Scene view digital map-driven intelligent evaluation method and system for power distribution network planning solution

Abstract

The present invention relates to the field of intelligent power distribution network planning evaluation. Disclosed are a scene view digital map-driven intelligent evaluation method and system for a power distribution network planning solution. The method comprises: acquiring a scene view digital map image; using a stacked autoencoder-based U-Net convolutional neural network land parcel identification policy to acquire land parcel information; and, on the basis of an indicator hierarchical structure and data-driven evaluation, selecting a candidate planning solution. The present invention can accurately infer the area and load curve of each land parcel, thereby improving the accuracy of evaluation; not only improves training efficiency, but also reduces the complexity of manually adjusting network structures, thus rendering evaluation processes more efficient and automated; can more comprehensively consider the requirements of power distribution systems in various aspects; provides clear bases and rankings for decision-making; and not only technically improves the accuracy and efficiency of evaluating power distribution network planning solutions, but also provides important support and guidance for the optimization and future development of power distribution systems.

Description

A Smart Evaluation Method and System for Distribution Network Planning Schemes Driven by Real-Scene Digital Maps Technical Field

[0001] This invention relates to the field of intelligent power distribution network planning and evaluation, and in particular to an intelligent evaluation method and system for distribution network planning schemes driven by real-scene digital maps. Background Technology

[0002] Today, the importance of evaluating power distribution network planning schemes is increasingly prominent. With the advancement of energy transition and changes in energy consumption patterns, traditional power distribution systems are struggling to meet increasingly complex and high-efficiency requirements. Therefore, systematically evaluating different planning schemes can ensure optimal performance of the power distribution system in terms of efficiency, reliability, and sustainability.

[0003] Traditional distribution network planning and evaluation suffers from shortcomings such as inconsistent planning standards, significant subjective influence, and chaotic post-planning operation and management, failing to meet the basic requirements of modern distribution networks. While current distribution network evaluation methods aim to improve the accuracy and reliability of distribution network planning schemes, thus providing a reference for practical applications, they still require excessive manual intervention, limiting the efficiency of planning and evaluation. Machine learning methods, widely used in load forecasting, have not yet been fully utilized in fully automated distribution network evaluation; the intelligence level and accuracy of distribution network evaluation need further improvement. Summary of the Invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] Therefore, the technical problem solved by this invention is: how to improve the efficiency, reliability and adaptability of power distribution systems through effective evaluation of power distribution network planning schemes.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] In a first aspect, embodiments of the present invention provide an intelligent evaluation method for power distribution network planning schemes driven by real-scene digital maps, including:

[0008] We obtain real-world map data and land parcel type information from the map as the raw dataset, and build a U-Net convolutional neural network using a self-stacked encoder architecture. The raw dataset is then used as the input for training the convolutional neural network.

[0009] A wake-up-sleep mechanism is used to train the convolutional neural network;

[0010] Using the parameters of the convolutional neural network trained with a wake-up-sleep mechanism, the real-scene map data of the area to be planned is input into the convolutional neural network to obtain the area data of each type of land parcel;

[0011] The load curves for each plot are obtained by multiplying the typical load curves and load densities of each plot with the plot area obtained in advance.

[0012] Based on the load curves of each plot, for each candidate planning scheme, four indicators are calculated: equipment utilization rate, power supply radius, renewable energy installed capacity, and peak-valley difference.

[0013] The weights of the four indicators are calculated using a data-driven information entropy weighting method, and the final scores of the candidate planning schemes are calculated to complete the evaluation.

[0014] As a preferred method for intelligent evaluation of power distribution network planning schemes driven by real-scene digital maps, the following is provided:

[0015] The method of training the convolutional neural network using a wake-up-sleep mechanism includes:

[0016] The training is divided into two phases: wake-up and sleep; let R... i Let L represent the i-th layer of the convolutional neural network, and L represent the number of layers in the convolutional neural network.

[0017] During the wake-up phase, the first operation is performed: after the second convolutional neural network R2, a virtual layer R3 is added that is structurally symmetrical to the first convolutional neural network R1. virtual R1, R2, R3 virtual The original dataset serves as the network to be trained in this step; it is used as R1, R2, and R3 simultaneously. virtual The input and output are used for training; after training, the weight parameters of R1 are obtained, and the output of R2 is used as the input for the next training step.

