Vehicle network interactive scheduling method and system based on dynamic partition

Abstract

The invention discloses a vehicle network interactive scheduling method and system based on dynamic partitioning, and relates to the technical field of vehicle network interaction.The method comprises the steps that electric vehicle demand data is collected, a vehicle network interactive platform is constructed, the weight of the electric vehicle demand data and historical data is dynamically adjusted based on a long-short-term memory network, and electric vehicle prediction data is obtained; based on the electric vehicle demand data and the electric vehicle prediction data, dividing the whole service area into a plurality of intervals by adopting a dynamic partitioning algorithm, setting boundary adjustment rules of the intervals, and when a specific condition is monitored, dynamically adjusting boundaries of the intervals; and dynamically adjusting a scheduling strategy based on the electric vehicle demand data and the electric vehicle prediction data in the interval, and formulating a priority scheduling strategy for the user by combining real-time traffic flow data according to the electric vehicle charging demand and the emergency degree of the user. According to the invention, the complexity in traffic operation and the diversity of electric vehicle user demands can be effectively dealt with, and the efficiency and reliability of vehicle network interaction are improved.

Description

Technical Field

[0001] The present invention relates to the field of vehicle-grid interaction technology, and more specifically to a vehicle-grid interaction scheduling method and system based on dynamic partitioning. Background Art

[0002] With the rapid popularization of electric vehicles around the world, vehicle-grid interaction, as one of the key directions for the development of smart grids, is gaining increasing attention. Vehicle-grid interaction aims to achieve efficient collaboration between electric vehicles and the grid, optimize energy distribution, and improve the stability and reliability of the grid. However, the traditional vehicle-grid interactive scheduling method has gradually exposed many problems in the face of the increasingly complex power grid operation environment and the diversified needs of electric vehicle users. For example, the traditional vehicle-grid interactive scheduling method usually adopts a fixed partitioning method, which means that in the process of power grid planning and scheduling, the division of each area is predetermined and remains unchanged for a long time. This fixed partitioning method is difficult to adapt to the dynamic changes in the power grid operation environment. For example, with the development of the city and the change of traffic flow, the number of electric vehicles and electricity demand in different areas may change significantly. However, due to the fixed partitioning, the scheduling system cannot adjust the resource allocation in time, resulting in insufficient or excessive power supply in some areas, affecting the stability of the power grid and the charging experience of users. In a highly dynamic and complex traffic environment, emergencies (traffic accidents, bad weather, etc.) occur from time to time. These events may have an impact on the operation of the power grid and the charging needs of electric vehicles. For example, traffic accidents may cause road congestion, affecting the driving route and charging time of electric vehicles; bad weather may cause some charging facilities to be damaged or unable to work properly, affecting the charging plan of electric vehicles; and the traditional vehicle-grid interactive scheduling method is often difficult to respond to these emergencies quickly, and cannot adjust the partition boundaries and resource allocation in time, resulting in a decrease in the efficiency of power grid operation.

[0003] Therefore, how to provide a vehicle-grid interaction scheduling method and system based on dynamic partitioning, effectively adjust the partition boundaries to adapt to emergencies, and improve the efficiency and reliability of vehicle-grid interaction is a problem that technical personnel in this field urgently need to solve. Summary of the invention

[0004] In view of this, the present invention provides a vehicle-grid interaction scheduling method and system based on dynamic partitioning, which effectively adjusts the partition boundaries to adapt to emergencies and improves the efficiency and reliability of vehicle-grid interaction.

[0005] In order to achieve the above-mentioned purpose, the present invention adopts the following technical solution: a vehicle-grid interactive scheduling method based on dynamic partitioning, comprising:

[0006] Collecting electric vehicle demand data, the electric vehicle demand data including: traffic flow, electric vehicle charging demand and power grid load data;

[0007] Building a vehicle-grid interactive platform, the vehicle-grid interactive platform integrating a data analysis algorithm, a dynamic partitioning algorithm, and an interactive scheduling algorithm;

[0008] The data analysis algorithm dynamically adjusts the weights of electric vehicle demand data and historical data based on a long short-term memory network to obtain electric vehicle forecast data;

[0009] Based on the electric vehicle demand data and the electric vehicle prediction data, a dynamic partitioning algorithm is used to divide the entire service area into multiple intervals, and a boundary adjustment rule of the interval is set. When a specific condition is monitored, the boundary of the interval is dynamically adjusted;

[0010] The interactive scheduling algorithm dynamically adjusts the scheduling strategy based on the electric vehicle demand data and electric vehicle forecast data within the interval, and formulates a priority scheduling strategy for users based on the electric vehicle charging demand and the user's urgency, combined with real-time traffic flow data.

