A line loss prediction method, device and terminal of a photovoltaic grid-connected power distribution network

Abstract

The application provides a line loss prediction method, device and terminal for a photovoltaic grid-connected power distribution network. The method comprises: obtaining a first influence feature of the photovoltaic grid-connected power distribution network, and determining an influence degree of the photovoltaic grid-connected power distribution network on line loss according to the first influence feature, wherein the first influence feature comprises an access position, an access capacity and a proportion of on-grid power; if the influence degree is greater than a set threshold, performing line loss prediction based on a first line loss factor index for influencing line loss and using a trained first line loss prediction model to obtain a line loss prediction result; and if the influence degree is not greater than the set threshold, performing line loss prediction based on a second line loss factor index for influencing line loss and using the trained first line loss prediction model to obtain a line loss prediction result. The application reduces the prediction difficulty of line loss of the photovoltaic grid-connected power distribution network, improves the prediction accuracy of line loss, and provides a reliable reference for energy saving and loss reduction of the power distribution network after distributed photovoltaic access and realization of the double-carbon target of the power grid.

Description

Technical Field

[0001] This invention belongs to the field of distribution network line loss prediction technology, and particularly relates to a method, device and terminal for predicting line losses in a photovoltaic grid-connected distribution network. Background Technology

[0002] Line loss refers to the percentage of electrical energy lost during power grid transmission, usually expressed as the line loss rate. Currently, in power systems, the line loss rate is a crucial indicator reflecting the economic efficiency of grid operation. It accurately reflects the rationality of the grid's planning and design, the operational management level of power companies, and provides a reference for companies to formulate energy-saving and loss-reduction policies. To better balance power supply quality and economic benefits, accurate prediction of distribution network line losses is necessary during distribution network planning.

[0003] However, numerous factors influence line losses, including transmission and distribution equipment, grid topology, and operating methods. Even management factors such as meter reading errors, inconsistent meter readings, poor insulation of live equipment, and electricity theft can affect line losses. Furthermore, the increasing integration of distributed photovoltaic systems into the distribution network further exacerbates the difficulty and inaccuracy of line loss prediction. On the other hand, existing deep learning-based line loss prediction methods require a large number of samples due to the numerous influencing factors, and even with a sufficient number of samples, the training and learning workload becomes substantial. Summary of the Invention

[0004] This invention provides a method, device, and terminal for predicting line losses in a photovoltaic grid-connected distribution network, in order to solve the problems of inaccurate line loss prediction or large training workload in the prior art.

[0005] In a first aspect, the present invention provides a method for predicting line losses in a photovoltaic grid-connected distribution network, comprising: acquiring a first impact feature of photovoltaic grid connection, and determining the degree of impact of photovoltaic grid connection on line losses based on the first impact feature, wherein the first impact feature includes the connection location, the connection capacity, and the proportion of electricity fed into the grid;

[0006] If the impact exceeds the set threshold, then based on the predetermined first line loss factor index that affects line loss, the trained first line loss prediction model is used to predict line loss and obtain the line loss prediction result.

[0007] If the impact is not greater than the set threshold, then based on the predetermined second line loss factor index that affects line loss, the trained second line loss prediction model is used to predict line loss and obtain the line loss prediction result.

[0008] The first line loss factor index includes line loss factor indices related to photovoltaic grid connection, while the second line loss factor index does not include line loss factor indices related to photovoltaic grid connection.

[0009] Secondly, the present invention provides a line loss prediction device for a photovoltaic grid-connected distribution network, comprising:

[0010] The impact determination unit is used to obtain the first impact characteristics of photovoltaic grid connection and determine the impact of photovoltaic grid connection on line loss based on the first impact characteristics. The first impact characteristics include the access location, access capacity and the proportion of grid-connected electricity.

[0011] The first prediction unit is used to predict line loss based on a pre-determined first line loss factor index that affects line loss, and to obtain the line loss prediction result if the impact is greater than a set threshold.

[0012] The second prediction unit is used to predict line loss based on the pre-determined second line loss factor index that affects line loss, and to obtain the line loss prediction result if the impact is not greater than a set threshold.

[0013] The first line loss factor index includes line loss factor indices related to photovoltaic grid connection, while the second line loss factor index does not include line loss factor indices related to photovoltaic grid connection.

[0014] Thirdly, the present invention provides a terminal including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the method for predicting line losses in a photovoltaic grid-connected distribution network as described in the first aspect or any possible implementation thereof.