[0018] Second operation in the wake-up phase: Let l = 2;

[0019] The third operation in the wake-up phase: in R l+1 Later supplements and R l symmetrical virtual layer R l+2 virtual , will R l+1 R l R l+2 virtual As the network to be trained in this step; R in the first operation of the wake-up phase l The output result is also used as R l+1 R l R l+2 virtual The input and output are used for training; after training, R is obtained. l The weight parameters are determined, and R is obtained. l+1 The output is used as the input for the next training step;

[0020] The fourth operation of the wake-up phase: Let l = l + 1. If l < L, then jump to the third operation of the wake-up phase; if l = L, then the wake-up phase training is complete and enter the sleep phase.

[0021] As a preferred method for intelligent evaluation of power distribution network planning schemes driven by real-scene digital maps, the following is provided:

[0022] The method of using a wake-up-sleep mechanism to train a convolutional neural network also includes:

[0023] During the sleep phase, the first operation is performed: the weights of each layer of the convolutional neural network obtained during the wake-up phase are used as the initial weights.

[0024] The second operation during the sleep phase: setting the input and output of the convolutional neural network based on the original dataset;

[0025] The third operation in the sleep phase: Use backpropagation to train the convolutional neural network to calculate the final weights.

[0026] As a preferred method for intelligent evaluation of power distribution network planning schemes driven by real-scene digital maps, the following is provided:

[0027] For each candidate planning scheme, the calculation of four indicators includes: equipment utilization rate, power supply radius, renewable energy installed capacity, and peak-valley difference.

[0028] Equipment utilization rate: Calculate the sum of the load curves of the plots under each substation to obtain the substation load curve E. ij trans Let represent the load value of the i-th substation at time j; the formula for calculating equipment utilization rate is as follows: A1 = average(max(E ij trans ) / C i trans )

[0029] Among them, C i trans Let represent the capacity of the i-th substation, max be the function to find the maximum value, and average be the function to find the average value.

[0030] Power supply radius: Calculate the average distance from each plot of land to its respective substation. The formula for calculating the power supply radius is as follows: A2 = average(R ij )

[0031] Among them, R ij This represents the distance from the i-th plot of land to the j-th substation.

[0032] As a preferred method for intelligent evaluation of power distribution network planning schemes driven by real-scene digital maps, the following is provided:

[0033] The calculation of the four indicators—equipment utilization rate, power supply radius, renewable energy installed capacity, and peak-valley difference—for each candidate planning scheme also includes:

[0034] The installed capacity of renewable energy is expressed as: A3 = sum(S) i )

[0035] Among them, S i Let represent the installed capacity of renewable energy in the i-th plot, and sum be the summation function;

[0036] The peak-to-valley difference is expressed as: A4 = average(max(E ij trans )-min(E ij trans ))

[0037] Where min is the function for finding the minimum value.

[0038] As a preferred method for intelligent evaluation of power distribution network planning schemes driven by real-scene digital maps, the following is provided:

[0039] The data-driven information entropy weight calculation method for calculating the weights of the four indicators includes:

[0040] The weight W of the k-th indicator is calculated using a data-driven information entropy weight calculation method. k W k =(1-D k ) / (n-sum(D k D k = -ln(n) -1 ×sum(A ik ×ln(A ik ))

[0041] Among them, D k Let A represent the information entropy of the k-th indicator, n represent the number of indicators, and A ik This represents the value of the k-th indicator in the i-th planning scheme.

[0042] As a preferred method for intelligent evaluation of power distribution network planning schemes driven by real-scene digital maps, the following is provided:

[0043] The final score for calculating the candidate planning schemes includes: T i =sum(W k ×A ik )

[0044] Among them, T i Let represent the final score of the i-th scheme.

[0045] Secondly, embodiments of the present invention provide an intelligent evaluation system for power distribution network planning schemes driven by real-scene digital maps, comprising:

[0046] The acquisition module is used to acquire real-scene map data and land parcel type information in the map as the raw dataset. A U-Net convolutional neural network is built using a self-stacked encoder architecture, and the raw dataset is used as the input for training the convolutional neural network.

