Method for analyzing access bearing capacity of low-voltage power distribution network

Abstract

The invention provides a low-voltage distribution network access bearing capacity analysis method, and relates to the technical field of power grids. The method comprises the following steps: S1, acquiring historical data within a preset time; s2, preprocessing the historical data, forming a data set, and dividing the data set into a model training set and a model test set; s3, constructing a model based on machine learning based on the model training set and the model test set; s4, calling the model to calculate the number of electric vehicles of the target low-voltage distribution network in the planning year, the charging facility data of the target low-voltage distribution network in the planning year, the charging load data of the target low-voltage distribution network in the planning year and the charging load data of the target low-voltage distribution network under a vehicle-network interaction strategy; and S5, analyzing to obtain the low-voltage distribution network access bearing capacity in the planning year. The method can improve the scientificity and credibility of the planning scheme, supports the early discovery of potential problems of the low-voltage distribution network, and improves the safety, reliability and overall quality.

Description

Technical Field

[0001] The present invention relates to the technical field of power grids, and in particular to a method for analyzing access carrying capacity of a low-voltage distribution network. Background Art

[0002] As the number of new energy vehicles increases rapidly, the development of charging piles is also rapid, and the charging volume and charging load are growing rapidly. With the rapid growth of charging volume and charging load, the load brought by charging facilities will become an important consideration for medium-voltage and low-voltage distribution networks, directly affecting important indicators such as power supply reliability, power quality, and new energy consumption. It is a new challenge faced by multiple professional departments of the power grid, including planning, and is also the key research content of the new generation of smart distribution networks.

[0003] Traditional distribution network planning technology does not consider the load of electric vehicle charging separately, lacks corresponding methods, and has no relevant specifications to adapt to the new demand brought by the growth of electric vehicles. At present, relevant demand parties generally complete this kind of work through expert consultation, and have not formed a methodology, mature system or industry consensus, which brings problems such as high cost, long cycle and large error. Summary of the invention

[0004] The purpose of the present invention includes providing a low-voltage distribution network access carrying capacity analysis method, which can support the improvement of low-voltage distribution network professional planning work, expand its scope of consideration factors, adapt to the new form brought by electric vehicles, improve the scientificity and credibility of the planning scheme, support early discovery of potential problems in the low-voltage distribution network, and improve safety, reliability and overall quality.

[0005] The embodiments of the present invention can be implemented as follows:

[0006] The present invention provides a method for analyzing the access carrying capacity of a low-voltage distribution network, the method comprising:

[0007] S1: Get historical data within a preset time period;

[0008] S2: Preprocess the historical data and form a data set, and divide the data set into a model training set and a model test set;

[0009] S3: Based on the model training set and the model test set, a machine learning-based model is constructed, wherein the model includes an electric vehicle quantity prediction model, a charging facility quantity prediction model, an electric vehicle charging load prediction model, and a quantitative analysis model of the impact of typical vehicle-grid interaction strategies on charging load;

[0010] S4: Call the model to calculate the number of electric vehicles in the target low-voltage distribution network in the planning year, the charging facility data of the target low-voltage distribution network in the planning year, the charging load data of the target low-voltage distribution network in the planning year, and the charging load data of the target low-voltage distribution network under the vehicle-grid interaction strategy;

[0011] S5: Analyze and obtain the access carrying capacity of the low-voltage distribution network in the planned year.

[0012] In an optional implementation, in S1, historical data includes electric vehicle data and power grid data. The electric vehicle data includes the number of electric vehicles in use, electric vehicle market penetration, charging facility data, and vehicle-to-charging pile ratio; the power grid data includes charging load data, power grid load data, average distribution and transformation capacity per household in residential areas, electricity price data, vehicle-grid interaction strategy historical record data set, and low-voltage distribution network planning data.

[0013] In an optional implementation, in S1, the scenes of the acquired historical data include urban residential areas, rural areas, road service areas, public service areas, and solar storage and charging stations;

[0014] Urban residential areas are divided into four sub-scenarios: old communities, communities built between 15 and 25 years old, communities built between 5 and 15 years old, and newly built communities. Public service areas are divided into five sub-scenarios: office areas, commercial centers, industrial centers, rest centers, and public parking areas.