[0015] This invention provides a method, device, and terminal for predicting line losses in a photovoltaic (PV) grid-connected distribution network. It obtains first influencing characteristics of PV grid connection, including connection location, connection capacity, and the proportion of electricity fed into the grid. Based on these first influencing characteristics, the impact of PV grid connection on line losses is determined. When the impact exceeds a set threshold, the influence of PV-related line loss factor indicators on the distribution network line loss rate is considered, and a corresponding first line loss prediction model is used for line loss prediction. When the impact is not greater than the set threshold, due to the limited impact, the influence of PV-related line loss factor indicators on the distribution network line loss rate can be disregarded, and only the influence of the distribution network's own line loss factor indicators is considered. A trained second line loss prediction model is then used for line loss prediction. On one hand, this approach considers the magnitude of the impact of PV grid connection on the line loss rate under different influencing characteristics and uses different prediction models to predict the line loss rate, improving prediction accuracy. On the other hand, it optimizes the line loss factor indicators, considering only the main influencing factors affecting the line loss rate, thus solving the problem of large sample data requirements and computational complexity during training caused by numerous line loss factor indicators. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart illustrating the implementation of the line loss prediction method for photovoltaic grid-connected distribution networks provided in this embodiment of the invention.

[0018] Figure 2 This is a schematic diagram of the structure of the line loss prediction device for photovoltaic grid-connected distribution network provided in an embodiment of the present invention;

[0019] Figure 3 This is a schematic diagram of the terminal provided in an embodiment of the present invention. Detailed Implementation

[0020] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of the invention. However, those skilled in the art will understand that the invention can be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods are omitted so as not to obscure the description of the invention with unnecessary detail.

[0021] To make the objectives, technical solutions, and advantages of the present invention clearer, specific embodiments will be described below in conjunction with the accompanying drawings.

[0022] Currently, line losses in my country's medium and low voltage distribution networks account for approximately 70% of the total power grid losses. This significant energy loss not only reduces the profitability of power grid companies but also contradicts my country's dual-carbon development goals. This invention uses a 10kV distribution network as an example for line loss prediction, which also incorporates distributed photovoltaic power generation.

[0023] This solution can improve the accuracy of line loss prediction, which is conducive to providing a reliable reference for energy saving and loss reduction in the distribution network after the integration of distributed photovoltaic power and the realization of the dual carbon target of the power grid.

[0024] See Figure 1 The document illustrates a flowchart of the line loss prediction method for photovoltaic grid-connected distribution networks provided in this embodiment of the invention, detailed below:

[0025] Step 101: Obtain the first impact characteristics of photovoltaic grid connection, and determine the degree of impact of photovoltaic grid connection on line loss based on the first impact characteristics.

[0026] In this embodiment of the invention, the first influencing feature refers to the indicator among the characteristic indicators in photovoltaic grid-connected related data that has a significant impact on the overall line loss rate of the distribution network. In practical applications, this can be obtained through correlation analysis.

[0027] For example, the characteristic indicators of photovoltaic grid connection may include irradiance, average temperature, grid connection capacity, grid connection location, and the proportion of electricity generated to the grid. Among them, grid connection capacity can be 1000 kW, 2000 kW, 3000 kW, etc., representing the rated grid connection capacity of photovoltaic grid connection; grid connection location indicates which grid connection node of the distribution network the photovoltaic grid connection is located at; and the proportion of electricity generated to the grid refers to the ratio of the actual electricity generated by the photovoltaic grid connection to the total electricity distributed by the distribution network.

[0028] In this embodiment, relevant data from the distribution network can be collected in advance, normalized, and preprocessed. Using these characteristic indicators as independent variables and the upward fluctuation of the line loss rate as the dependent variable, a single-factor analysis method is used to analyze the correlation between each characteristic indicator and the dependent variable. Then, several characteristic indicators with high correlation are selected as the first influencing characteristics of photovoltaic grid connection. In this embodiment, the first influencing characteristics determined through analysis include the connection location, connection capacity, and the proportion of electricity fed into the grid.

[0029] In this embodiment, the first influencing feature reflects the degree of impact of photovoltaic grid connection on the line loss rate. When the impact of photovoltaic grid connection on the line loss rate is small, the photovoltaic grid connection index can be disregarded, and the impact of photovoltaic grid connection on the line loss rate can be regarded as an error term. Based on the influence shadow of the distribution network itself, a deep learning model can be used for prediction, which can obtain a relatively accurate prediction result, thus requiring too many data samples and reducing the training workload. Conversely, when the impact of photovoltaic grid connection on the line loss rate is large, the corresponding error term is large, which has affected the accuracy of the line loss rate prediction. It is necessary to consider the photovoltaic grid connection-related influencing factors as a major influencing factor and achieve accurate prediction of the line loss rate through another line loss prediction model.