[0047] The training module is used to train the convolutional neural network using a wake-up-sleep mechanism;

[0048] The area data calculation module is used to input the real-scene map data of the area to be planned into the convolutional neural network by using the parameters of the convolutional neural network trained with a wake-up-sleep mechanism, and obtain the area data of each type of land parcel.

[0049] The load curve calculation module is used to multiply the typical load curves and load densities of each plot in advance with the obtained plot area to obtain the load curves of each plot.

[0050] The indicator calculation module is used to calculate four indicators—equipment utilization, power supply radius, renewable energy installed capacity, and peak-valley difference—for each candidate planning scheme based on the load curves of each plot.

[0051] The evaluation module is used to calculate the weights of the four indicators using a data-driven information entropy weight calculation method, and to calculate the final score of the candidate planning schemes to complete the evaluation.

[0052] Thirdly, embodiments of the present invention provide a computing device, including:

[0053] Memory and processor;

[0054] The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the one or more programs are executed by the one or more processors, the one or more processors implement the intelligent evaluation method for power distribution network planning scheme driven by real-scene digital map as described in any embodiment of the present invention.

[0055] Fourthly, embodiments of the present invention provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the aforementioned intelligent evaluation method for power distribution network planning schemes driven by real-scene digital maps.

[0056] The beneficial effects of this invention are as follows: By using real-world map data and land parcel type information, combined with deep learning technology, especially the U-Net convolutional neural network, this invention can accurately infer the area and load curve of each land parcel, thereby improving the accuracy of the assessment. Introducing a wake-up-sleep mechanism to train the convolutional neural network not only improves training efficiency but also reduces the complexity of manually adjusting the network structure, making the assessment process more efficient and automated. By calculating multiple indicators such as equipment utilization, power supply radius, renewable energy installed capacity, and peak-valley difference, the advantages and disadvantages of various planning schemes are comprehensively evaluated, allowing for a more comprehensive consideration of the various needs of the power distribution system. The use of an information entropy weight calculation method analyzes the weights of each indicator based on actual data, making the assessment more objective and scientific. By combining the weights of each indicator with the data results of specific schemes, the final score is calculated, providing a clear basis and ranking for decision-making. This method not only technically improves the accuracy and efficiency of power distribution network planning scheme assessment but also better adapts to increasingly complex and high-efficiency energy demands, providing important support and guidance for the optimization and future development of power distribution systems. Attached Figure Description

[0057] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. 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.

[0058] Figure 1 is an overall flowchart of the intelligent evaluation method for power distribution network planning scheme driven by real-scene digital map according to the first embodiment of the present invention.

[0059] Figure 2 is a U-Net convolutional neural network structure diagram of the intelligent evaluation method for power distribution network planning driven by real-scene digital map according to the first embodiment of the present invention;

[0060] Figure 3 shows the U-Net recognition results of the planning area and a schematic diagram of the planning scheme in a simulation example of the intelligent evaluation method for power distribution network planning scheme driven by real-scene digital map according to the second embodiment of the present invention.

[0061] Figure 4 is a schematic diagram of the weight calculation results in a simulation example of the intelligent evaluation method for power distribution network planning driven by real-scene digital map according to the second embodiment of the present invention. Detailed Implementation

[0062] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0063] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0064] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0065] Example 1

[0066] Referring to Figures 1-2, the first embodiment of the present invention provides a method for intelligent evaluation of power distribution network planning schemes driven by real-scene digital maps, including:

[0067] S1: Obtain real-world map data and land parcel type information from the map as the raw dataset. Build a U-Net convolutional neural network using a self-stacked encoder architecture. Use the raw dataset as the input for training the convolutional neural network. Employ a wake-up-sleep mechanism to train the convolutional neural network.

[0068] In this embodiment of the application, the training of the convolutional neural network using a wake-up-sleep mechanism includes:

[0069] The training is divided into two phases: wake-up and sleep; let R... i Let L represent the i-th layer of the convolutional neural network, and L represent the number of layers in the convolutional neural network.

[0070] During the wake-up phase, the first operation is performed: after the second convolutional neural network R2, a virtual layer R3 is added that is structurally symmetrical to the first convolutional neural network R1. virtual R1, R2, R3 virtual The original dataset serves as the network to be trained in this step; it is used as R1, R2, and R3 simultaneously. virtual The input and output are used for training; after training, the weight parameters of R1 are obtained, and the output of R2 is used as the input for the next training step.

