Processing method, device and equipment of power distribution network bearing capacity evaluation model

Abstract

The invention relates to a processing method and device of a power distribution network bearing capacity evaluation model, computer equipment, a computer readable storage medium and a computer program product. The method comprises the steps that a training sample corresponding to the power distribution network and a corresponding label are acquired, the training sample comprises sample power grid index data representing the power quality of the power distribution network and sample inverter index data representing an inverter connected into the power distribution network, and the label is used for representing the evaluation grade of the power distribution network bearing capacity corresponding to the training sample; constructing an initial evaluation model based on a neural network, and obtaining a sample estimation grade based on the training sample through the initial evaluation model; and training the initial evaluation model based on the label and the sample estimation level, taking the trained initial evaluation model as a power distribution network bearing capacity evaluation model, and using the power distribution network bearing capacity evaluation model to estimate the power distribution network bearing capacity, thereby realizing accurate and effective evaluation of the bearing capacity of the power distribution network connected with the inverter.

Description

Technical Field

[0001] The present application relates to the technical field of distribution network, and in particular to a processing method, device, computer equipment, computer-readable storage medium and computer program product for a distribution network carrying capacity assessment model. Background Art

[0002] With the continuous development of distribution network technology, factors such as the widespread application of renewable energy, the construction of smart grids and the rapid development of distributed power generation systems have greatly promoted the integration of distributed energy in urban power grids, showing a blowout development trend. This trend not only enhances the flexibility and reliability of the power grid, but also brings new challenges, especially in terms of power quality.

[0003] As more and more inverters are connected to the urban power grid, the carrying capacity of the distribution network will be affected. However, since there is no analysis on the carrying capacity of the distribution network, it is difficult to evaluate the carrying capacity of the distribution network. Summary of the invention

[0004] Based on this, it is necessary to provide a processing method, device, computer equipment, computer-readable storage medium and computer program product for a distribution network carrying capacity assessment model that can effectively evaluate the distribution network carrying capacity in response to the above technical problems.

[0005] In a first aspect, the present application provides a method for processing a distribution network carrying capacity evaluation model, comprising:

[0006] Acquire training samples corresponding to the distribution network and corresponding labels, wherein the training samples include sample grid indicator data characterizing the power quality of the distribution network and sample inverter indicator data characterizing the inverters incorporated into the distribution network, and the labels are used to characterize the evaluation level of the distribution network carrying capacity corresponding to the training samples;

[0007] Constructing an initial evaluation model based on a neural network, and using the initial evaluation model to estimate the carrying capacity of the distribution network based on the training samples to obtain a sample estimation level;

[0008] Based on the labels and the sample estimated levels, the initial evaluation model is trained, and the trained initial evaluation model is used as a distribution network carrying capacity evaluation model, and the distribution network carrying capacity evaluation model is used to estimate the distribution network carrying capacity.

[0009] In a second aspect, the present application also provides a processing device for a distribution network carrying capacity evaluation model, comprising:

[0010] An acquisition module, used to acquire training samples corresponding to the distribution network and corresponding labels, wherein the training samples include sample grid indicator data representing the power quality of the distribution network and sample inverter indicator data representing the inverters incorporated into the distribution network, and the labels are used to represent the evaluation level of the distribution network carrying capacity corresponding to the training samples;

[0011] An estimation module is used to construct an initial estimation model based on a neural network, and to estimate the carrying capacity of the distribution network based on the training samples through the initial estimation model to obtain a sample estimation level;

[0012] A training module is used to train the initial evaluation model based on the label and the sample estimation level, and use the trained initial evaluation model as a distribution network carrying capacity evaluation model, wherein the distribution network carrying capacity evaluation model is used to estimate the distribution network carrying capacity.

[0013] In a third aspect, the present application further provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0014] Acquire training samples corresponding to the distribution network and corresponding labels, wherein the training samples include sample grid indicator data characterizing the power quality of the distribution network and sample inverter indicator data characterizing the inverters incorporated into the distribution network, and the labels are used to characterize the evaluation level of the distribution network carrying capacity corresponding to the training samples;

[0015] Constructing an initial evaluation model based on a neural network, and using the initial evaluation model to estimate the carrying capacity of the distribution network based on the training samples to obtain a sample estimation level;

[0016] Based on the labels and the sample estimated levels, the initial evaluation model is trained, and the trained initial evaluation model is used as a distribution network carrying capacity evaluation model, and the distribution network carrying capacity evaluation model is used to estimate the distribution network carrying capacity.

[0017] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the following steps are implemented:

[0018] Acquire training samples corresponding to the distribution network and corresponding labels, wherein the training samples include sample grid indicator data characterizing the power quality of the distribution network and sample inverter indicator data characterizing the inverters incorporated into the distribution network, and the labels are used to characterize the evaluation level of the distribution network carrying capacity corresponding to the training samples;

[0019] Constructing an initial evaluation model based on a neural network, and using the initial evaluation model to estimate the carrying capacity of the distribution network based on the training samples to obtain a sample estimation level;

[0020] Based on the labels and the sample estimated levels, the initial evaluation model is trained, and the trained initial evaluation model is used as a distribution network carrying capacity evaluation model, and the distribution network carrying capacity evaluation model is used to estimate the distribution network carrying capacity.

[0021] In a fifth aspect, the present application further provides a computer program product, including a computer program, which implements the following steps when executed by a processor:

[0022] Acquire training samples corresponding to the distribution network and corresponding labels, wherein the training samples include sample grid indicator data characterizing the power quality of the distribution network and sample inverter indicator data characterizing the inverters incorporated into the distribution network, and the labels are used to characterize the evaluation level of the distribution network carrying capacity corresponding to the training samples;

[0023] Constructing an initial evaluation model based on a neural network, and using the initial evaluation model to estimate the carrying capacity of the distribution network based on the training samples to obtain a sample estimation level;

[0024] Based on the labels and the sample estimated levels, the initial evaluation model is trained, and the trained initial evaluation model is used as a distribution network carrying capacity evaluation model, and the distribution network carrying capacity evaluation model is used to estimate the distribution network carrying capacity.