[0030] In one implementation, the degree of influence of each feature in the first influencing feature on line loss can be calculated separately, and the weighted summation can be used to obtain the degree of influence of photovoltaic grid connection on line loss.

[0031] In one implementation, step 101 above can be specifically achieved through the following steps:

[0032] Calculate the first degree of influence of the access location on the target distribution network line loss rate;

[0033] Calculate the second degree of influence of access capacity on the target distribution network line loss rate;

[0034] Calculate the third degree of influence of the proportion of electricity generated in the grid on the target distribution network line loss rate;

[0035] The total impact of photovoltaic grid connection on line loss is obtained based on the first, second, and third impact degrees.

[0036] In this embodiment, a model with the same architecture as the target distribution network can be constructed. Using this model, different access capacities are set, and line loss rate data at different access locations are simulated for each access capacity. Correlation analysis is performed on the line loss rate and access location using the simulated data to obtain a first linear relationship. Similarly, using the same model, different access locations are set, and line loss rate data corresponding to different access capacities are simulated at each access location. Correlation analysis is performed on the line loss rate and access capacity using the simulated data to obtain a second linear relationship. Furthermore, different access capacities and access locations are set, and line loss rate data corresponding to the proportion of electricity generated into the grid is simulated. Correlation analysis is performed on the line loss rate and the proportion of electricity generated into the grid using the simulated data to obtain a third linear relationship. Therefore, given a fixed access capacity, a first correlation coefficient corresponding to the access location can be obtained based on the first linear relationship; given a fixed access location, a second correlation coefficient corresponding to the access capacity can be obtained based on the second linear relationship; and given a fixed access capacity and access location, a third correlation coefficient corresponding to the proportion of electricity generated into the grid can be obtained based on the third linear relationship.

[0037] In this embodiment, a multi-factor analysis method is used to analyze each feature in the first influencing feature to determine the influence weight of each feature on the line loss rate. For example, the first influence weight of access location on line loss rate, the second influence weight of access capacity on line loss rate, and the third influence weight of the proportion of electricity used for internet access on line loss rate can be obtained.

[0038] The first influence of grid connection location on line loss rate can be obtained based on the first correlation coefficient and the first influence weight. The second influence of grid connection capacity on line loss rate can be obtained based on the second correlation coefficient and the second influence weight. The third influence of grid-connected electricity ratio on line loss rate can be obtained based on the third correlation coefficient and the third influence weight. The three influence values ​​are added together to obtain the influence of photovoltaic grid connection on line loss.

[0039] Step 102: If the impact is greater than the set threshold, then based on the predetermined first line loss factor index that affects line loss, use the trained first line loss prediction model to predict line loss and obtain the line loss prediction result.

[0040] In this embodiment, two line loss prediction models are pre-trained to predict line losses for different scenarios. The first line loss prediction model is for scenarios where the impact is greater than a set threshold and considers the prediction of line loss factor indicators related to photovoltaic grid connection. The second line loss prediction model is for scenarios where the impact is not greater than the set threshold and does not consider the prediction of line loss factor indicators related to photovoltaic grid connection. For example, the set threshold can be 0.5%.

[0041] In step 102, since the impact of photovoltaic grid connection on line loss is greater than the set threshold, it is necessary to consider the line loss factor index related to photovoltaic grid connection.

[0042] In this embodiment, the first line loss factor index includes line loss factor indices related to photovoltaic grid connection. Specifically, in one implementation, the step of determining the first line loss factor index may include:

[0043] Obtain sample data for the first line loss indicator;

[0044] A first reference sequence is generated based on the actual line loss data in the first line loss index sample data;

[0045] A first comparison sequence is generated based on the first line loss index data related to photovoltaic grid connection in the first line loss index sample data;

[0046] A second comparison sequence is generated based on the second line loss index data related to the distribution network from the first line loss index sample data;

[0047] The first theoretical line loss data is calculated based on the first line loss index data and the second line loss index data in the first line loss index sample data, and a second reference sequence is generated based on the first theoretical line loss data;

[0048] Grey relational analysis was performed on the first comparison sequence, the second comparison sequence, and the first reference sequence to obtain the first correlation degree.

[0049] Grey relational analysis was performed on the first comparison sequence, the second comparison sequence, and the second reference sequence to obtain the second correlation degree.

[0050] The first and second correlation degrees are fused to obtain the first fused correlation degree between each line loss index and the line loss rate;

[0051] The line loss indicators with the highest first degree of fusion correlation are determined as the first line loss factor indicators.

