in this document is only used to describe the association relationship of the associated objects, which means that there can be three relationships, for example, A and B can represent the following three cases: A exists alone, A and B exist together, and B exists alone. In addition, the description of "first", "second" and the like in the present application is only for the purpose of description, and cannot be understood as indicating or implying the relative importance of the indicated technical features or implicitly indicating the number of the indicated technical features. Therefore, the features limited by "first", "second" can explicitly or implicitly include at least one of the features. In addition, the technical solutions of each embodiment can be combined with each other, but it must be based on the fact that a person skilled in the art can realize it. When the combination of technical solutions appears contradictory or unachievable, it should be considered that the combination of technical solutions does not exist, and is not within the scope of protection claimed by the present application.

[0035] Based on the embodiments in the present application, all other embodiments obtained by a person skilled in the art without creative labor are within the scope of protection of the present application.

[0036] The embodiments of the present application provide a meta-reinforcement learning model migration training method based on topology transformation. Referring to Figure 1 , Figure 1 A flowchart of a meta-reinforcement learning model migration training method based on topology transformation is provided for the embodiments of the present application. The method comprises the following steps:

[0037] S101, acquiring power data of multiple regions to construct a region data set.

[0038] S102, comparing the data features of the region data set with the model data set of the preset training model, and determining the update strategy of the preset training model according to the data features.

[0039] S103, migrating the data in the region data set to the model data set, and performing an update operation according to the update strategy to obtain a target model.

[0040] The regional data sets include: training data sets, validation data sets, and test data sets; the power data include: power grid topology and power data of various equipment.

[0041] Based on a topology transformation-based meta-reinforcement learning model transfer training method provided by an embodiment of the present invention, this method effectively applies the transfer training method to model updates by collecting multi-regional power data to construct regional datasets and comparing data features with the model dataset of a preset training model to determine the model update strategy. This method not only accelerates the training process and improves model performance, but also reduces computing resource requirements, thereby achieving an efficient and economical model update method and enhancing the model's adaptability and accuracy in power fault location.

[0042] In one implementation, the grid topology in the power data changes due to upgrades or other reasons, so the model needs to be upgraded based on the current situation. Model training through migration training can greatly reduce the amount of training, improve resource utilization, and reduce unnecessary costs.

[0043] In one implementation, the update strategy includes: a first update mode, a second update mode, a third update mode and a fourth update mode; wherein, if the data volume of the regional data set is small and the similarity is high, the mode is determined to be the first update mode; if the data volume of the regional data set is small and the similarity is low, the mode is determined to be the second update mode; if the data volume of the regional data set is large and the similarity is high, the mode is determined to be the third update mode; if the data volume of the regional data set is large and the similarity is low, the mode is determined to be the fourth update mode; the first update mode can reduce unnecessary training costs and enable the model to adapt to the new environment; the second update mode can enable the model to adapt to the new environment with a small amount of training cost; the third update mode adapts the model to the new data set through step-by-step training, and can also maintain the performance of the model on the original data set, thereby achieving better migration effects; the fourth update mode is large because the new data set is large and the differences are also large, and retraining the model can make full use of these rich data resources, thereby further improving the performance and accuracy of the model.

[0044] In one implementation, by comparing the data features of a regional dataset with the model dataset of a preset training model and determining the model update strategy accordingly, the model can be efficiently adjusted to better determine the update mode, thereby improving the accuracy of the model in the new environment without sacrificing the original performance, achieving more accurate grid fault location, and saving time and computing resources.

[0045] In one embodiment, step S102 further includes:

[0046] extracting key features of the regional data set and the model data set; the key features include: key node features in the power grid topology and power data features of the equipment.

[0047] calculating an absolute difference value of the key features of the regional data set and the preset training model, and determining that the similarity between the regional data set and the preset training model is high if the absolute difference value is less than a preset threshold.

[0048] comparing the data amount of the regional data set with a data amount threshold, and determining that the data amount of the regional data set is small if the data amount of the regional data set is less than the data amount threshold.