[0015] In an optional implementation, in S3, building an electric vehicle quantity prediction model based on machine learning includes:

[0016] A regression-based machine learning algorithm for time series prediction is used to perform classification predictions according to different vehicle types, and the prediction results for the number of electric vehicles are superimposed.

[0017] In an optional implementation, in S3, building a charging facility quantity prediction model based on machine learning includes:

[0018] According to the two first dimensions of public and private, the four second dimensions of private charging piles, slow charging piles, fast charging piles and super fast charging piles are predicted separately, and the prediction results of the number of charging facilities are obtained by superimposing them.

[0019] In an optional implementation, in S3, building a charging facility quantity prediction model based on machine learning includes:

[0020] A regression-based machine learning algorithm for time series prediction is used to perform supervised machine learning to predict the car-to-charging pile ratio in the planned year. The result data of the electric vehicle quantity prediction model is then used as input to obtain the number of charging piles in the planned year according to the preset formula; the preset formula is:

[0021] Planned annual number of charging piles = ∑ (number of electric vehicles in the sub-scenario * car-to-pile ratio in the sub-scenario).

[0022] In an optional implementation, in S3, constructing a quantitative analysis model of the impact of a typical vehicle-grid interaction strategy on charging load based on machine learning includes:

[0023] Based on the historical record data set of vehicle-grid interaction strategies obtained in S1, data is structured according to the application of strategy x to the scenario, where strategy x includes: time-of-use electricity price, two-way interaction, short-term response, long-term agreement direct control, and photovoltaic storage and charging interaction;

[0024] A multivariate regression machine learning algorithm is used to train a quantitative analysis model of the impact of typical vehicle-grid interaction strategies on charging load based on the model training set.

[0025] In an optional embodiment, S4 includes:

[0026] Call the electric vehicle quantity prediction model based on machine learning to calculate the number of electric vehicles in the target low-voltage distribution network in the planning year;

[0027] Call the charging facility quantity prediction model based on machine learning to calculate the charging facility data of the target low-voltage distribution network in the planning year;

[0028] Calculate the base load of the target low-voltage distribution network in the planning year;

[0029] Call the electric vehicle charging load prediction model based on machine learning to calculate the charging load data of the target low-voltage distribution network in the planning year;

[0030] A quantitative analysis model of the impact of typical vehicle-grid interaction strategies on charging load based on machine learning is called, and the output of the electric vehicle charging load prediction model is used as part of the input to calculate the charging load data of the target low-voltage distribution network under the vehicle-grid interaction strategy.

[0031] In an optional embodiment, S4 includes:

[0032] After constructing the electric vehicle charging load prediction model, the data of the target low-voltage distribution network is sorted and input to obtain the charging load data of the target low-voltage distribution network in the target planning year, and then superimposed on the basic load data of the planning year to obtain the final load data.

[0033] In an optional embodiment, S5 includes:

[0034] The charging load data after the target year strategy of the target low-voltage distribution network is superimposed on its basic load data, and the numerical result is output; the load curve data of the target low-voltage distribution network after the vehicle-grid interaction strategy in the target year is obtained;

[0035] Compare and analyze the distribution transformer capacity input by the user to obtain the maximum load rate of the distribution transformer;

[0036] Analyze whether the maximum load rate of the distribution transformer exceeds the load rate alarm line. If it exceeds the alarm line, output alarm information.

[0037] The beneficial effects of the low-voltage distribution network access carrying capacity analysis method provided by the embodiment of the present invention include:

[0038] Multiple machine learning models were constructed and used to predict the number of electric vehicles, the number of charging facilities, and the charging load. At the same time, a machine learning model with typical vehicle-grid interaction strategy characteristics was constructed to realize the charging load prediction under policy guidance. The low-voltage distribution network capacity was compared to obtain the analysis results of the low-voltage distribution network access carrying capacity in the planned year. The method can support the improvement of the professional planning of low-voltage distribution networks, expand the scope of its considerations, adapt to the new forms brought by electric vehicles, improve the scientificity and credibility of planning schemes, support the early discovery of potential problems in low-voltage distribution networks, and improve safety, reliability and overall quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0040] Figure 1 A flow chart of a method for analyzing the access carrying capacity of a low-voltage distribution network provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0041] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0042] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0043] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0044] It should be noted that, in the absence of conflict, the features in the embodiments of the present invention may be combined with each other.