[0025] The processing method, device, computer equipment, computer-readable storage medium and computer program product of the distribution network carrying capacity assessment model mentioned above obtain training samples corresponding to the distribution network and corresponding labels. The training samples include sample grid indicator data that characterize the power quality of the distribution network and sample inverter indicator data that characterize the inverters incorporated into the distribution network. The labels are used to characterize the assessment level of the distribution network carrying capacity corresponding to the training samples. That is, information of two dimensions, the grid dimension and the inverter dimension incorporated into the grid, is obtained to ensure the effectiveness and accuracy of subsequent model training. An initial assessment model based on a neural network is constructed. Through the initial assessment model, the carrying capacity of the distribution network is estimated based on the training samples to obtain the sample estimation level; based on the labels and the sample estimation levels, the initial assessment model is trained, and the trained initial assessment model is used as the distribution network carrying capacity assessment model. The distribution network carrying capacity assessment model is used to estimate the distribution network carrying capacity. In this way, by training with the indicator data of the power provider and the access party respectively, the model can mine and learn the detailed information of the power provider and the access party respectively, so as to make a more comprehensive and accurate estimate, thereby realizing an accurate and effective evaluation of the carrying capacity of the distribution network connected to the inverter. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the drawings required for use in the embodiments of the present application or related technical descriptions 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 related drawings can be obtained based on these drawings without paying creative work.

[0027] Figure 1 An application environment diagram of a method for processing a distribution network carrying capacity evaluation model in one embodiment;

[0028] Figure 2 A schematic flow chart of a method for processing a distribution network carrying capacity evaluation model in one embodiment;

[0029] Figure 3 A schematic diagram of a neural network in one embodiment;

[0030] Figure 4 A schematic diagram of a network parameter optimization process according to an embodiment;

[0031] Figure 5 A schematic diagram of pre-training effect comparison in one embodiment;

[0032] Figure 6 It is a structural block diagram of a processing device of a distribution network carrying capacity evaluation model in one embodiment;

[0033] Figure 7 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0034] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0035] The processing method of the distribution network carrying capacity evaluation model provided in the embodiment of the present application can be applied to Figure 1 In the application environment shown. Among them, the terminal 102 communicates with the server 104 through the network. The data storage system can store the data that the server 104 needs to process. The data storage system can be integrated on the server 104, or it can be placed on the cloud or other network servers. The processing method of the distribution network carrying capacity assessment model provided in the embodiment of the present application can be executed by the server 104 alone, or by the terminal 102 alone, or by the terminal 102 and the server 104 in collaboration, without specific limitation.

[0036] In some embodiments, the terminal 102 collects or obtains the training samples and corresponding labels corresponding to different historical periods for the same distribution network, and sends them to the server 104. Each training sample includes sample grid indicator data characterizing the power quality of the distribution network, and sample inverter indicator data characterizing the inverter incorporated into the distribution network. After obtaining the training samples and corresponding labels, the server 104 constructs an initial evaluation model based on a neural network, and estimates the carrying capacity of the distribution network based on the training samples through the initial evaluation model to obtain the sample estimation level; based on the labels and the sample estimation levels, the initial evaluation model is trained, and the trained initial evaluation model is used as the distribution network carrying capacity evaluation model, and the distribution network carrying capacity evaluation model is used to estimate the distribution network carrying capacity.

[0037] In some embodiments, after acquiring the distribution network carrying capacity assessment model (the trained initial assessment model), if the server 104 receives the target data returned by the terminal 102 within a preset time period, it calls the distribution network carrying capacity assessment model and processes the target data to estimate the assessment level of the distribution network's carrying capacity within the preset time period.

[0038] In other embodiments, after obtaining the distribution network carrying capacity assessment model, the server 104 may also send the distribution network carrying capacity assessment model to the terminal 102 for deployment. After receiving the assessment request, the terminal obtains the target data within a preset time period, and processes the target data through the distribution network carrying capacity assessment model to estimate the assessment level of the distribution network's carrying capacity within the preset time period.

[0039] The terminal 102 may be, but is not limited to, various personal computers, laptops, smart phones, tablet computers, devices with data collection functions, etc. The server 104 may be an independent physical server, a server cluster or distributed system consisting of multiple physical servers, or a cloud server providing cloud computing services.

[0040] In an exemplary embodiment, Figure 2 As shown, a method for processing a distribution network carrying capacity evaluation model is provided, and the method is applied to a computer device (which can be Figure 1 The terminal 102 may also be Figure 1 The server 104) is used as an example to illustrate, including the following steps S202 to S206. Among them:

[0041] Step S202, obtaining training samples corresponding to the distribution network and corresponding labels, the training samples include sample grid indicator data characterizing the power quality of the distribution network, and sample inverter indicator data characterizing the inverters incorporated into the distribution network, and the labels are used to characterize the evaluation level of the distribution network carrying capacity corresponding to the training samples.

[0042] Among them, the training samples are obtained by collecting data on the distribution network in the historical period, and there are corresponding training samples for each historical period. The sample power grid indicator data contained in the training samples reflects the power quality of the distribution network in the historical period. The distribution network is connected to at least one inverter (such as a distributed power supply multi-inverter). The sample power grid indicator data contains data of at least one power quality indicator under the evaluation indicator system corresponding to the distribution network. Exemplarily, according to the actual situation of the distribution network, an evaluation indicator system corresponding to the distribution network and related to the carrying capacity is determined. The evaluation indicator system includes multiple indicators such as voltage deviation, voltage fluctuation, harmonics, average power outage frequency, annual average line load rate, net load fluctuation, etc. Correspondingly, the sample power grid indicator data at least includes voltage deviation data, voltage fluctuation data, harmonic data, average power outage frequency data, annual average line load rate data, and net load fluctuation data.