[0049] In an implementation manner, the absolute difference value of the key features of the regional data set and the model data set is calculated by using the cosine similarity, and this method is used to determine the similarity between the two, which can provide a quantitative index to evaluate the similarity between the two data sets. The advantage of this method is its intuitiveness and accuracy. By calculating the absolute difference value, the difference between the two data sets in the key features can be obtained, so that the updating strategy of the model can be quickly determined. Using the preset threshold as the judgment standard can ensure that only when the difference reaches a certain degree, the similarity difference between the two is large.

[0050] In an implementation manner, the key features of the regional data set and the model data set, especially the key node features in the power grid topology and the power data features of the equipment, are extracted, which can ensure the accuracy and effectiveness of the model updating.

[0051] In an embodiment, determining the updating strategy of the preset training model according to the data features further includes:

[0052] If the data amount of the regional data set is small and the similarity is high, the updating strategy of the preset training model is determined to be the first updating mode.

[0053] Freezing the model parameters of the convolutional layer and the pooling layer of the preset training model, unfreezing the model parameters of the fully connected layer and the classification layer of the preset training model, inputting the data of the regional data set into the preset training model for model training, and updating the parameters of the fully connected layer and the classification layer to obtain a target model.

[0054] In an implementation manner, freezing is to lock the model parameters of the layers in the model and prohibit the updating of the model parameters of each layer; and unfreezing is to unlock the model parameters of the layers in the model and allow the updating of the model parameters of each layer.

[0055] In one implementation, the transfer training process brings significant benefits when the model is updated in the first update mode. Because the regional dataset has a small amount of data but high similarity, the transfer training utilizes the knowledge in the pre-trained preset training model, i.e., the learned general feature representation. This feature representation can still be effective on the new task, thereby reducing the large amount of time and computing resources required to train the model from scratch. The transfer training can preserve the feature extraction ability learned by the preset training model on similar tasks by updating the parameters of the fully connected layer and the classification layer in the model, while enabling the model to adapt to the specific task of the new dataset. This adaptability enables the model to quickly converge on the new dataset, improves training efficiency, and often achieves better performance, reducing resource waste.

[0056] In one implementation, when only the parameters of the fully connected layer and the classification layer are updated, in the deep neural network, the convolutional layer and the pooling layer are mainly responsible for learning low-level feature representations of data, such as edges, textures, etc., which are often common between multiple tasks. The fully connected layer and the classification layer are more concerned with learning high-level feature representations and decision boundaries of data, which are closely related to specific tasks. In the first update mode, because the regional dataset has high similarity with the preset training model, it can be considered that the convolutional layer and the pooling layer have learned sufficient feature representations, and therefore do not need to be updated. The fully connected layer and the classification layer need to be fine-tuned according to the new dataset to adapt to the specific requirements of the new task. The benefit of this selection is that it can maintain the performance of the model on similar tasks while improving the adaptability and accuracy of the model on the new dataset. By updating only the parameters of the fully connected layer and the classification layer, the model can quickly and effectively complete the transfer training, save computing resources, and achieve better model performance.

[0057] In one embodiment, determining the update strategy of the preset training model according to the data characteristics further includes:

[0058] If the regional dataset has a small amount of data and low similarity, the update strategy of the preset training model is determined to be the second update mode.

[0059] Thaw the model parameters of the fully connected layer of the preset model, input the regional dataset into the preset training model for model training, and update the parameters for the fully connected layer to obtain the target model.

[0060] In one implementation, when the model is updated in the second update mode, due to the small amount of data and low similarity of the regional data set, training a new model from scratch is not only inefficient, but also may be difficult to capture enough feature information to adapt to the new task. Migration training allows experimenters to use the knowledge learned by the pre-set training model on a large amount of similar data, especially its ability in feature extraction and representation learning. By unfreezing and updating the fully connected layer of the pre-set model, the model is allowed to learn specifically for the new data set while maintaining these general feature representations. The benefit of this training process is that it can minimize the consumption of computing resources while ensuring the adaptability of the model to the new data set, improving training efficiency.