[0045] Please refer to Figure 1 This embodiment provides a method for analyzing the access carrying capacity of a low-voltage distribution network (hereinafter referred to as the method), and the method includes the following steps:

[0046] S1: Obtain historical data within a preset time, where the historical data includes electric vehicle data and power grid data.

[0047] The preset time can be the past 5 years. Electric vehicle data includes electric vehicle ownership, electric vehicle market penetration, charging facility data, vehicle-to-pile ratio, etc. Grid data includes charging load data, grid load data, average distribution and transformation capacity per household in residential areas, electricity price data, vehicle-grid interaction strategy historical record data set, low-voltage distribution network planning data, etc.

[0048] The scope of historical data acquisition includes data of target planning objects and data of similar objects at home and abroad. The scenes of historical data acquired include urban residential areas, rural areas, road service areas, public service areas, solar storage and charging stations, etc.

[0049] Among them, urban residential areas can be divided into four sub-scenarios: old residential areas, 15-25 year-old residential areas, 5-15 year-old residential areas, and newly built residential areas. The public service area can be further divided into five sub-scenarios: office area, commercial center, industrial center, rest center, and public parking area. The final scene expansion includes: old residential areas in urban residential areas, 15-25 year-old residential areas in urban residential areas, 5-15 year-old residential areas in urban residential areas, newly built residential areas in urban residential areas, rural areas, road service areas, photovoltaic storage and charging stations, office areas in public areas, commercial centers in public areas, industrial centers in public areas, rest centers in public areas, and public parking areas in public areas.

[0050] S2: Preprocess the historical data and form a data set, and divide the data set into a model training set and a model test set.

[0051] Specifically, the historical data is preprocessed according to the training requirements of multiple models in S3 to form multiple data sets. The original data can be reused, the results can be referenced unidirectionally, and the data sets are divided into model training sets and model test sets.

[0052] S3: Build a machine learning-based model based on the model training set and the model test set.

[0053] Among them, the models include electric vehicle quantity prediction model, charging facility quantity prediction model, electric vehicle charging load prediction model and quantitative analysis model of the impact of typical vehicle-grid interaction strategies on charging load.

[0054] Specifically, a model for predicting the number of electric vehicles based on machine learning is constructed; a model for predicting the number of charging facilities based on machine learning is constructed; a model for predicting the charging load of electric vehicles based on machine learning is constructed; and a quantitative analysis model for the impact of typical vehicle-grid interaction strategies on charging load is constructed based on machine learning.

[0055] Based on the model training set in S2, the electric vehicle quantity prediction model, the charging facility quantity prediction model, the electric vehicle charging load prediction model and the quantitative analysis model of the impact of typical vehicle-grid interaction strategies on the charging load are trained, and based on the model testing set in S2, the electric vehicle quantity prediction model, the charging facility quantity prediction model, the electric vehicle charging load prediction model and the quantitative analysis model of the impact of typical vehicle-grid interaction strategies on the charging load are tested until the model meets the test requirements.

[0056] Among them, the construction of the electric vehicle quantity prediction model based on machine learning includes the use of regression machine learning algorithms for time series prediction, classification prediction according to different vehicle types, and superposition to obtain the prediction results of the number of electric vehicles. The prediction results of the number of electric vehicles include the number of electric vehicles of types such as buses, passenger and tourist buses, online taxis, official passenger cars, private passenger cars, light logistics vehicles, medium and heavy trucks (including special vehicles such as sanitation), etc. In this way, the model can eventually input the historical data of the number of electric vehicles within the target low-voltage distribution network, and output the prediction data of the number of electric vehicles in the target low-voltage distribution network in the planning year.

[0057] Building a charging facility quantity prediction model based on machine learning includes predicting the number of charging facilities according to the two first dimensions of public and private, and the four second dimensions of private charging piles, slow charging piles, fast charging piles, and super fast charging piles, and superimposing them to obtain the prediction results of the number of charging facilities.