[0043] The sample inverter index data includes the attribute information of multiple inverters in the distribution area deployed by the distribution network. For example, the distribution area contains random inverters, and randomness means that the usage data applied on the corresponding equipment is random. For example, for the inverter applied to electric vehicles, after the electric vehicle is connected to the distribution network, since the distribution pile of the electric vehicle is random, the driving behavior of the driver of the electric vehicle is random, the driving road condition is also random, and the charging habits of the driver are also random, then the corresponding usage data includes driving behavior data, road condition data, charging data, etc. in the corresponding historical period. It can be understood that the usage data of inverters on different electric vehicles are different and random. For another example, the distribution area contains inverters that are not random, such as photovoltaic equipment, and the usage data of the inverter applied to photovoltaics is fixed, such as fixed time, and is not random.

[0044] Thus, for the first inverter with randomness, the corresponding attribute information includes a randomness flag and usage data, and the randomness flag is used to identify that the usage data of the corresponding inverter is random. For the second inverter without randomness, the corresponding attribute information includes a fixed flag, which is used to identify that the usage data of the corresponding inverter is not random. Exemplarily, the sample inverter indicator data includes the number of first inverters and second inverters in the power distribution area, the attribute information of each first inverter, and the attribute information of each second inverter.

[0045] The carrying capacity of the distribution network refers to the maximum load that the distribution network can bear. The evaluation level is used to reflect the carrying capacity of the distribution network. For example, the stronger the carrying capacity, the better the evaluation level. For example, it is divided into multiple evaluation levels, namely excellent, good, fair, and poor, which can be represented by corresponding values, such as 1, 2, 3, and 4.

[0046] Optionally, the computer device obtains training samples and corresponding labels corresponding to different historical periods. In each historical period, the distribution network is connected to different inverters, and the corresponding training samples are different.

[0047] Exemplarily, the computer device obtains historical data of each detection point in the distribution network within a historical time interval, performs data preprocessing on the historical data, and normalizes the preprocessed data to obtain a data set. The computer device divides the data set into a training set, a validation set, and a test set, such as in a ratio of 7:2:1. The training set includes training samples corresponding to multiple different historical periods.

[0048] Data preprocessing can be data cleaning processing, and normalization processing can accelerate the convergence speed and accuracy of the neural network, such as normalizing the data value to [0,1] or [-1,1].

[0049] Step S204, constructing an initial evaluation model based on a neural network, using the initial evaluation model to estimate the carrying capacity of the distribution network based on training samples to obtain a sample estimation level.

[0050] The initial evaluation model is constructed based on a neural network. For example, the initial evaluation model is a BP (Back Propagation Neural Network) neural network model. Figure 3 FIG. 1 is a schematic diagram of a neural network in one embodiment. Figure 3 Take BP neural network as an example.

[0051] Exemplarily, the computer device constructs an initial evaluation model based on a neural network, inputs training samples into the initial evaluation model, and outputs the estimated grade of the samples.

[0052] Step S206, based on the labels and the sample estimated levels, the initial evaluation model is trained, and the trained initial evaluation model is used as the distribution network carrying capacity evaluation model, and the distribution network carrying capacity evaluation model is used to estimate the distribution network carrying capacity.

[0053] Exemplarily, the server adjusts the model parameters of the initial evaluation model according to the labels and the estimated levels of the samples to obtain a trained initial evaluation model.

[0054] Exemplarily, the server adjusts the model parameters of the initial evaluation model through an optimization algorithm according to the labels and the sample estimated levels to obtain a trained initial evaluation model. For example, the optimization algorithm is a genetic optimization algorithm.

[0055] In some embodiments, the initial evaluation model is trained based on the labels and sample prediction levels, including: constructing a target loss based on the difference between the labels and sample prediction levels; and iteratively training the initial evaluation model with minimizing the target loss as the training goal until a trained initial evaluation model is obtained.

[0056] Exemplarily, based on the difference between the label and the estimated level of the sample, the target loss is constructed by the Softmax (function for binary classification problems) or Sigmoid (function for multi-classification problems) function, and the minimization of the target loss is used as the training goal. The initial evaluation model is iteratively trained by the gradient descent method or Adam (Adaptive Moment Estimation) or RMSprop (Root Mean Square Propagation) to obtain a trained initial evaluation model.

[0057] In this implementation, the target loss is constructed based on the difference between the label and the sample estimation level, and the training goal is to minimize the objective function, which can ensure the accuracy of model training, enhance the generalization ability of the model, and thus ensure the evaluation accuracy of the distribution network carrying capacity evaluation model obtained after subsequent training.

[0058] In the processing method of the distribution network carrying capacity evaluation model, by obtaining training samples corresponding to the distribution network and corresponding labels, the training samples include sample grid indicator data representing the power quality of the distribution network and sample inverter indicator data representing the inverter incorporated into the distribution network, and the labels are used to represent the evaluation level of the distribution network carrying capacity corresponding to the training samples. That is, the information of the grid dimension and the inverter dimension incorporated into the grid is obtained to ensure the effectiveness and accuracy of subsequent model training. An initial evaluation model based on a neural network is constructed, and the carrying capacity of the distribution network is estimated based on the training samples through the initial evaluation model to obtain the sample estimation level; based on the labels and the sample estimation level, the initial evaluation model is trained, and the trained initial evaluation model is used as the distribution network carrying capacity evaluation model, and the distribution network carrying capacity evaluation model is used to estimate the distribution network carrying capacity. In this way, by training as the indicator data of the power provider and the access party, the model can mine and learn the detailed information of the power provider and the access party, so as to achieve accurate and effective evaluation of the carrying capacity of the distribution network connected to the inverter with more comprehensive and accurate estimation.

[0059] In some embodiments, the initial evaluation model includes a first feature extraction network and a second feature extraction network. Through the initial evaluation model, based on the training samples, the carrying capacity of the distribution network is estimated to obtain the sample estimation level, including: extracting features from the training samples through the first feature extraction network to obtain the first feature; extracting features from the training samples through the second feature extraction network to obtain the second feature; and determining the sample estimation level based on the first feature and the second feature.