[0061] In one implementation, in the second update mode, the fully connected layer plays an important role in mapping the learned features to the final output in the neural network, and it is responsible for making the final decision or prediction based on the input feature representation. Since the regional data set has low similarity with the pre-set training model, the decision boundary and feature combination learned by the fully connected layer may need to be adjusted to adapt to the characteristics of the new data set. Therefore, by updating only the parameters of the fully connected layer, the model can be fine-tuned specifically for the new data set while maintaining the ability of the model in feature extraction. The benefit of this selection is that it can ensure the adaptability of the model to the new data set while avoiding large-scale modification of the overall structure of the model, thereby maintaining the stability and reliability of the model.

[0062] In one embodiment, determining the update strategy of the pre-set training model according to the data characteristics further comprises:

[0063] If the amount of data of the regional data set is large and the similarity is high, the update strategy of the pre-set training model is determined to be the third update mode.

[0064] The model parameters of each layer of the pre-set training model are unfrozen in a step-by-step incremental mode, and the data of the regional data set is input into the pre-set training model for incremental training until the parameter update of the parameters of each layer of the pre-set training model is completed, and the target model is obtained.

[0065] In one implementation, in the third update mode, when the data set of the regional data set is large and the similarity is high, the parameters of each layer of the pre-set training model are unfrozen and updated in a step-by-step incremental mode for migration training, which fine-tunes the model by making full use of a large amount of data to ensure that the model can achieve better performance on the new data set. By unfreezing the model layers in a step-by-step incremental manner, the feature information of the new data set can be gradually integrated into the model, allowing the model to gradually adapt to the distribution and pattern of the new data during the update process. This not only improves the accuracy of the model on the new data set, but also maintains the performance of the model on the original data set, achieving better migration effect.

[0066] In an implementation, in the third update mode, the reason for selecting each layer for incremental updating is that different network layers have different functions and importance in the model. By incrementally unfreezing and updating the parameters of each layer, it can be ensured that the model can gradually absorb the information of new data during the update process, and better adapt to the new data set on the basis of maintaining the original feature extraction capability. This incremental updating method can avoid the problem of model instability and performance degradation caused by unfreezing all layers at once. At the same time, the update order between each layer is determined according to their role and importance in the model, so as to ensure that the model can gradually optimize its performance during the update process, and finally obtain a target model that performs excellently on the new data set. This way of selecting each layer for incremental updating not only improves the training efficiency of the model, but also guarantees the generalization ability and accuracy of the model on the new data set.

[0067] In an embodiment, determining the update strategy of the preset training model according to the data characteristics further includes:

[0068] If the data amount of the regional data set is large and the similarity is low, the update strategy of the preset training model is determined as the fourth update mode.

[0069] The parameters of the preset training model are initialized, the data in the regional data set is added to the model data set of the preset training model for model training, and the model parameters of the preset training model are updated to obtain a target model.

[0070] In an implementation, in the fourth update mode, when the data set of the regional data set is large and the similarity is low, the parameters of the preset training model are initialized, and the data in the regional data set is added to the model data set for complete model training. This training process brings significant benefits. Because the data set is large, the model can learn more details and patterns from it, which helps to improve the generalization ability of the model on new tasks. Because the similarity of the data set is low, it is necessary to completely retrain the model to ensure that the model can adapt to the new data distribution and features. This transfer training process can fully utilize the information of a large amount of new data to deeply adjust and optimize the model, thereby obtaining a more accurate and powerful target model.

[0071] In one implementation, in the fourth update mode, the reason for selecting to re-update the entire model is that when the new data has a large difference with the original data in the feature representation, simply fine-tuning part of the parameters of the model can not be able to fully adapt to the characteristics of the new data. At this time, re-initializing the parameters of the model and adding the new data to the model data set for complete model training can ensure that the model can learn the feature representation of the new data from the beginning, so as to better adapt to the new task. The advantage of this way of re-updating the model is that it can break the inherent cognition of the original model to the data features, and re-learn the inherent laws and patterns of the new data, thereby obtaining a more flexible and more adaptable model. In addition, since the new data set is large, re-training the model can fully utilize these rich data resources to further improve the performance and accuracy of the model.