[0011] Preferably, the weights of electric vehicle demand data and historical data are dynamically adjusted based on the long short-term memory network, and the model parameters of the long short-term memory network are adjusted and optimized, and the historical data weight α t The specific adjustment calculation rules are as follows:

[0012] α t =α0+(α f -α0)·(1-e -kt );

[0013] Among them, α t represents the weight of historical data at time step t; t represents the time step; α0 represents the weight of initial historical data; α f It represents the weight of historical data at the final time; k represents the rate parameter of the adjustment function change; e represents the base of the natural logarithm.

[0014] Preferably, the long short-term memory network comprises:

[0015] S100, determine the number of LSTM layers and the number of hidden units;

[0016] S200, inputting electric vehicle demand data into the LSTM network and determining the time step required for prediction;

[0017] S300, use electric vehicle demand data and historical data to train the LSTM network, dynamically adjust the weights of electric vehicle demand data and historical data, optimize model parameters through back propagation algorithm, and introduce maximum time step limit historical data to optimize model learning;

[0018] S400, using the trained LSTM network, inputting the current electric vehicle demand data, and predicting the electric vehicle prediction data for the next time step.

[0019] Preferably, based on the electric vehicle demand data and the electric vehicle prediction data, a dynamic partitioning algorithm is used to divide the entire service area into multiple intervals, including:

[0020] Using the Apriori algorithm to perform correlation analysis on the electric vehicle demand data and the electric vehicle forecast data to identify potential risks;

[0021] Intelligent classification of potential risks based on clustering algorithms;

[0022] The service area is divided into a plurality of sections according to the classification result.

[0023] Preferably, the clustering algorithm comprises:

[0024] Generate a charging station attraction graph network using charging station location information and electric vehicle charging demand data;

[0025] A hierarchical graph clustering algorithm is used to extract and divide areas with potential risk constraints and electric vehicle charging density constraints by selecting similar and active nodes and combining edge weights and geographic distance constraints;

[0026] Compare the intervals with potential risk constraints and electric vehicle charging density constraints with the charging station attraction graph network, filter out the intervals with high potential risks in the clustering results, and accurately recommend all intervals where electric vehicles can be arranged for charging;

[0027] And set the priority order according to the distance between electric vehicles and charging stations.

[0028] Preferably, the charging station attraction graph network uses the locations of charging stations as nodes in the graph, and uses the charging density of electric vehicles in each charging station to obtain the node access intensity as the node feature; in order to represent the correlation between nodes, a hierarchical weighted edge graph is designed and constructed, and the neighbor nodes are divided into direct neighbors and indirect neighbors according to geographical isolation. The adjacency matrix of the edge graph determines the edge weights of the two types of neighbors through the similarity of potential risks.

[0029] Preferably, the node access intensity is estimated by node spatial attraction and node temporal attraction; wherein, node spatial attraction is expressed as the spatial attraction of node pi to each electric vehicle; node temporal attraction represents the number of electric vehicles charged at a certain node pi recorded through the vehicle-network interaction platform within a certain time period t.

[0030] Preferably, a vehicle-grid interactive scheduling system based on dynamic partitioning includes:

[0031] A data collection module and a vehicle-network interaction platform, data transmission is performed between the data collection module and the vehicle-network interaction platform;

[0032] The data acquisition module is used to collect electric vehicle demand data, and the electric vehicle demand data includes: traffic flow, electric vehicle charging demand and power grid load data;

[0033] The vehicle-grid interaction platform comprises: a data analysis unit, a dynamic partitioning unit and an interactive scheduling unit connected in sequence;

[0034] A data analysis unit, used to dynamically adjust the weights of electric vehicle demand data and historical data based on a long short-term memory network to obtain electric vehicle forecast data;

[0035] A dynamic partitioning unit, for dividing the entire service area into a plurality of intervals using a dynamic partitioning algorithm based on the electric vehicle demand data and the electric vehicle prediction data, setting interval boundary adjustment rules, and dynamically adjusting interval boundaries when specific conditions are monitored;

[0036] The interactive scheduling unit is used for the interactive scheduling algorithm to dynamically adjust the scheduling strategy based on the electric vehicle demand data and electric vehicle prediction data within the interval, and formulate a priority scheduling strategy for the user in combination with real-time traffic flow data according to the electric vehicle charging demand and the user's urgency.