[0052] In this embodiment, each sample in the first line loss index sample data includes both line loss factor indicators related to photovoltaic grid connection and line loss factor indicators of the distribution network itself. For example, line loss factor indicators related to photovoltaic grid connection may include: irradiance, average temperature, access capacity, access location, and actual grid-connected electricity ratio, etc. Line loss factor indicators of the distribution network itself may include line attribute indicators, such as trunk line length, total line length, number of distribution transformers, distribution transformer capacity, distribution transformer model, line current carrying capacity, and cable conversion rate, etc.; they may also include basic operation indicators, such as active power supply, reactive power supply, iron loss, copper loss, maximum line load rate, line outlet power factor, and average power factor of distribution transformers, etc.; and they may also include some management factor indicators, such as equipment aging degree, power supply area, meter reading accuracy, meter loss, operating years, and smart meter penetration rate, etc.

[0053] In this embodiment, a first reference sequence can be generated by preprocessing the actual line loss data (line loss rate) in the first line loss index sample data. A first comparison sequence can be generated by the first line loss index data related to photovoltaic grid connection. There can be multiple first comparison sequences. A second comparison sequence can be generated by the second line loss index data related to distribution network. There can also be multiple second comparison sequences. The first correlation degree between the comparison sequence and the first reference sequence can be obtained through grey relational analysis.

[0054] In this embodiment, the actual line loss data (line loss rate) in the first line loss index sample data is affected by many incidental factors. Due to the limited sample size, these incidental factors directly cause fluctuations in the line loss rate, affecting the correlation analysis results. This scheme uses the first and second line loss index data from the first line loss index sample data to theoretically calculate the first theoretical line loss data, which is the ideal data excluding the influence of incidental factors. A second reference sequence can be generated based on the first theoretical line loss data; then, grey relational analysis is used to obtain the second correlation degree between the comparison sequence and the second reference sequence.

[0055] Because the first correlation degree is significantly affected by unforeseen factors, while the second correlation degree completely ignores them, both exhibit errors compared to actual conditions, and these errors are in opposite directions. Therefore, the first and second correlation degrees can be fused to obtain a first fused correlation degree for each line loss indicator and line loss rate. The fusion method can be weighted summation. The resulting first fused correlation degree reduces the impact of extreme unforeseen factors and has higher accuracy. Consequently, the first line loss factor indicator selected based on the first fused correlation degree is also more accurate.

[0056] In this embodiment, the first line loss factor index is ultimately selected as the line loss factor index that can play a major role in the line loss rate, taking into account the impact of photovoltaic grid connection. After determining the first line loss factor index, the actual data corresponding to the first line loss factor index of the distribution network can be obtained and input into the trained second line loss prediction model to predict line loss and obtain the line loss prediction result.

[0057] In this embodiment, the first line loss prediction model is a Stacking ensemble model, which consists of a photovoltaic grid-connected line loss factor index base learner, a distribution network line loss factor index base learner, and a line loss rate prediction meta-learner. The first line loss prediction model is trained on a first sample set formed by the first line loss factor index and its corresponding historical line loss rate data.

[0058] In this embodiment, the line loss prediction result obtained by using the trained first line loss prediction model can include:

[0059] The photovoltaic grid-connected line loss factor index data in the first line loss factor index is input into the photovoltaic grid-connected line loss factor index base learner to obtain the first output;

[0060] Input the distribution network line loss factor index data from the first line loss factor index into the distribution network line loss factor index base learner to obtain the second output;

[0061] The first and second outputs are input into the line loss rate prediction meta-learner to obtain the line loss prediction result.

[0062] In this embodiment, the training process of the first line loss prediction model is also the deep learning process of the model. The first sample set is used to form a training set and a validation set. The first line loss prediction model is trained using the training set, and then the model parameters of the first line loss prediction model are corrected using the validation set until the loss function of the first line loss prediction model meets the expectations.

[0063] In this embodiment, a Stacking ensemble model is adopted, which consists of two layers: a base learner and a meta-learner. The two base learners are used to learn the photovoltaic grid-connected line loss factor index data and the distribution network line loss factor index data, respectively. Then, the meta-learner is used to learn and predict the output results of the two base learners, and outputs the predicted line loss rate. This reduces the training workload and difficulty on the one hand, and achieves higher accuracy on the other.

[0064] Step 103: If the impact is not greater than the set threshold, then based on the predetermined second line loss factor index that affects line loss, use the trained second line loss prediction model to predict line loss and obtain the line loss prediction result.

[0065] In this embodiment, the second line loss prediction model is designed for prediction scenarios where the impact is no greater than a set threshold and does not consider line loss factor indicators related to photovoltaic grid connection.