[0072] Based on the same inventive concept, the embodiment of the present application also provides a topological transformation-based meta-reinforcement learning model migration training device. Referring to Figure 2 , Figure 2 A topological transformation-based meta-reinforcement learning model migration training device provided by the embodiment of the present application has a structure diagram, which includes a data set construction module, a similarity comparison module, and a model update module.

[0073] The data set construction module is used to obtain power data of multiple regions and construct a regional data set; the regional data set includes a training data set, a verification data set, and a test data set; the power data includes power data of various devices and a power grid topology structure.

[0074] The similarity comparison module is used to compare data features of the regional data set and a model data set of a preset training model, and determine an update strategy of the preset training model according to the data features.

[0075] The model update module is used to migrate data in the regional data set to the model data set, and perform an update operation according to the update strategy to obtain a target model.

[0076] The topological transformation-based meta-reinforcement learning model migration training device provided by the embodiment of the present application collects power data of multiple regions to construct a regional data set, compares data features of the regional data set and a model data set of a preset training model, and determines a model update strategy, so that the migration training method is effectively applied to model updating. This method not only accelerates the training process, improves the model performance, but also reduces the demand for computing resources, thereby realizing an efficient and economical model updating method, and enhancing the adaptability and accuracy of the model in power fault positioning.

[0077] The above has been described in detail one embodiment of the present application, but the content is only the preferred embodiment of the present application, cannot be considered for limiting the scope of the present application. Any equivalent changes and improvements made in the scope of the present application, should still belong to the scope of the present application.

Claims

1. A meta-reinforcement learning model transfer training method based on topological transformation, characterized in that: Obtaining a preset training model, wherein the preset training model is used to identify power grid faults in a target area, the method comprising: Obtain power data from multiple regions and build regional datasets; The regional data set includes: a training data set, a verification data set, and a test data set; the power data includes: power grid topology and power data of various devices; Comparing data features of the regional dataset with the model dataset of the preset training model, and determining an update strategy for the preset training model based on the data features; Migrating the data in the regional dataset to the model dataset, and performing an update operation according to the update strategy to obtain a target model; By collecting multi-regional power data to build a regional dataset, and comparing data features with the model dataset of the preset training model, the model update strategy is determined, and the transfer training method is effectively applied to the model update; The update strategy includes: a first update mode, a second update mode, a third update mode, and a fourth update mode; wherein, if the data volume of the regional dataset is small and the similarity is high, the mode is determined to be the first update mode; if the data volume of the regional dataset is small and the similarity is low, the mode is determined to be the second update mode; if the data volume of the regional dataset is large and the similarity is high, the mode is determined to be the third update mode; and if the data volume of the regional dataset is large and the similarity is low, the mode is determined to be the fourth update mode; By comparing the data characteristics of the regional dataset with the model dataset of the preset training model, and determining the model update strategy accordingly, the model can be efficiently adjusted to better determine the update mode; Comparing data features of the regional dataset with the model dataset of the preset training model includes: Extracting key features of the regional dataset and the model dataset; the key features include: key node features in the power grid topology and power data features of the equipment; Calculating an absolute difference between the key features of the regional dataset and the preset training model; if the absolute difference is less than a preset threshold, determining that the regional dataset has a high similarity to the preset training model; comparing the data volume of the regional data set with a data volume threshold, and if the data volume of the regional data set is less than the data volume threshold, determining that the data volume of the regional data set is small; By calculating the absolute difference between the key features of the regional dataset and the model dataset through cosine similarity, and using this method to determine the similarity between the two, a quantitative indicator can be provided to evaluate the similarity between the two datasets. By calculating the absolute difference, the degree of difference between the two datasets in key features can be obtained, thereby quickly determining the model update strategy.