[0060] Among them, the first feature extraction network is used to focus on (focus on) obtaining the power quality characteristics of the distribution network and the attribute characteristics of the first inverter (an inverter with randomness) from the training samples, that is, focusing on extracting the detailed information of the power quality and the detailed information of the first inverter from the training samples. The second feature extraction network is used to focus on (focus on) extracting the power quality characteristics of the power quality and the attribute characteristics of the second inverter (an inverter without randomness), that is, focusing on extracting the detailed information of the power quality and the detailed information of the second inverter from the training samples.

[0061] The first feature is used to reflect the detailed information of the power quality and the detailed information of the first inverter. The second feature is used to reflect the detailed information of the power quality and the detailed information of the second inverter.

[0062] It should be noted that the distribution network is connected to the inverter in different situations at different time periods, for example, it may be situation 1: connecting to the first inverter and the second inverter at the same time, or situation 2: only connecting to the first inverter, or situation 3: only connecting to the second inverter.

[0063] For situation 1, detailed information of the first inverter and the second inverter can be extracted; for situation 2, detailed information of the second inverter cannot be extracted, or the amount of information of the second inverter is less than that of the first inverter; for situation 3, detailed information of the first inverter cannot be extracted, or the amount of information of the first inverter is less than that of the second inverter.

[0064] To this end, in order to ensure that the trained initial evaluation model can be generally applied to access conditions of different distribution networks in different time periods, a first feature extraction network and a second feature extraction network are set to extract features of the first inverter and features of the second inverter respectively.

[0065] Exemplarily, for each training sample corresponding to a historical period, after the computer device inputs the training sample into the initial evaluation model, the sample power grid index data and the sample inverter index data of the training sample are input into the first feature extraction network to extract the first feature. The sample power grid index data and the sample inverter index data of the training sample are input into the second feature extraction network to extract the second feature.

[0066] The first feature and the second feature are integrated to obtain a fused feature, and based on the fused feature, the carrying capacity of the distribution network in the historical period is predicted to obtain a sample estimation level.

[0067] In this embodiment, through the first feature extraction network and the second feature extraction network, the usage of different inverters in the corresponding historical period can be considered more carefully, thereby providing the accuracy of model training. In this way, the trained initial evaluation model can more flexibly and accurately make reasonable and precise estimates of the access status of the distribution network in different time periods.

[0068] In some embodiments, model training may include a first training stage and a second training stage. First, enter the first training stage: in the first training stage, the first initial network and the second initial network are pre-trained respectively, that is, the first initial network is pre-trained by using a network parameter optimization processing method that matches the first initial network, and the pre-trained first initial network is used as the first feature extraction network in the foregoing text. At the same time, the second initial network is pre-trained by using a network parameter optimization processing method that matches the second initial network, and the pre-trained second initial network is used as the second feature extraction network in the foregoing text.

[0069] Then, the second training phase is entered, and the first feature extraction network and the second feature extraction network are jointly trained, that is, the process returns to step S202 to continue execution.

[0070] In this way, two feature extraction networks with different focuses are trained separately through pre-training in the first training stage, and then the networks are jointly trained in the second training stage, which can ensure the accuracy of model training.

[0071] In some embodiments, the pre-training step of the first initial network includes: obtaining a first pre-training sample and a corresponding pre-training label, the first pre-training sample includes the pre-trained power grid indicator data and the first inverter indicator data, the first inverter indicator data includes the attribute information of the first inverter; based on the first pre-training sample, the first sample feature is output through the first initial network. The output first sample feature is passed through the trained classification layer to obtain the first sample estimation level, and based on the first sample estimation level and the pre-training label, the network parameters of the first initial network are optimized using the corresponding network parameter optimization processing method to obtain the pre-trained first initial network.

[0072] The network parameter optimization processing method corresponding to the first initial network may be a sparrow algorithm. The first pre-training sample is a sample connected to the first inverter, which may not include the second inverter indicator data. The first initial network may be a BP network.

[0073] In this embodiment, the network parameters of the first initial network are correspondingly pre-trained through the first pre-training sample, and the first initial network is preliminarily adjusted to have the ability to extract the attribute characteristics of the first inverter.

[0074] In some embodiments, the pre-training step of the second initial network includes: obtaining a second pre-training sample and a corresponding pre-training label, the second pre-training sample includes the pre-trained power grid indicator data and the second inverter indicator data, the second inverter indicator data includes the attribute information of the second inverter; based on the second pre-training sample, the second sample feature is output through the second initial network. The output second sample feature is passed through the trained classification layer to obtain the second sample estimated level, based on the second sample estimated level and the pre-training label, the network parameters of the second initial network are optimized using the corresponding network parameter optimization processing method to obtain the pre-trained second initial network.

[0075] The network parameter optimization processing method corresponding to the second initial network may be a genetic optimization algorithm. The second pre-training sample is a sample connected to the second inverter, which may not include the first inverter indicator data. The second initial network may be a BP network.

[0076] like Figure 4FIG. 1 is a schematic diagram of a network parameter optimization process according to an embodiment. Figure 4 ( Figure 4 The BP neural network in is the second initial network) The specific steps are as follows:

[0077] Step 1: Determine the number of input layer nodes of the second initial network. The number of nodes is the number of inputs to the second initial network. Then, determine the appropriate number of hidden layers and the number of nodes per layer based on experience. You can start with a single hidden layer and gradually increase the complexity. The number of nodes is usually 2 to 3 times the number of input layer nodes. The number of output layer nodes is 1, corresponding to the output of the carrying capacity assessment level. Then select the activation function. The hidden layer generally uses ReLU or Sigmoid activation function. The output layer can select an appropriate activation function according to the actual situation of the problem. Then train the second initial network, pass the input data in the training set layer by layer, calculate the output through the weighted summation of each layer and the activation function, until the output layer obtains the predicted value, calculate the error between the predicted value and the actual value, calculate the total error according to the selected loss function, and then use the chain rule to propagate the error from the input layer to the front layer according to the error, calculate the error contribution of each node, update the weights and biases, and use a random number generator to assign initial values ​​to the weights and biases of the neural network through multiple iterations. The initial value is close to 0 to avoid problems such as gradient disappearance or gradient explosion.