[0066] In step 103, since the impact of photovoltaic grid connection on line loss is no greater than a set threshold, there is no need to consider the line loss factor index related to photovoltaic grid connection. Therefore, the second line loss factor index does not include the line loss factor index related to photovoltaic grid connection. Specifically, the steps for determining the second line loss factor index may include:

[0067] Obtain sample data for the second line loss indicator;

[0068] A third reference sequence is generated based on the actual line loss data in the second line loss index sample data;

[0069] A third comparison sequence is generated based on the third line loss index data in the second line loss index sample data;

[0070] The second theoretical line loss data is calculated based on the third line loss index data in the second line loss index sample data, and the fourth reference sequence is generated based on the second theoretical line loss data;

[0071] Grey relational analysis was performed on the third comparison sequence and the third reference sequence to obtain the third correlation degree.

[0072] Grey relational analysis was performed on the third comparison sequence and the fourth reference sequence to obtain the fourth correlation degree.

[0073] The third and fourth correlation degrees are fused to obtain the second fused correlation degree for each line loss index and line loss rate;

[0074] The line loss indicators with the highest pre-defined number of second fusion correlations are determined as the second line loss factor indicators.

[0075] In this embodiment, the determination process of the first line loss factor index can be referred to above. The difference is that the first line loss index data related to photovoltaic grid connection is no longer considered. Since its impact on the line loss rate result is small, it is treated as an error term.

[0076] In this embodiment, the second line loss prediction model can be a single-layer deep learning model, which is trained based on a second sample set formed by the second line loss factor index and its corresponding historical line loss rate data. The training and learning process can refer to the relevant processes of deep learning models in the prior art, and will not be described in detail here.

[0077] In summary, this invention provides a method for predicting line losses in a photovoltaic (PV) grid-connected distribution network. It obtains the first influencing characteristics of PV grid connection, including connection location, connection capacity, and the proportion of electricity fed into the grid. Based on these first influencing characteristics, the impact of PV grid connection on line losses is determined. When the impact exceeds a set threshold, the influence of PV grid-connected related line loss factor indicators on the distribution network line loss rate is considered, and a corresponding first line loss prediction model is used for line loss prediction. When the impact is not greater than the set threshold, due to the implicit impact, the influence of PV grid-connected related line loss factor indicators on the distribution network line loss rate can be disregarded; only the influence of the distribution network's own line loss factor indicators is considered, and a trained second line loss prediction model is used for line loss prediction. On the one hand, this method considers the magnitude of the impact of PV grid connection on the line loss rate under different influencing characteristics and uses different prediction models to predict the line loss rate, improving prediction accuracy. On the other hand, it optimizes the line loss factor indicators, considering only the main influencing factors affecting the line loss rate, thus solving the problem of large sample data requirements and computational complexity during training caused by numerous line loss factor indicators.

[0078] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0079] The following are device embodiments of the present invention. For details not described in detail, please refer to the corresponding method embodiments described above.

[0080] Figure 2 A schematic diagram of the line loss prediction device for a photovoltaic grid-connected distribution network provided in an embodiment of the present invention is shown. For ease of explanation, only the parts related to the embodiment of the present invention are shown, and are described in detail below:

[0081] like Figure 2 As shown, the line loss prediction device 2 for photovoltaic grid-connected distribution network includes: an impact determination unit 21, a first prediction unit 22, and a second prediction unit 23.

[0082] The impact determination unit 21 is used to obtain the first impact characteristics of photovoltaic grid connection and determine the impact of photovoltaic grid connection on line loss based on the first impact characteristics. The first impact characteristics include the access location, access capacity and the proportion of grid-connected electricity.

[0083] The first prediction unit 22 is used to predict line loss based on a pre-determined first line loss factor index that affects line loss, and to obtain the line loss prediction result if the impact is greater than a set threshold.

[0084] The second prediction unit 23 is used to predict line loss based on the pre-determined second line loss factor index that affects line loss, and to obtain the line loss prediction result if the impact degree is not greater than the set threshold.

[0085] The first line loss factor index includes line loss factor indices related to photovoltaic grid connection, while the second line loss factor index does not include line loss factor indices related to photovoltaic grid connection.

[0086] In this embodiment of the invention, the influence determination unit 21 is specifically used for:

[0087] Calculate the first degree of influence of grid connection location on the target distribution network line loss rate; calculate the second degree of influence of grid connection capacity on the target distribution network line loss rate; calculate the third degree of influence of grid-connected electricity ratio on the target distribution network line loss rate; and obtain the degree of influence of photovoltaic grid connection on line loss based on the first, second, and third degree of influence.