[0071] Second operation in the wake-up phase: Let l = 2;

[0072] The third operation in the wake-up phase: in R l+1 Later supplements and R l symmetrical virtual layer R l+2 virtual , will R l+1 R l R l+2 virtual As the network to be trained in this step; R in the first operation of the wake-up phase l The output result is also used as R l+1 R l R l+2 virtual The input and output are used for training; after training, R is obtained. l The weight parameters are determined, and R is obtained. l+1 The output is used as the input for the next training step;

[0073] The fourth operation of the wake-up phase: Let l = l + 1. If l < L, then jump to the third operation of the wake-up phase; if l = L, then the wake-up phase training is complete and enter the sleep phase.

[0074] During the sleep phase, the first operation is performed: the weights of each layer of the convolutional neural network obtained during the wake-up phase are used as the initial weights.

[0075] The second operation during the sleep phase: setting the input and output of the convolutional neural network based on the original dataset;

[0076] The third operation in the sleep phase: Use backpropagation to train the convolutional neural network to calculate the final weights.

[0077] It should be noted that the wake-up-sleep mechanism effectively improves the training efficiency and convergence speed of convolutional neural networks, enabling the model to reach better performance levels more quickly in complex tasks. By using real-world map data and land parcel type information as input, combined with the convolutional neural network trained by the wake-up-sleep mechanism, the model can more accurately infer the area and load curve of the land parcel, thereby improving the accuracy and reliability of the evaluation results. This not only improves the automation and efficiency of power distribution network planning scheme evaluation technically, but also fully considers the integration and operability of the model during implementation, providing a reliable solution for practical applications.

[0078] S2: Using the parameters of the convolutional neural network trained with a wake-up-sleep mechanism, input the real-scene map data of the area to be planned into the convolutional neural network to obtain the area data of each type of land parcel;

[0079] Specifically, the real-scene map data needs to be preprocessed: ensure that the real-scene map data of the area to be planned is consistent with the data type and format used during training, including image resolution, color space, etc.; embed the land parcel type information into the input data in an appropriate format to ensure that the convolutional neural network can effectively identify and analyze different types of land parcels.

[0080] It should be noted that the convolutional neural network obtained through training can accurately infer the area data of each plot, which is one of the key inputs for the subsequent load curve calculation and evaluation process.

[0081] S3: Based on the typical load curves and load densities of each plot obtained in advance, multiply them by the plot area to obtain the load curves of each plot;

[0082] It should be noted that by combining typical load curves and land parcel area data, the resulting load curves have high accuracy and reliability, and can accurately reflect the load characteristics of each land parcel. The load curves are important input parameters when evaluating each candidate planning scheme, and their accurate calculation helps to more accurately assess the power demand and system load of each scheme.

[0083] S4: Based on the load curves of each plot, calculate four indicators for each candidate planning scheme: equipment utilization rate, power supply radius, renewable energy installed capacity, and peak-valley difference.

[0084] In this embodiment of the application, for each candidate planning scheme, the calculation of four indicators includes: equipment utilization rate, power supply radius, renewable energy installed capacity, and peak-valley difference.

[0085] Equipment utilization rate: Calculate the sum of the load curves of the plots under each substation to obtain the substation load curve E. ij trans Let represent the load value of the i-th substation at time j; the formula for calculating equipment utilization rate is as follows: A1 = average(max(E ij trans ) / C i trans )

[0086] Among them, C i trans Let represent the capacity of the i-th substation, max be the function to find the maximum value, and average be the function to find the average value.

[0087] Power supply radius: Calculate the average distance from each plot of land to its respective substation. The formula for calculating the power supply radius is as follows: A2 = average(R ij )

[0088] Among them, R ijThis represents the distance from the i-th plot of land to the j-th substation.

[0089] The installed capacity of renewable energy is expressed as: A3 = sum(S) i )

[0090] Among them, S i Let represent the installed capacity of renewable energy in the i-th plot, and sum be the summation function;

[0091] The peak-to-valley difference is expressed as: A4 = average(max(E ij trans )-min(E ij trans ))

[0092] Where min is the function for finding the minimum value.