[0058] Building a prediction model for the number of charging facilities based on machine learning includes using a regression-based machine learning algorithm for time series prediction to perform supervised machine learning and predict the car-to-pile ratio for the planned year. The result data of the electric vehicle number prediction model is then used as input to obtain the number of charging piles for the planned year according to the preset formula. The model can ultimately input historical data on the number of charging facilities within the target low-voltage distribution network and predicted data on the number of electric vehicles, and output predicted data on the number of charging facilities within the target low-voltage distribution network for the planned year. The preset formula is:

[0059] Planned annual number of charging piles = ∑ (number of electric vehicles in the sub-scenario * ratio of vehicles to charging piles in the sub-scenario)

[0060] The expression of the electric vehicle charging load prediction model is as follows:

[0061] Charging load at a certain moment = predicted number of charging facilities × probability of charging facilities charging at the same time × average power of charging facilities

[0062] Among them, the predicted value of the number of charging facilities is the output value of the charging facility number prediction model. A regression-based machine learning algorithm for time series prediction is used to perform supervised machine learning based on historical data to predict the probability of simultaneous charging of charging facilities at a certain moment and the average power of charging facilities. Among them, the probability of simultaneous charging of charging facilities at a certain moment can be extracted from the charging load data obtained from S1; the average power of charging facilities can be extracted from the charging facility data obtained from S1. Through the expression of the electric vehicle charging load prediction model, the charging load at a certain moment can be calculated, and the process is repeated at a first preset time interval (which can be 15 minutes to 1 hour) to complete the construction of the electric vehicle charging load prediction model. The model can eventually input the target low-voltage distribution network electric vehicle charging load historical data and charging facility data prediction data, and output the planning year target low-voltage distribution network electric vehicle charging load prediction data.

[0063] Constructing a quantitative analysis model of the impact of typical vehicle-grid interaction strategies on charging load based on machine learning includes applying the vehicle-grid interaction strategy historical record dataset obtained in S1 to the scenario for data structuring according to strategy x, where strategy x includes: time-of-use electricity price, two-way interaction, short-term response, long-term agreement direct control, and photovoltaic storage and charging interaction. The scenario can be seen as shown in S1.

[0064] Among them, the vehicle-grid interaction strategy historical record data set can be structured into data: strategy type, action scenario, active electricity price time period, subsidy price, typical day charging load before strategy, typical day total load before strategy, typical day charging load after strategy, and typical day total load after strategy. Among them, the typical day charging load before strategy, the typical day total load before strategy, the typical day charging load after strategy, and the typical day total load after strategy are all vector data with the typical day as the second preset time interval (generally 15 minutes to 1 hour). Taking 15 minutes as an example, the vector has 96 points. The data used for training and verification models need to be manually labeled. Using a multivariate regression machine learning algorithm, based on the model training set, a quantitative analysis model of the impact of typical vehicle-grid interaction strategies on charging load is trained. The model can eventually input low-voltage distribution network basic load data, low-voltage distribution network charging load data, and vehicle-grid interaction strategy parameters, and output low-voltage distribution network charging load data after the strategy.

[0065] S4: Call the model to calculate the number of electric vehicles in the target low-voltage distribution network in the planning year, the charging facility data of the target low-voltage distribution network in the planning year, the charging load data of the target low-voltage distribution network in the planning year, and the charging load data of the target low-voltage distribution network under the vehicle-grid interaction strategy.

[0066] Among them, the electric vehicle quantity prediction model based on machine learning is called to calculate the number of electric vehicles in the target low-voltage distribution network in the planning year. The charging facility quantity prediction model based on machine learning is called to calculate the charging facility data of the target low-voltage distribution network in the planning year. The traditional mature solution is used to calculate the basic load of the target low-voltage distribution network in the planning year; the electric vehicle charging load prediction model based on machine learning is called to calculate the charging load data of the target low-voltage distribution network in the planning year. The quantitative analysis model of the impact of typical vehicle-grid interaction strategies on charging load based on machine learning is called, and the output of the electric vehicle charging load prediction model is used as part of the input to calculate the charging load data of the target low-voltage distribution network under the vehicle-grid interaction strategy.