[0078] Step 2: Using real number encoding, encode the weights and biases of the second initial network in the previous step into real number strings, and randomly generate a certain number of network parameter combinations (including multiple groups of network parameter combinations, each group of network parameters can be regarded as an individual) to form an initial population. According to the evaluation requirements, design the fitness function. Usually, the error between the predicted output and the actual output (such as mean square error MSE) can be used as the fitness value. The smaller the error, the higher the fitness. Next, perform genetic operations, select excellent individuals according to the fitness value as the parent generation, and individuals with high fitness are selected first. Then select two individuals from the parent generation, generate offspring individuals through real number crossover method, and then perform mutation operations on the offspring individuals to increase the diversity of the population. The mutation operation can be performed by randomly adjusting the weights and biases in small amplitudes. Repeat the selection, crossover, and mutation operations, replace the old population with the newly generated offspring, generate a new generation of population, and evaluate the fitness of the new population. Select the best individual among them, and judge whether the maximum number of iterations has been reached or the fitness value is no longer changing significantly. If not, continue to iterate. The iteration stops until the final condition is met (such as reaching the predetermined number of generations or the fitness no longer improves).

[0079] Step 3: After the iteration of the genetic optimization algorithm is completed, the individual with the highest fitness is extracted from the final population. The network parameters of this individual are the optimal parameters, that is, the weights and biases of the optimal second initial network; the second initial network is initialized using the selected optimal weights and biases, and the second initial network is pre-trained until the training error is less than the set threshold, and the pre-trained second initial network is obtained.

[0080] Among them, the parameters of the genetic optimization algorithm, such as population size, crossover rate, mutation rate, etc., can be adjusted according to actual conditions to improve the optimization effect. At the same time, complexity expansion can be carried out to increase the evolutionary strategies of the genetic algorithm, such as elite retention, dynamic adjustment of the crossover mutation rate, etc., to improve the optimization efficiency.

[0081] In order to compare the benefits of the genetic optimization algorithm, the pre-training effects of the genetic optimization algorithm and the non-genetic optimization algorithm were compared. Figure 5 As shown, it is a schematic diagram of pre-training effect comparison in one embodiment. Figure 5 From top to bottom in the figure, they are the accuracy graph without genetic optimization algorithm, the accuracy graph without genetic optimization algorithm, the accuracy graph with genetic optimization algorithm, and the accuracy graph with genetic optimization algorithm.

[0082] In this embodiment, the global optimization capability of the genetic optimization algorithm and the nonlinear mapping capability of the BP neural network are combined to significantly improve the accuracy of the pre-training process of the second initial network.

[0083] In some embodiments, based on the first feature and the second feature, determining the sample estimation level includes: respectively determining the weight corresponding to the first feature and the weight corresponding to the second feature; weighting the first feature and the second feature according to the weight corresponding to the first feature and the weight corresponding to the second feature to obtain the sample estimation level.

[0084] Exemplarily, the computer device determines the respective weights of the first feature and the second feature according to the sample power grid indicator data.

[0085] In this embodiment, the respective weights of the first feature and the second feature can be used to adjust the respective attention levels to the first inverter and the second inverter, thereby more accurately obtaining the sample estimation level for the training sample and improving the estimation accuracy.

[0086] In some embodiments, the initial evaluation model also includes a weight determination network, which determines the weight corresponding to the first feature and the weight corresponding to the second feature, respectively, including: obtaining sample inverter indicator data from training samples; inputting the sample inverter indicator data into the weight determination network, and outputting the weight corresponding to the first feature and the weight corresponding to the second feature.

[0087] The weight determination network is used to determine the weight of the first feature extraction network (ie, determine the weight of the first feature) and determine the weight of the second feature extraction network (ie, determine the weight of the second feature).

[0088] The weights determined by different training samples are not necessarily the same.

[0089] Exemplarily, after the training samples and corresponding labels are input into the initial evaluation model, the training samples are respectively input into the first feature extraction network and the second feature extraction network to obtain the first feature output by the first feature extraction network and the second feature output by the second feature extraction network.

[0090] The sample inverter index data in the training sample is input into the weight determination network, and the weight determination network outputs the weight corresponding to the first feature and the weight corresponding to the second feature.

[0091] Subsequently, the first feature and the second feature are weighted according to the weight corresponding to the first feature and the weight corresponding to the second feature through the fusion network of the initial evaluation model to obtain the fused feature, and the sample estimated level corresponding to the fused feature is determined, and the sample estimated level is output through the initial evaluation model.

[0092] In this embodiment, the weight corresponding to the training sample is adaptively determined by the weight determination network, so that the degree of attention to the first inverter and the second inverter can be adjusted subsequently, so that the sample estimation level for the training sample can be obtained more accurately, thereby improving the estimation accuracy.

[0093] In some embodiments, the method also includes: obtaining target data obtained by collecting data on the distribution network within a preset time period, the target data including target grid indicator data and target inverter indicator data corresponding to the preset time period; inputting the target data into a distribution network carrying capacity assessment model, and outputting an assessment level of the distribution network's carrying capacity within the preset time period.

[0094] As mentioned above, the trained initial evaluation model is used as the distribution network carrying capacity evaluation model. It can be understood that the first feature extraction network, the second feature extraction network and the weight determination network in the distribution network carrying capacity evaluation model are all trained.

[0095] Exemplarily, after obtaining the distribution network carrying capacity assessment model, the computer device determines that the assessment conditions are met and obtains target data obtained by collecting data on the distribution network within a preset time period. The target data includes target grid indicator data and target inverter indicator data corresponding to the preset time period.