[0093] S5: Calculate the weights of the four indicators using a data-driven information entropy weighting method, and calculate the final score of the candidate planning schemes to complete the evaluation.

[0094] In this embodiment of the application, the calculation of the weights of the four indicators using a data-driven information entropy weight calculation method includes:

[0095] The weight W of the k-th indicator is calculated using a data-driven information entropy weight calculation method. k W k =(1-D k ) / (n-sum(D k D k = -ln(n) -1 ×sum(A ik ×ln(A ik ))

[0096] Among them, D k Let A represent the information entropy of the k-th indicator, n represent the number of indicators, and A ik This represents the value of the k-th indicator in the i-th planning scheme.

[0097] The final score for each candidate planning scheme includes: T i =sum(W k ×A ik )

[0098] Among them, T i Let represent the final score of the i-th scheme.

[0099] It should be noted that using the information entropy method to calculate weights can objectively reflect the contribution of each indicator to the final evaluation result, avoiding the bias of subjective weighting; comprehensively considering multiple indicators such as equipment utilization rate, power supply radius, renewable energy installed capacity and peak-valley difference, it ensures the comprehensiveness and integration of the evaluation results; the final score provides an intuitive and quantitative evaluation result, providing decision-makers with a clear basis for ranking planning schemes, supporting the scientific and operable nature of decision-making.

[0100] The above is an illustrative scheme of the intelligent evaluation method for distribution network planning driven by real-scene digital maps in this embodiment. It should be noted that the technical solution of the intelligent evaluation system for distribution network planning driven by real-scene digital maps and the technical solution of the intelligent evaluation method for distribution network planning driven by real-scene digital maps belong to the same concept. Details not described in detail in the technical solution of the intelligent evaluation system for distribution network planning driven by real-scene digital maps in this embodiment can be found in the description of the technical solution of the intelligent evaluation method for distribution network planning driven by real-scene digital maps described above.

[0101] The intelligent evaluation system for power distribution network planning schemes driven by real-scene digital maps in this embodiment includes:

[0102] The acquisition module is used to acquire real-scene map data and land parcel type information in the map as the raw dataset. A U-Net convolutional neural network is built using a self-stacked encoder architecture, and the raw dataset is used as the input for training the convolutional neural network.

[0103] The training module is used to train the convolutional neural network using a wake-up-sleep mechanism;

[0104] The area data calculation module is used to input the real-scene map data of the area to be planned into the convolutional neural network by using the parameters of the convolutional neural network trained with a wake-up-sleep mechanism, and obtain the area data of each type of land parcel.

[0105] The load curve calculation module is used to multiply the typical load curves and load densities of each plot in advance with the obtained plot area to obtain the load curves of each plot.

[0106] The indicator calculation module is used to calculate four indicators—equipment utilization, power supply radius, renewable energy installed capacity, and peak-valley difference—for each candidate planning scheme based on the load curves of each plot.

[0107] The evaluation module is used to calculate the weights of the four indicators using a data-driven information entropy weight calculation method, and to calculate the final score of the candidate planning schemes to complete the evaluation.

[0108] This embodiment also provides a computing device applicable to intelligent evaluation methods for power distribution network planning schemes driven by real-scene digital maps, including:

[0109] The system includes a memory and a processor. The memory stores computer-executable instructions, and the processor executes these instructions to implement the intelligent evaluation method for power distribution network planning driven by real-scene digital maps, as proposed in the above embodiments.

[0110] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, it implements the intelligent evaluation method for power distribution network planning scheme driven by real-scene digital map as proposed in the above embodiment.

[0111] The storage medium proposed in this embodiment and the intelligent evaluation method for power distribution network planning driven by real-scene digital maps proposed in the above embodiments belong to the same inventive concept. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.

[0112] Example 2

[0113] Referring to Figures 3-4 and Table 1, an embodiment of the present invention provides an intelligent evaluation method for power distribution network planning driven by a real-scene digital map. To verify the beneficial effects of the present invention, a simulation experiment is conducted for scientific demonstration.

[0114] 1. Data preparation and preprocessing

[0115] Obtain real-world map data and land parcel type information as the raw dataset; prepare typical load curves and load density data.