[0067] The calculation method of each parameter is as follows:

[0068] (1) Number of electric vehicles in the planning year = Number of electric vehicles in the current year × average annual growth rate^(n)

[0069] Where n is the difference between the planning year and the current year.

[0070] (2) Charging facility data for the planning year = ∑ 不同场景 (Number of electric vehicles planned for the sub-scenario in a year * Average number of kilometers driven per day in the sub-scenario * Average power consumption per 100 kilometers in the sub-scenario / Average charging power per pile in the sub-scenario)

[0071] (3) The annual charging rated load data of the sub-scenario plan = the number of charging facilities planned for the sub-scenario in the year * the average charging power of a single pile in the sub-scenario

[0072] Sub-scenario planned annual charging load = sub-scenario planned annual charging rated load data * sub-scenario charging load seed curve

[0073] Charging load data for the planning year = ∑ 不同场景 (Sub-scenario planning of annual charging load)

[0074] (4) Charging load data under vehicle-grid interaction strategy = ∑ 不同场景 (f 子场景 (Sub-scenario planning annual charging load), sub-scenario strategy parameters)

[0075] Specifically, after building the electric vehicle charging load prediction model, the data of the target low-voltage distribution network is sorted and input, and the charging load data of the target low-voltage distribution network in the target planning year is gradually obtained, and superimposed on the basic load data of the planning year to obtain the final load data. Among them, the basic load data of the low-voltage distribution network in the planning year belongs to the field of traditional distribution network planning, and the technology is relatively mature, and the staff can calculate and input it according to the corresponding standards.

[0076] Among them, the charging load data of the target low-voltage distribution network in the target planning year and the vehicle-grid interaction strategy parameters of the area where the target low-voltage distribution network is located in the target planning year are used as input, and the charging load data of the target low-voltage distribution network under the vehicle-grid interaction strategy can be obtained.

[0077] S5: Analyze and obtain the access carrying capacity of the low-voltage distribution network in the planned year.

[0078] Total load = Charging load data under vehicle-grid interaction strategy + grid load

[0079] Carrying capacity = max (total load) / feeder maximum transmission capacity

[0080] Specifically, the charging load data after the target year strategy of the target low-voltage distribution network is superimposed on its basic load data, and the numerical result is output; finally, the load curve data of the target low-voltage distribution network after the vehicle-grid interaction strategy in the target year can be obtained. On this basis, the distribution transformer capacity input by the user is compared and analyzed to obtain the maximum load rate of the distribution transformer. Analyze whether the maximum load rate of the distribution transformer exceeds the load rate alarm line. If it exceeds the alarm line, output the alarm information. For example, compare the load rate alarm line set by the user, which is generally around 80%. If it exceeds the load rate alarm line, output the alarm information.

[0081] Example

[0082] The low-voltage distribution network access carrying capacity analysis method provided in this embodiment includes the following steps:

[0083] Step 1: Get historical data.

[0084] Historical data is generally exported from the systems and databases of relevant departments or institutions by authorized managers. For some low-voltage distribution network areas without micro data, it can be obtained after a certain degree of scaling and inference from macro data and local business forms. The amount of data is one of the key factors affecting the accuracy of the model in the subsequent step three. Therefore, if the accuracy of the method needs to be improved, a richer data set is required.

[0085] Step 2: Preprocess historical data.

[0086] The preprocessing of historical data can be achieved with mature tools, such as spreadsheet software, csv document editing software or json editing software. Data preprocessing can also be performed through programming. The split of the model training set and the model test set of the data can be performed in an 8:2 ratio or other scientific ratios.

[0087] Step 3: Build the model.

[0088] The model can be constructed by secondary development of standard machine learning software libraries. Python language can be selected and PyTorch open source library can be introduced. Recurrent neural networks (RNNs) can be selected as machine learning algorithms to build electric vehicle quantity prediction models based on machine learning, charging facility quantity prediction models based on machine learning, and electric vehicle charging load prediction models based on machine learning; regression or causal inference machine learning algorithms can be selected to build a quantitative analysis model of the impact of typical vehicle-grid interaction strategies on charging load based on machine learning, such as the autoregressive integrated moving average model (ARIMA).