[0096] The computer device inputs the target data into the distribution network carrying capacity assessment model, extracts the features of the target data through the first feature extraction network in the distribution network carrying capacity assessment model, and obtains the first feature. The computer device extracts the features of the target data through the second feature extraction network in the distribution network carrying capacity assessment model, and obtains the second feature.

[0097] The network is determined by the weights in the distribution network carrying capacity assessment model, and the first weight and the second weight corresponding to the target data are determined according to the target inverter index data. The first weight is the weight corresponding to the first feature for the target data, and the second weight is the weight corresponding to the second feature for the target data.

[0098] The first feature and the second feature are weighted according to the first weight and the second weight to obtain a fusion feature corresponding to the target data, and based on the fusion feature, an assessment level of the carrying capacity of the distribution network within a preset time period is determined and output.

[0099] In this embodiment, based on the trained distribution network carrying capacity assessment model, it is possible to make a more comprehensive and accurate estimate based on the detailed information of the power provider and the access party, thereby achieving an accurate and effective assessment of the carrying capacity of the distribution network connected to the inverter.

[0100] In a specific embodiment, the first training phase, the second training phase and the reasoning phase are specifically involved, as follows:

[0101] First training stage: The computer equipment constructs the first initial network and the second initial network respectively, and pre-trains the first initial network and the second initial network respectively, that is, the first initial network is pre-trained by using the network parameter optimization processing method matching the first initial network, and the pre-trained first initial network is used as the first feature extraction network mentioned above, and at the same time, the second initial network is pre-trained by using the network parameter optimization processing method matching the second initial network, and the pre-trained second initial network is used as the second feature extraction network mentioned above. Thus, the second training stage is entered.

[0102] The second training stage: the computer device obtains the training samples and corresponding labels corresponding to the distribution network. The training samples include sample grid indicator data representing the power quality of the distribution network and sample inverter indicator data representing the inverters incorporated into the distribution network. The labels are used to represent the evaluation level of the distribution network carrying capacity corresponding to the training samples. A weight determination network is constructed, and an initial evaluation model based on a neural network is constructed according to the weight determination network, the first initial network obtained in the first training stage, and the second initial network. The training samples are feature extracted through the first feature extraction network to obtain the first feature; the training samples are feature extracted through the second feature extraction network to obtain the second feature. Sample inverter indicator data is obtained from the training samples; the sample inverter indicator data is input into the weight determination network, and the weight corresponding to the first feature and the weight corresponding to the second feature are output. The first feature and the second feature are weighted according to the weight corresponding to the first feature and the weight corresponding to the second feature to obtain the sample estimated level. Based on the difference between the label and the sample estimated level, a target loss is constructed; the initial evaluation model is iteratively trained with minimizing the target loss as the training goal until a trained initial evaluation model is obtained. Then, the inference stage is entered.

[0103] Reasoning stage: After the computer device determines that the evaluation conditions are met, it obtains the target data obtained by collecting data on the distribution network within the preset period, and the target data includes the target grid index data and the target inverter index data corresponding to the preset period. The computer device inputs the target data into the distribution network carrying capacity evaluation model, and extracts the features of the target data through the first feature extraction network in the distribution network carrying capacity evaluation model to obtain the first feature. Through the second feature extraction network in the distribution network carrying capacity evaluation model, the target data is extracted to obtain the second feature. Through the weight determination network in the distribution network carrying capacity evaluation model, according to the target inverter index data, the first weight and the second weight corresponding to the target data are determined, the first weight is the weight corresponding to the first feature for the target data, and the second weight is the weight corresponding to the second feature for the target data. According to the first weight and the second weight, the first feature and the second feature are weighted to obtain the fusion feature corresponding to the target data, and according to the fusion feature, the evaluation level of the carrying capacity of the distribution network within the preset period is determined and output.

[0104] In this embodiment, by obtaining training samples and corresponding labels corresponding to the distribution network, the training samples include sample grid indicator data characterizing the power quality of the distribution network, and sample inverter indicator data characterizing the inverter incorporated into the distribution network, and the label is used to characterize the evaluation level of the distribution network carrying capacity corresponding to the training sample. That is, information of two dimensions, the grid dimension and the inverter dimension incorporated into the grid, is obtained to ensure the effectiveness and accuracy of subsequent model training. An initial evaluation model based on a neural network is constructed, and the carrying capacity of the distribution network is estimated based on the training samples through the initial evaluation model to obtain the sample estimation level; based on the label and the sample estimation level, the initial evaluation model is trained, and the trained initial evaluation model is used as the distribution network carrying capacity evaluation model, and the distribution network carrying capacity evaluation model is used to estimate the distribution network carrying capacity. In this way, by training as the indicator data of the power provider and the access party, the model can mine and learn the detailed information of the power provider and the access party, so as to achieve accurate and effective evaluation of the carrying capacity of the distribution network connected to the inverter with more comprehensive and accurate estimation.

[0105] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0106] Based on the same inventive concept, the embodiment of the present application also provides a processing device for a distribution network carrying capacity assessment model for implementing the processing method of the distribution network carrying capacity assessment model involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in the embodiments of the processing device for one or more distribution network carrying capacity assessment models provided below can refer to the limitations of the processing method for the distribution network carrying capacity assessment model above, and will not be repeated here.

[0107] In an exemplary embodiment, Figure 6 As shown, a processing device 600 for a distribution network carrying capacity evaluation model is provided, comprising: an acquisition module 602, an estimation module 604 and a training module 606, wherein:

[0108] An acquisition module 602 is used to acquire training samples corresponding to the distribution network and corresponding labels, wherein the training samples include sample grid indicator data representing the power quality of the distribution network and sample inverter indicator data representing the inverters incorporated into the distribution network, and the labels are used to represent the evaluation level of the distribution network carrying capacity corresponding to the training samples;

[0109] An estimation module 604 is used to construct an initial estimation model based on a neural network, and to estimate the carrying capacity of the distribution network based on training samples through the initial estimation model to obtain a sample estimation level;

[0110] The training module 606 is used to train the initial evaluation model based on the labels and sample estimation levels, and use the trained initial evaluation model as the distribution network carrying capacity evaluation model, which is used to estimate the distribution network carrying capacity.