[0116] 2. Construction of U-Net Convolutional Neural Network

[0117] A U-Net-structured convolutional neural network was constructed using an autoencoder; real-world map data was used as training input to predict the area data of various land parcel types.

[0118] 3. Wake-Sleep Mechanism Training

[0119] Awakening phase:

[0120] Virtual layers are added sequentially, and the network is trained to obtain the weight parameters for each layer.

[0121] Sleep stages:

[0122] The weight parameters obtained during the wake-up phase are used as initial weights; the backpropagation algorithm is used to further optimize the network parameters.

[0123] 4. Load curve calculation

[0124] Based on the predicted plot area data and load density, the load curve for each plot is calculated.

[0125] 5. Performance Index Calculation

[0126] For each candidate planning scheme, calculate the following four indicators: equipment utilization rate A 1、 Power supply radius A 2、 Renewable energy installed capacity A 3、 Peak-to-valley difference A4.

[0127] 6. Information Entropy Weight Calculation

[0128] The weights of the four indicators mentioned above are calculated using a data-driven information entropy weight calculation method.

[0129] 7. Evaluation of planning schemes

[0130] Based on the calculated weights and indicator values, the final score T for each planning scheme is calculated. i .

[0131] Figure 3 shows the land parcel identification and planning scheme of the planning area. This information directly affects the input data and training process of the U-Net model. Figure 4 shows the weight calculation results, which affect the final evaluation of the planning scheme. The weights calculated using the information entropy method reflect the importance of different indicators in the evaluation. The following are some experimental results:

[0132] Table 1 Partial Experimental Results

[0133] This simulation example demonstrates the effectiveness and practicality of the intelligent evaluation method for power distribution network planning driven by real-world digital maps, providing important technical support and decision-making basis for intelligent planning in practical application scenarios.

[0134] It should be noted that the above 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 preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for intelligent evaluation of power distribution network planning schemes driven by real-scene digital maps, characterized in that, include: We obtain real-world map data and land parcel type information from the map as the raw dataset, and build a U-Net convolutional neural network using a self-stacked encoder architecture. The raw dataset is then used as the input for training the convolutional neural network. A wake-up-sleep mechanism is used to train the convolutional neural network; Using the parameters of the convolutional neural network trained with a wake-up-sleep mechanism, the real-scene map data of the area to be planned is input into the convolutional neural network to obtain the area data of each type of land parcel; The load curves for each plot are obtained by multiplying the typical load curves and load densities of each plot with the plot area obtained in advance. Based on the load curves of each plot, for each candidate planning scheme, four indicators are calculated: equipment utilization rate, power supply radius, renewable energy installed capacity, and peak-valley difference. The weights of the four indicators are calculated using a data-driven information entropy weighting method, and the final scores of the candidate planning schemes are calculated to complete the evaluation.

2. The intelligent evaluation method for power distribution network planning schemes driven by real-scene digital maps as described in claim 1, characterized in that, The method of training the convolutional neural network using a wake-up-sleep mechanism includes: The training is divided into two phases: wake-up and sleep; let R... i Let L represent the i-th layer of the convolutional neural network, and L represent the number of layers in the convolutional neural network. During the wake-up phase, the first operation is performed: after the second convolutional neural network R2, a virtual layer R3 is added that is structurally symmetrical to the first convolutional neural network R1. virtual R1, R2, R3 virtual The original dataset serves as the network to be trained in this step; it is used as R1, R2, and R3 simultaneously. virtual The input and output are used for training; after training, the weight parameters of R1 are obtained, and the output of R2 is used as the input for the next training step. Second operation in the wake-up phase: Let l = 2; The third operation in the wake-up phase: in R l+1 Later supplements and R l symmetrical virtual layer R l+2 virtual , will R l+1 R l R l+2 virtual As the network to be trained in this step; R in the first operation of the wake-up phase l The output result is also used as R l+1 R l R l+2 virtual The input and output are used for training; after training, R is obtained. l The weight parameters are determined, and R is obtained. l+1 The output is used as the input for the next training step; The fourth operation of the wake-up phase: Let l = l + 1. If l < L, then jump to the third operation of the wake-up phase; if l = L, then the wake-up phase training is complete and enter the sleep phase.