[0088] In one embodiment, the line loss prediction device 2 further includes a first line loss index determination unit, which is configured to: acquire first line loss index sample data; generate a first reference sequence based on the actual line loss data in the first line loss index sample data; generate a first comparison sequence based on the first line loss index data related to photovoltaic grid connection in the first line loss index sample data; generate a second comparison sequence based on the second line loss index data related to distribution network in the first line loss index sample data; calculate a first theoretical line loss data based on the first and second line loss index data in the first line loss index sample data, and generate a second reference sequence based on the first theoretical line loss data; perform grey relational analysis on the first comparison sequence, the second comparison sequence, and the first reference sequence to obtain a first correlation degree; perform grey relational analysis on the first comparison sequence, the second comparison sequence, and the second reference sequence to obtain a second correlation degree; fuse the first correlation degree and the second correlation degree to obtain a first fusion correlation degree for each line loss index and the line loss rate; and determine a preset number of line loss indices with the highest first fusion correlation degree as first line loss factor indices.

[0089] In one embodiment, the first line loss prediction model is a Stacking ensemble model, which consists of a photovoltaic grid-connected line loss factor index base learner, a distribution network line loss factor index base learner, and a line loss rate prediction meta learner; the first line loss prediction model is trained based on a first sample set formed by the first line loss factor index and its corresponding historical line loss rate data.

[0090] Accordingly, the process of using the trained first line loss prediction model to predict line loss and obtain the line loss prediction result includes:

[0091] The photovoltaic grid-connected line loss factor index data from the first line loss factor index is input into the photovoltaic grid-connected line loss factor index base learner to obtain the first output; the distribution network line loss factor index data from the first line loss factor index is input into the distribution network line loss factor index base learner to obtain the second output; the first output and the second output are input into the line loss rate prediction element learner to obtain the line loss prediction result.

[0092] In one embodiment, the line loss prediction device 2 further includes a second line loss index determination unit, which is specifically used to: acquire second line loss index sample data; generate a third reference sequence based on the actual line loss data in the second line loss index sample data; generate a third comparison sequence based on the third line loss index data in the second line loss index sample data; calculate second theoretical line loss data based on the third line loss index data in the second line loss index sample data, and generate a fourth reference sequence based on the second theoretical line loss data; perform grey relational analysis on the third comparison sequence and the third reference sequence to obtain a third correlation degree; perform grey relational analysis on the third comparison sequence and the fourth reference sequence to obtain a fourth correlation degree; fuse the third correlation degree and the fourth correlation degree to obtain a second fusion correlation degree for each line loss index and the line loss rate; and determine the line loss indexes with the highest preset number of second fusion correlation degrees as second line loss factor indices.

[0093] In one embodiment, the second line loss prediction model is a single-layer deep learning model, which is trained based on a second sample set formed by the second line loss factor index and its corresponding historical line loss rate data.

[0094] In summary, this invention provides a line loss prediction device for a photovoltaic (PV) grid-connected distribution network. It acquires the first influencing characteristics of PV grid connection, including connection location, connection capacity, and the proportion of electricity fed into the grid. Based on these first influencing characteristics, it determines the degree of influence of PV grid connection on line loss. When the degree of influence exceeds a set threshold, it considers the impact of PV grid-connected related line loss factor indicators on the distribution network line loss rate and uses a corresponding first line loss prediction model for line loss prediction. When the degree of influence is not greater than the set threshold, due to the implicit nature of the influence, it can disregard the impact of PV grid-connected related line loss factor indicators on the distribution network line loss rate and only consider the impact of the distribution network's own line loss factor indicators on the distribution network line loss rate, using a trained second line loss prediction model for line loss prediction. On the one hand, it considers the magnitude of the impact of PV grid connection on the line loss rate under different influencing characteristics and uses different prediction models to predict the line loss rate, improving prediction accuracy. On the other hand, it optimizes the line loss factor indicators, considering only the main influencing factors affecting the line loss rate, solving the problem of large sample data requirements and large computational workload during training caused by numerous line loss factor indicators.

[0095] Figure 3 This is a schematic diagram of a terminal provided in an embodiment of the present invention. Figure 3 As shown, the terminal 3 in this embodiment includes: a processor 30, a memory 31, and a computer program 32 stored in the memory 31 and executable on the processor 30. When the processor 30 executes the computer program 32, it implements the steps in the above-described embodiments of the photovoltaic grid-connected distribution network line loss prediction method, for example... Figure 1 Steps 101 to 103 are shown. Alternatively, when the processor 30 executes the computer program 32, it implements the functions of each unit in the above-described device embodiments, for example... Figure 2 The functions of units 21 to 23 are shown.