[0111] In some embodiments, the initial evaluation model includes a first feature extraction network and a second feature extraction network, and an estimation module 604 is used to extract features of the training sample through the first feature extraction network to obtain a first feature; extract features of the training sample through the second feature extraction network to obtain a second feature; and determine the sample estimation level based on the first feature and the second feature.

[0112] In some embodiments, the estimation module 604 is used to determine the weight corresponding to the first feature and the weight corresponding to the second feature respectively; weight the first feature and the second feature according to the weight corresponding to the first feature and the weight corresponding to the second feature to obtain the sample estimation level.

[0113] In some embodiments, the initial evaluation model also includes a weight determination network, an estimation module 604, which is used to obtain sample inverter indicator data from training samples; input the sample inverter indicator data into the weight determination network, and output the weight corresponding to the first feature and the weight corresponding to the second feature.

[0114] In some embodiments, the training module 606 is used to construct a target loss based on the difference between the label and the sample estimated level; and iteratively train the initial evaluation model with minimizing the target loss as the training goal until a trained initial evaluation model is obtained.

[0115] In some embodiments, the device also includes an inference module, which is used to obtain target data obtained by collecting data on the distribution network within a preset time period, and the target data includes target power grid indicator data and target inverter indicator data corresponding to the preset time period; the target data is input into the distribution network carrying capacity assessment model, and the assessment level of the distribution network's carrying capacity within the preset time period is output.

[0116] Each module in the processing device of the distribution network carrying capacity assessment model can be implemented in whole or in part by software, hardware and a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute operations corresponding to each module.

[0117] In an exemplary embodiment, a computer device is provided. The computer device may be a server or a terminal. The internal structure diagram thereof may be as follows: Figure 7 As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a processing method for a distribution network carrying capacity evaluation model is implemented.

[0118] Those skilled in the art will understand that Figure 7 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0119] In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the following steps when executing the computer program: obtaining training samples corresponding to a distribution network and corresponding labels, wherein the training samples include sample power grid indicator data characterizing the power quality of the distribution network and sample inverter indicator data characterizing the inverters incorporated into the distribution network, and the labels are used to characterize the evaluation level of the distribution network carrying capacity corresponding to the training samples; constructing an initial evaluation model based on a neural network, and using the initial evaluation model to estimate the carrying capacity of the distribution network based on the training samples to obtain the sample estimation level; training the initial evaluation model based on the labels and the sample estimation levels, and using the trained initial evaluation model as a distribution network carrying capacity evaluation model, and the distribution network carrying capacity evaluation model is used to estimate the distribution network carrying capacity.

[0120] In one embodiment, the initial evaluation model includes a first feature extraction network and a second feature extraction network, and the processor also implements the following steps when executing the computer program: extracting features from the training samples through the first feature extraction network to obtain a first feature; extracting features from the training samples through the second feature extraction network to obtain a second feature; and determining the sample estimation level based on the first feature and the second feature.

[0121] In one embodiment, when the processor executes the computer program, the following steps are also implemented: respectively determining the weight corresponding to the first feature and the weight corresponding to the second feature; weighting the first feature and the second feature according to the weight corresponding to the first feature and the weight corresponding to the second feature to obtain the sample estimation level.

[0122] In one embodiment, the initial evaluation model also includes a weight determination network, and the processor also implements the following steps when executing the computer program: obtaining sample inverter indicator data from training samples; inputting the sample inverter indicator data into the weight determination network, and outputting the weight corresponding to the first feature and the weight corresponding to the second feature.

[0123] In one embodiment, when the processor executes the computer program, the following steps are also implemented: constructing a target loss based on the difference between the label and the sample estimated level; and iteratively training the initial evaluation model with minimizing the target loss as the training goal until a trained initial evaluation model is obtained.

[0124] In one embodiment, when the processor executes the computer program, the following steps are also implemented: obtaining target data obtained by collecting data on the distribution network within a preset time period, the target data including target power grid index data and target inverter index data corresponding to the preset time period; inputting the target data into the distribution network carrying capacity assessment model, and outputting the assessment level of the distribution network's carrying capacity within the preset time period.

[0125] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented: obtaining training samples corresponding to a distribution network and corresponding labels, the training samples include sample power grid indicator data characterizing the power quality of the distribution network, and sample inverter indicator data characterizing the inverters incorporated into the distribution network, and the labels are used to characterize the evaluation level of the distribution network carrying capacity corresponding to the training samples; constructing an initial evaluation model based on a neural network, and using the initial evaluation model, based on the training samples, estimating the carrying capacity of the distribution network to obtain the sample estimation level; based on the labels and the sample estimation levels, training the initial evaluation model, and using the trained initial evaluation model as a distribution network carrying capacity evaluation model, and the distribution network carrying capacity evaluation model is used to estimate the distribution network carrying capacity.

[0126] In one embodiment, the initial evaluation model includes a first feature extraction network and a second feature extraction network. When the computer program is executed by a processor, the following steps are also implemented: feature extraction is performed on the training sample through the first feature extraction network to obtain a first feature; feature extraction is performed on the training sample through the second feature extraction network to obtain a second feature; based on the first feature and the second feature, the sample estimation level is determined.

[0127] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented: respectively determining the weight corresponding to the first feature and the weight corresponding to the second feature; weighting the first feature and the second feature according to the weight corresponding to the first feature and the weight corresponding to the second feature to obtain the sample estimation level.

[0128] In one embodiment, the initial evaluation model also includes a weight determination network, and when the computer program is executed by the processor, the following steps are also implemented: inputting the sample inverter index data into the weight determination network, and outputting the weight corresponding to the first feature and the weight corresponding to the second feature.

[0129] In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: constructing a target loss based on the difference between the label and the sample estimated level; iteratively training the initial evaluation model with minimizing the target loss as the training goal until a trained initial evaluation model is obtained.