2. The meta-reinforcement learning model transfer training method based on topological transformation according to claim 1, characterized in that: Determining an update strategy for the preset training model according to the data characteristics includes: If the data volume of the regional data set is small and the similarity is high, determining that the update strategy of the preset training model is the first update mode; Freeze the model parameters of the convolution layer and pooling layer of the preset training model, unfreeze the model parameters of the fully connected layer and classification layer of the preset training model, input the data of the regional data set into the preset training model for model training, and update the parameters of the fully connected layer and the classification layer to obtain the target model.

3. The meta-reinforcement learning model transfer training method based on topological transformation according to claim 1, characterized in that: Determining an update strategy for the preset training model according to the data characteristics includes: If the data volume of the regional data set is small and the similarity is low, determining that the update strategy of the preset training model is the second update mode; Unfreeze the model parameters of the fully connected layer of the preset training model, input the regional data set into the preset training model for model training, and update the parameters of the fully connected layer to obtain a target model.

4. The meta-reinforcement learning model transfer training method based on topological transformation according to claim 1, characterized in that: Determining an update strategy for the preset training model according to the data characteristics includes: If the data volume of the regional data set is large and the similarity is high, determining that the update strategy of the preset training model is the third update mode; The model parameters of each layer in the preset training model are unfrozen in a step-by-step incremental mode, and the data of the regional data set is input into the preset training model for incremental training until the parameters of each layer in the preset training model are updated to obtain the target model.

5. The meta-reinforcement learning model transfer training method based on topological transformation according to claim 1, characterized in that: Determining an update strategy for the preset training model according to the data characteristics includes: If the data volume of the regional data set is large and the similarity is low, determining that the update strategy of the preset training model is the fourth update mode; The parameters of the preset training model are initialized, the data in the regional data set is added to the model data set of the preset training model for model training, and the model parameters of the preset training model are updated to obtain the target model.

6. A meta-reinforcement learning model transfer training device based on topological transformation, characterized in that: The device includes: a data set construction module, a similarity comparison module and a model update module: The data set construction module is used to obtain power data of multiple regions and construct regional data sets; the regional data sets include: training data sets, verification data sets, and test data sets; the power data includes: power grid topology and power data of various devices; The similarity comparison module is used to compare data features of the regional dataset with a model dataset of a preset training model, and determine an update strategy for the preset training model based on the data features; The model updating module is used to migrate the data in the regional dataset to the model dataset, and perform an update operation according to the update strategy to obtain a target model; By collecting multi-regional power data to build a regional dataset, and comparing data features with the model dataset of the preset training model, the model update strategy is determined, and the transfer training method is effectively applied to the model update; The update strategy includes: a first update mode, a second update mode, a third update mode, and a fourth update mode; wherein, if the data volume of the regional dataset is small and the similarity is high, the mode is determined to be the first update mode; if the data volume of the regional dataset is small and the similarity is low, the mode is determined to be the second update mode; if the data volume of the regional dataset is large and the similarity is high, the mode is determined to be the third update mode; and if the data volume of the regional dataset is large and the similarity is low, the mode is determined to be the fourth update mode; By comparing the data characteristics of the regional dataset with the model dataset of the preset training model, and determining the model update strategy accordingly, the model can be efficiently adjusted to better determine the update mode; The similarity comparison module is further configured to: Extracting key features of the regional dataset and the model dataset; the key features include: key node features in the power grid topology and power data features of the equipment; Calculating an absolute difference between the key features of the regional dataset and the preset training model; if the absolute difference is less than a preset threshold, determining that the regional dataset has a high similarity to the preset training model; comparing the data volume of the regional data set with a data volume threshold, and if the data volume of the regional data set is less than the data volume threshold, determining that the data volume of the regional data set is small; By calculating the absolute difference between the key features of the regional dataset and the model dataset through cosine similarity, and using this method to determine the similarity between the two, a quantitative indicator can be provided to evaluate the similarity between the two datasets. By calculating the absolute difference, the degree of difference between the two datasets in key features can be obtained, thereby quickly determining the model update strategy.

Images