[0089] Step 4: Call the model to calculate the number of electric vehicles in the target low-voltage distribution network in the planning year, the charging facility data of the target low-voltage distribution network in the planning year, the charging load data of the target low-voltage distribution network in the planning year, and the charging load data of the target low-voltage distribution network under the vehicle-grid interaction strategy.

[0090] The model output is based on a typical day, that is, the full-day data of the day when the charging load of the target low-voltage distribution network is high or the highest. The time interval can be selected as 15 minutes, 1 hour, or more detailed or rough. When 15 minutes is selected, the charging load data of the target low-voltage distribution network for the target planning year can be obtained at 96 points of real numbers throughout the day.

[0091] A mature solution is used to calculate the basic load of the target low-voltage distribution network in the planning year. Generally, distribution network planning professionals directly enter the results based on their own experience or other work basis.

[0092] Based on a typical day, the time interval can be 15 minutes, 1 hour, or more detailed or rough. Taking 15 minutes as an example, 96 real data points of charging load of the target low-voltage distribution network under the vehicle-grid interaction strategy in the target planning year and 96 real data points of total load of the target low-voltage distribution network after the vehicle-grid interaction strategy in the target year can be obtained. The sources of the vehicle-grid interaction strategy include foreseeable commercial strategies and policy strategies, and can be summarized as time-of-use electricity prices, two-way interaction, short-term response, long-term agreement direct control, and photovoltaic storage and charging interaction.

[0093] Step 5: Analyze and obtain the access carrying capacity of the low-voltage distribution network in the planned year.

[0094] The method provided in this embodiment can be carried by software or hardware systems and interacted through a human-computer interaction interface. The carrying form includes but is not limited to websites, computer software, APPs, applets, etc., and the data input and output and delivery methods include but are not limited to spreadsheet software, text documents, direct database connections, interface communications, etc.

[0095] The beneficial effects of the low-voltage distribution network access carrying capacity analysis method provided in this embodiment include:

[0096] Multiple machine learning models were constructed and used to predict the number of electric vehicles, the number of charging facilities, and the charging load. At the same time, a machine learning model with typical vehicle-grid interaction strategy characteristics was constructed to realize the charging load prediction under policy guidance. The low-voltage distribution network capacity was compared to obtain the analysis results of the low-voltage distribution network access carrying capacity in the planned year. The method can support the improvement of the professional planning of low-voltage distribution networks, expand the scope of its considerations, adapt to the new forms brought by electric vehicles, improve the scientificity and credibility of planning schemes, support the early discovery of potential problems in low-voltage distribution networks, and improve safety, reliability and overall quality.

[0097] The method of this embodiment can be implemented by combining a variety of open source software and data sources using software, digital or automated methods, and falls within the scope of protection.

[0098] The above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed by the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. A method for analyzing the access carrying capacity of a low-voltage distribution network, characterized in that: The method comprises: S1: Get historical data within a preset time period; S2: preprocessing the historical data to form a data set, and dividing the data set into a model training set and a model test set; S3: Based on the model training set and the model test set, a model based on machine learning is constructed, wherein the model includes an electric vehicle quantity prediction model, a charging facility quantity prediction model, an electric vehicle charging load prediction model, and a quantitative analysis model of the impact of typical vehicle-grid interaction strategies on charging load; S4: Call the model to calculate the number of electric vehicles in the target low-voltage distribution network in the planning year, the charging facility data of the target low-voltage distribution network in the planning year, the charging load data of the target low-voltage distribution network in the planning year, and the charging load data of the target low-voltage distribution network under the vehicle-grid interaction strategy; S5: Analyze and obtain the access carrying capacity of the low-voltage distribution network in the planned year.

2. The low voltage distribution network access carrying capacity analysis method according to claim 1, characterized in that: In S1, the historical data includes electric vehicle data and power grid data. The electric vehicle data includes the number of electric vehicles in use, electric vehicle market penetration rate, charging facility data, and vehicle-to-pile ratio; the power grid data includes charging load data, power grid load data, average distribution and transformation capacity per household in residential areas, electricity price data, vehicle-grid interaction strategy historical record data set, and low-voltage distribution network planning data.