[0096] In summary, this invention provides a terminal that acquires the first impact characteristics of photovoltaic grid connection, including connection location, connection capacity, and the proportion of electricity fed into the grid. Based on these first impact characteristics, it determines the degree of influence of photovoltaic grid connection on line loss. When the degree of influence exceeds a set threshold, it considers the impact of photovoltaic grid connection-related line loss factor indicators on the distribution network line loss rate and uses a corresponding first line loss prediction model for line loss prediction. When the degree of influence is not greater than the set threshold, due to the implicit impact, it can disregard the impact of photovoltaic grid connection-related line loss factor indicators on the distribution network line loss rate and only consider the impact of the distribution network's own line loss factor indicators on the distribution network line loss rate, using a trained second line loss prediction model for line loss prediction. On the one hand, it considers the magnitude of the impact of photovoltaic grid connection on the line loss rate under different impact characteristics and uses different prediction models to predict the line loss rate, improving prediction accuracy. On the other hand, it optimizes the line loss factor indicators, considering only the main influencing factors affecting the line loss rate, solving the problem of large sample data requirements and large computational load during training caused by numerous line loss factor indicators.

[0097] The terminal 3 can be a desktop computer, laptop, handheld computer, or cloud server, etc. The terminal 3 may include, but is not limited to, a processor 30 and a memory 31. Those skilled in the art will understand that... Figure 3 This is merely an example of terminal 3 and does not constitute a limitation on terminal 3. It may include more or fewer components than shown, or combine certain components, or different components. For example, the terminal may also include input / output devices, network access devices, buses, etc.

[0098] The processor 30 may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0099] The memory 31 can be an internal storage unit of the terminal 3, such as a hard disk or memory of the terminal 3. The memory 31 can also be an external storage device of the terminal 3, such as a plug-in hard disk, smart media card (SMC), secure digital card (SD), flash card, etc., equipped on the terminal 3. Furthermore, the memory 31 can include both internal storage units and external storage devices of the terminal 3. The memory 31 is used to store the computer program and other programs and data required by the terminal. The memory 31 can also be used to temporarily store data that has been output or will be output.

[0100] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0101] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A line loss prediction method for a photovoltaic grid-connected distribution network, characterized in that, include: The first impact characteristics of photovoltaic grid connection are obtained, and the impact of photovoltaic grid connection on line loss is determined based on the first impact characteristics. The first impact characteristics include the access location, access capacity and the proportion of grid-connected electricity. If the impact exceeds the set threshold, then based on the predetermined first line loss factor index that affects line loss, the trained first line loss prediction model is used to predict line loss and obtain the line loss prediction result. If the impact is not greater than the set threshold, then based on the predetermined second line loss factor index that affects line loss, the trained second line loss prediction model is used to predict line loss and obtain the line loss prediction result. The first line loss factor index includes line loss factor indices related to photovoltaic grid connection, while the second line loss factor index does not include line loss factor indices related to photovoltaic grid connection. The first line loss prediction model is a Stacking ensemble model, which consists of a photovoltaic grid-connected line loss factor index base learner, a distribution network line loss factor index base learner, and a line loss rate prediction meta learner. The second line loss prediction model is a single-layer deep learning model, which is trained based on a second sample set formed by the second line loss factor index and its corresponding historical line loss rate data.

2. The method for line loss prediction of photovoltaic grid- connected distribution network according to claim 1, characterized in that, The determination of the impact of photovoltaic grid connection on line loss based on the first impact characteristic includes: Calculate the first degree of influence of the access location on the target distribution network line loss rate; Calculate the second degree of influence of access capacity on the target distribution network line loss rate; Calculate the third degree of influence of the proportion of electricity generated in the grid on the target distribution network line loss rate; The impact of photovoltaic grid connection on line loss is obtained based on the first, second, and third degree of impact.

3. The method for predicting line losses in a photovoltaic grid-connected distribution network according to claim 1 or 2, characterized in that, The steps for determining the first line loss factor index include: Obtain sample data for the first line loss indicator; A first reference sequence is generated based on the actual line loss data in the first line loss index sample data; A first comparison sequence is generated based on the first line loss index data related to photovoltaic grid connection in the first line loss index sample data; A second comparison sequence is generated based on the second line loss index data related to the distribution network from the first line loss index sample data; The first theoretical line loss data is calculated based on the first line loss index data and the second line loss index data in the first line loss index sample data, and a second reference sequence is generated based on the first theoretical line loss data; Grey relational analysis was performed on the first comparison sequence, the second comparison sequence, and the first reference sequence to obtain the first correlation degree. Grey relational analysis was performed on the first comparison sequence, the second comparison sequence, and the second reference sequence to obtain the second correlation degree. The first and second correlation degrees are fused to obtain the first fused correlation degree between each line loss index and the line loss rate; The line loss indicators with the highest first degree of fusion correlation are determined as the first line loss factor indicators.