[0130] In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: obtaining target data obtained by collecting data on the distribution network within a preset time period, the target data including target power grid indicator data and target inverter indicator data corresponding to the preset time period; inputting the target data into the distribution network carrying capacity assessment model, and outputting the assessment level of the distribution network's carrying capacity within the preset time period.

[0131] In one embodiment, a computer program product is provided, including a computer program, which implements the following steps when executed by a processor: obtaining training samples corresponding to a distribution network and corresponding labels, the training samples including sample power grid indicator data characterizing the power quality of the distribution network and sample inverter indicator data characterizing the inverters incorporated into the distribution network, and the labels are used to characterize the evaluation level of the distribution network carrying capacity corresponding to the training samples; constructing an initial evaluation model based on a neural network, and using the initial evaluation model to estimate the carrying capacity of the distribution network based on the training samples to obtain the sample estimation level; training the initial evaluation model based on the labels and the sample estimation levels, and using the trained initial evaluation model as a distribution network carrying capacity evaluation model, and the distribution network carrying capacity evaluation model is used to estimate the distribution network carrying capacity.

[0132] In one embodiment, the initial evaluation model includes a first feature extraction network and a second feature extraction network. When the computer program is executed by a processor, the following steps are also implemented: feature extraction is performed on the training sample through the first feature extraction network to obtain a first feature; feature extraction is performed on the training sample through the second feature extraction network to obtain a second feature; based on the first feature and the second feature, the sample estimation level is determined.

[0133] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented: respectively determining the weight corresponding to the first feature and the weight corresponding to the second feature; weighting the first feature and the second feature according to the weight corresponding to the first feature and the weight corresponding to the second feature to obtain the sample estimation level.

[0134] In one embodiment, the initial evaluation model also includes a weight determination network, and when the computer program is executed by the processor, the following steps are also implemented: inputting the sample inverter index data into the weight determination network, and outputting the weight corresponding to the first feature and the weight corresponding to the second feature.

[0135] In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: constructing a target loss based on the difference between the label and the sample estimated level; iteratively training the initial evaluation model with minimizing the target loss as the training goal until a trained initial evaluation model is obtained.

[0136] In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: obtaining target data obtained by collecting data on the distribution network within a preset time period, the target data including target power grid indicator data and target inverter indicator data corresponding to the preset time period; inputting the target data into the distribution network carrying capacity assessment model, and outputting the assessment level of the distribution network's carrying capacity within the preset time period.

[0137] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0138] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, an artificial intelligence (AI) processor, etc., but are not limited to this.

[0139] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above 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 application.

[0140] 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 understood as limiting the scope of the present 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 present application shall be subject to the attached claims.

Claims

1. A processing method for a distribution network carrying capacity evaluation model, characterized in that: The method comprises: Acquire training samples corresponding to the distribution network and corresponding labels, wherein the training samples include sample grid indicator data characterizing the power quality of the distribution network and sample inverter indicator data characterizing the inverters incorporated into the distribution network, and the labels are used to characterize the evaluation level of the distribution network carrying capacity corresponding to the training samples; Constructing an initial evaluation model based on a neural network, and using the initial evaluation model to estimate the carrying capacity of the distribution network based on the training samples to obtain a sample estimation level; Based on the labels and the sample estimated levels, the initial evaluation model is trained, and the trained initial evaluation model is used as a distribution network carrying capacity evaluation model, and the distribution network carrying capacity evaluation model is used to estimate the distribution network carrying capacity.

2. The method according to claim 1, characterized in that The initial evaluation model includes a first feature extraction network and a second feature extraction network. The initial evaluation model is used to estimate the carrying capacity of the distribution network based on the training samples to obtain a sample estimation level, including: Extracting features from the training sample using the first feature extraction network to obtain a first feature; Extracting features from the training samples using the second feature extraction network to obtain second features; Based on the first feature and the second feature, a sample estimation level is determined.

3. The method according to claim 2, characterized in that The step of determining the sample estimation level based on the first feature and the second feature includes: respectively determining a weight corresponding to the first feature and a weight corresponding to the second feature; The first feature and the second feature are weighted according to the weight corresponding to the first feature and the weight corresponding to the second feature to obtain a sample estimated level.

4. The method according to claim 3, characterized in that The initial evaluation model further includes a weight determination network, and the step of respectively determining the weight corresponding to the first feature and the weight corresponding to the second feature includes: Acquiring the sample inverter indicator data from the training sample; The sample inverter index data is input into the weight determination network, and the weight corresponding to the first feature and the weight corresponding to the second feature are output.

5. The method according to claim 1, characterized in that The training of the initial evaluation model based on the label and the sample estimated level includes: constructing a target loss based on the difference between the label and the estimated grade of the sample; Taking minimizing the target loss as the training goal, the initial evaluation model is iteratively trained until a trained initial evaluation model is obtained.

6. The method according to claim 1, characterized in that The method further comprises: Acquire target data obtained by collecting data on the distribution network within a preset time period, wherein the target data includes target power grid index data and target inverter index data corresponding to the preset time period; The target data is input into the distribution network carrying capacity assessment model, and an assessment level of the distribution network carrying capacity within a preset time period is output.

7. A processing device for a distribution network carrying capacity evaluation model, characterized in that: The device comprises: An acquisition module, used to acquire training samples corresponding to the distribution network and corresponding labels, wherein the training samples include sample grid indicator data representing the power quality of the distribution network and sample inverter indicator data representing the inverters incorporated into the distribution network, and the labels are used to represent the evaluation level of the distribution network carrying capacity corresponding to the training samples; An estimation module is used to construct an initial estimation model based on a neural network, and to estimate the carrying capacity of the distribution network based on the training samples through the initial estimation model to obtain a sample estimation level; A training module is used to train the initial evaluation model based on the label and the sample estimation level, and use the trained initial evaluation model as a distribution network carrying capacity evaluation model, wherein the distribution network carrying capacity evaluation model is used to estimate the distribution network carrying capacity.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images