3. The intelligent evaluation method for power distribution network planning schemes driven by real-scene digital maps as described in claim 2, characterized in that, The method of using a wake-up-sleep mechanism to train a convolutional neural network also includes: During the sleep phase, the first operation is performed: the weights of each layer of the convolutional neural network obtained during the wake-up phase are used as the initial weights. The second operation during the sleep phase: setting the input and output of the convolutional neural network based on the original dataset; The third operation in the sleep phase: Use backpropagation to train the convolutional neural network to calculate the final weights.

4. The intelligent evaluation method for power distribution network planning schemes driven by real-scene digital maps as described in claim 3, characterized in that, For each candidate planning scheme, the calculation of four indicators includes: equipment utilization rate, power supply radius, renewable energy installed capacity, and peak-valley difference. Equipment utilization rate: Calculate the sum of the load curves of the plots under each substation to obtain the substation load curve E. ij trans Let represent the load value of the i-th substation at time j; the formula for calculating equipment utilization rate is as follows: in, Let represent the capacity of the i-th substation, max be the function to find the maximum value, and average be the function to find the average value. Power supply radius: Calculate the average distance from each plot of land to its respective substation. The formula for calculating the power supply radius is as follows: A2=average(R ij ) Among them, R ij This represents the distance from the i-th plot of land to the j-th substation.

5. The intelligent evaluation method for power distribution network planning schemes driven by real-scene digital maps as described in claim 4, characterized in that, The calculation of the four indicators—equipment utilization rate, power supply radius, renewable energy installed capacity, and peak-valley difference—for each candidate planning scheme also includes: Renewable energy installed capacity is expressed as: A3=sum(S i ) Among them, S i Let represent the installed capacity of renewable energy in the i-th plot, and sum be the summation function; Peak-to-valley difference, expressed as: A4 = average(max(E ij trans )-min(E ij trans )) Where min is the function for finding the minimum value.

6. The intelligent evaluation method for power distribution network planning schemes driven by real-scene digital maps as described in claim 5, characterized in that, The data-driven information entropy weight calculation method for calculating the weights of the four indicators includes: The weight W of the k-th indicator is calculated using a data-driven information entropy weight calculation method. k : W k =(1-D k ) / (n-sum(D k )) D k =-ln(n) -1 ×sum(A ik ×ln(A ik )) Among them, D k Let A represent the information entropy of the k-th indicator, n represent the number of indicators, and A ik This represents the value of the k-th indicator in the i-th planning scheme.

7. The intelligent evaluation method for power distribution network planning schemes driven by real-scene digital maps as described in claim 6, characterized in that, The final score for calculating the candidate planning schemes includes: T i =sum(W k ×A ik ) Among them, T i Let represent the final score of the i-th scheme.

8. A system employing a real-scene digital map-driven intelligent evaluation method for power distribution network planning schemes as described in any one of claims 1 to 7, characterized in that, include: The acquisition module is used to acquire real-scene map data and land parcel type information in the map as the raw dataset. A U-Net convolutional neural network is built using a self-stacked encoder architecture, and the raw dataset is used as the input for training the convolutional neural network. The training module is used to train the convolutional neural network using a wake-up-sleep mechanism; The area data calculation module is used to input the real-scene map data of the area to be planned into the convolutional neural network by using the parameters of the convolutional neural network trained with a wake-up-sleep mechanism, and obtain the area data of each type of land parcel. The load curve calculation module is used to multiply the typical load curves and load densities of each plot in advance with the obtained plot area to obtain the load curves of each plot. The indicator calculation module is used to calculate four indicators—equipment utilization, power supply radius, renewable energy installed capacity, and peak-valley difference—for each candidate planning scheme based on the load curves of each plot. The evaluation module is used to calculate the weights of the four indicators using a data-driven information entropy weight calculation method, and to calculate the final score of the candidate planning schemes to complete the evaluation.

9. A computing device, comprising: Memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, they implement the steps of the intelligent evaluation method for power distribution planning scheme driven by real-scene digital map as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing computer-executable instructions that, when executed by a processor, implement the steps of the intelligent evaluation method for power distribution planning schemes driven by real-scene digital maps as described in any one of claims 1 to 7.

Images