4. The method for predicting line losses in a photovoltaic grid-connected distribution network according to claim 3, characterized in that, The first line loss prediction model is trained based on a first sample set formed by the first line loss factor index and its corresponding historical line loss rate data. Accordingly, the process of using the trained first line loss prediction model to predict line loss and obtain the line loss prediction result includes: The photovoltaic grid-connected line loss factor index data in the first line loss factor index is input into the photovoltaic grid-connected line loss factor index base learner to obtain the first output; Input the distribution network line loss factor index data from the first line loss factor index into the distribution network line loss factor index base learner to obtain the second output; The first and second outputs are input into the line loss rate prediction meta-learner to obtain the line loss prediction result.

5. The method for predicting line losses in a photovoltaic grid-connected distribution network according to claim 1 or 2, characterized in that, The steps for determining the second line loss factor index include: Obtain sample data for the second line loss indicator; A third reference sequence is generated based on the actual line loss data in the second line loss index sample data; A third comparison sequence is generated based on the third line loss index data in the second line loss index sample data; The second theoretical line loss data is calculated based on the third line loss index data in the second line loss index sample data, and the fourth reference sequence is generated based on the second theoretical line loss data; Grey relational analysis was performed on the third comparison sequence and the third reference sequence to obtain the third correlation degree. Grey relational analysis was performed on the third comparison sequence and the fourth reference sequence to obtain the fourth correlation degree. The third and fourth correlation degrees are fused to obtain the second fused correlation degree for each line loss index and line loss rate; The line loss indicators with the highest pre-defined number of second fusion correlations are determined as the second line loss factor indicators.

6. A line loss prediction device for a photovoltaic grid-connected distribution network, characterized in that, include: The impact determination unit is used to obtain the first impact characteristics of photovoltaic grid connection and determine the impact of photovoltaic grid connection on line loss based on the first impact characteristics. The first impact characteristics include the access location, access capacity and the proportion of grid-connected electricity. The first prediction unit is used to predict line loss based on a pre-determined first line loss factor index that affects line loss, and to obtain the line loss prediction result if the impact is greater than a set threshold. The second prediction unit is used to predict line loss based on the pre-determined second line loss factor index that affects line loss, and to obtain the line loss prediction result if the impact is not greater than a set threshold. The first line loss factor index includes line loss factor indices related to photovoltaic grid connection, while the second line loss factor index does not include line loss factor indices related to photovoltaic grid connection. The first line loss prediction model is a Stacking ensemble model, which consists of a photovoltaic grid-connected line loss factor index base learner, a distribution network line loss factor index base learner, and a line loss rate prediction meta learner. The second line loss prediction model is a single-layer deep learning model, which is trained based on a second sample set formed by the second line loss factor index and its corresponding historical line loss rate data.

7. The line loss prediction device for photovoltaic grid-connected distribution networks according to claim 6, characterized in that, The influence determination unit is specifically used for: Calculate the first degree of influence of the access location on the target distribution network line loss rate; Calculate the second degree of influence of access capacity on the target distribution network line loss rate; Calculate the third degree of influence of the proportion of electricity generated in the grid on the target distribution network line loss rate; The impact of photovoltaic grid connection on line loss is obtained based on the first, second, and third degree of impact.

8. The line loss prediction device for photovoltaic grid-connected distribution networks according to claim 6 or 7, characterized in that, The line loss prediction device further includes a first line loss index determination unit, used for: Obtain sample data for the first line loss indicator; A first reference sequence is generated based on the actual line loss data in the first line loss index sample data; A first comparison sequence is generated based on the first line loss index data related to photovoltaic grid connection in the first line loss index sample data; A second comparison sequence is generated based on the second line loss index data related to the distribution network from the first line loss index sample data; The first theoretical line loss data is calculated based on the first line loss index data and the second line loss index data in the first line loss index sample data, and a second reference sequence is generated based on the first theoretical line loss data; Grey relational analysis was performed on the first comparison sequence, the second comparison sequence, and the first reference sequence to obtain the first correlation degree. Grey relational analysis was performed on the first comparison sequence, the second comparison sequence, and the second reference sequence to obtain the second correlation degree. The first and second correlation degrees are fused to obtain the first fused correlation degree between each line loss index and the line loss rate; The line loss indicators with the highest first degree of fusion correlation are determined as the first line loss factor indicators.

9. A terminal, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method for predicting line losses in a photovoltaic grid-connected distribution network as described in any one of claims 1 to 5.

Images