3. The low voltage distribution network access carrying capacity analysis method according to claim 1, characterized in that: In S1, the scenarios of historical data obtained include urban residential areas, rural areas, road service areas, public service areas, and solar storage and charging stations; The urban residential area is divided into four sub-scenes: old residential areas, 15-25 year-old residential areas, 5-15 year-old residential areas and newly built residential areas. The public service area is divided into five sub-scenes: office area, commercial center, industrial center, rest center and public parking area.

4. The low voltage distribution network access carrying capacity analysis method according to claim 1, characterized in that: In S3, building a machine learning-based electric vehicle population prediction model includes: A regression-based machine learning algorithm for time series prediction is used to perform classification predictions according to different vehicle types, and the prediction results for the number of electric vehicles are superimposed.

5. The low voltage distribution network access carrying capacity analysis method according to claim 1, characterized in that: In S3, building a charging facility quantity prediction model based on machine learning includes: According to the two first dimensions of public and private, the four second dimensions of private charging piles, slow charging piles, fast charging piles and super fast charging piles are predicted separately, and the prediction results of the number of charging facilities are obtained by superimposing them.

6. The low voltage distribution network access carrying capacity analysis method according to claim 1, characterized in that: In S3, building a charging facility quantity prediction model based on machine learning includes: A regression-based machine learning algorithm for time series prediction is used to perform supervised machine learning to predict the car-to-charging pile ratio in the planned year. The result data of the electric vehicle quantity prediction model is then used as input to obtain the number of charging piles in the planned year according to the preset formula; the preset formula is: Planned annual number of charging piles = ∑ (number of electric vehicles in the sub-scenario * car-to-pile ratio in the sub-scenario).

7. The low voltage distribution network access carrying capacity analysis method according to claim 1, characterized in that: In S3, a quantitative analysis model based on machine learning on the impact of typical vehicle-grid interaction strategies on charging load is constructed, including: Based on the historical record data set of vehicle-grid interaction strategies obtained in S1, data is structured according to the application of strategy x to the scenario, where strategy x includes: time-of-use electricity price, two-way interaction, short-term response, long-term agreement direct control, and photovoltaic storage and charging interaction; A multivariate regression machine learning algorithm is used to train a quantitative analysis model of the impact of typical vehicle-grid interaction strategies on charging load based on the model training set.

8. The low-voltage distribution network access carrying capacity analysis method according to claim 1, characterized in that S4 include: Call the electric vehicle quantity prediction model based on machine learning to calculate the number of electric vehicles in the target low-voltage distribution network in the planning year; Call the charging facility quantity prediction model based on machine learning to calculate the charging facility data of the target low-voltage distribution network in the planning year; Calculate the base load of the target low-voltage distribution network in the planning year; Call the electric vehicle charging load prediction model based on machine learning to calculate the charging load data of the target low-voltage distribution network in the planning year; A quantitative analysis model of the impact of typical vehicle-grid interaction strategies on charging load based on machine learning is called, and the output of the electric vehicle charging load prediction model is used as part of the input to calculate the charging load data of the target low-voltage distribution network under the vehicle-grid interaction strategy.

9. The low-voltage distribution network access carrying capacity analysis method according to claim 8, characterized in that: S4 includes: After constructing the electric vehicle charging load prediction model, the data of the target low-voltage distribution network is sorted and input to obtain the charging load data of the target low-voltage distribution network in the target planning year, and then superimposed on the basic load data of the planning year to obtain the final load data.

10. The low-voltage distribution network access carrying capacity analysis method according to claim 1, characterized in that S5 include: Superimpose the charging load data after the target annual strategy of the target low-voltage distribution network onto its basic load data and output the numerical result; Obtain the load curve data of the target low-voltage distribution network after the vehicle-grid interaction strategy in the target year; Compare and analyze the distribution transformer capacity input by the user to obtain the maximum load rate of the distribution transformer; Analyze whether the maximum load rate of the distribution transformer exceeds the load rate alarm line. If it exceeds the alarm line, output alarm information.

Images