[0037] Through the above technical solutions, it can be known that compared with the prior art, the present invention discloses a vehicle-grid interactive scheduling method and system based on dynamic partitioning, including: collecting electric vehicle demand data, the electric vehicle demand data including: traffic flow, electric vehicle charging demand and power grid load data; constructing a vehicle-grid interactive platform, the vehicle-grid interactive platform integrating data analysis algorithm, dynamic partitioning algorithm and interactive scheduling algorithm; the data analysis algorithm dynamically adjusts the weight of electric vehicle demand data and historical data based on long short-term memory network to obtain electric vehicle prediction data; based on the electric vehicle demand data and the electric vehicle prediction data, the entire service area is divided into multiple intervals using a dynamic partitioning algorithm, and the boundary adjustment rules of the interval are set. When specific conditions are monitored, the boundary of the interval is dynamically adjusted; the interactive scheduling algorithm is based on the electric vehicle demand data and the electric vehicle prediction data in the interval. The scheduling strategy is dynamically adjusted, and the priority scheduling strategy is formulated for the user according to the electric vehicle charging demand and the user's urgency, combined with real-time traffic flow data. The vehicle-grid interactive scheduling method based on dynamic partitioning of the present invention can effectively cope with the complexity of traffic operation and the diversity of electric vehicle user needs. The implementation of dynamic partitioning makes the scheduling of power grid resources more flexible and efficient, while improving the stability of the power grid and the charging experience of users. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0039] Figure 1 A schematic flow chart of a vehicle-grid interactive scheduling method based on dynamic partitioning provided by the present invention.

[0040] Figure 2 A schematic diagram of the structure of a vehicle-grid interactive scheduling system based on dynamic partitioning provided by the present invention.

[0041] Figure 3 This is a schematic diagram of the structure of the vehicle-network interactive platform provided by the present invention. DETAILED DESCRIPTION

[0042] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. 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] The embodiment of the present invention discloses a vehicle-grid interactive scheduling method based on dynamic partitioning, such as Figure 1 As shown, including:

[0044] Collect electric vehicle demand data, including traffic flow, electric vehicle charging demand and grid load data; use sensors and smart devices to collect real-time data such as traffic flow, electric vehicle charging demand, grid load, etc.

[0045] Traffic flow data collection: Through the GPS devices installed on the vehicles, the vehicle’s location information and driving trajectory are obtained, the vehicle’s speed, driving route, etc. are analyzed, and the traffic flow and congestion conditions are inferred through information interaction between vehicles.

[0046] Electric vehicle charging demand data collection: The vehicle's onboard diagnostic system (OBD) is used to obtain information such as the vehicle's battery status, mileage, and energy consumption. By connecting to a cloud server, this data is used to analyze charging demand, predict when the vehicle needs to be charged, and provide charging reminders and suggestions to users.

[0047] Grid load data collection: Smart meters are installed on the user side and grid nodes to measure power usage in real time, including active power, reactive power, voltage, current and other parameters. Smart meters upload data to the vehicle-grid interactive platform through the communication network to achieve real-time monitoring and analysis of grid load. By analyzing the data of smart meters in different regions, the distribution and change trend of grid load can be determined, providing a basis for optimizing grid operation and dispatching.

[0048] Constructing a vehicle-grid interactive platform that integrates data analysis algorithms, dynamic partitioning algorithms, and interactive scheduling algorithms; monitoring the status of charging stations in real time to ensure the normal operation of charging facilities;

[0049] The data analysis algorithm dynamically adjusts the weights of electric vehicle demand data and historical data based on the long short-term memory network to obtain electric vehicle forecast data; the electric vehicle forecast data includes electric vehicle demand data and grid load change data in the future period;

[0050] Based on the electric vehicle demand data and the electric vehicle forecast data, a dynamic partitioning algorithm is used to divide the entire service area into multiple intervals, and a boundary adjustment rule of the interval is set. When specific conditions (such as traffic congestion, weather changes) are monitored, the boundary of the interval is dynamically adjusted;

[0051] The interactive scheduling algorithm dynamically adjusts the scheduling strategy based on the electric vehicle demand data and electric vehicle forecast data within the interval, and formulates a priority scheduling strategy for users based on the electric vehicle charging demand and the user's urgency, combined with real-time traffic flow data, to ensure that the user's charging needs are met first.

[0052] The vehicle-grid interactive scheduling method based on dynamic partitioning can effectively cope with the complexity of traffic operation and the diversity of electric vehicle user needs. The implementation of dynamic partitioning makes the scheduling of power grid resources more flexible and efficient, while improving the stability of the power grid and the charging experience of users.

[0053] It also includes: establishing an emergency detection mechanism to monitor traffic accidents, bad weather and other emergencies in real time and respond quickly;

[0054] When an emergency occurs, grid resources can be quickly reallocated and the load of charging stations can be adjusted to ensure that the charging needs of electric vehicles are met.

[0055] Specifically, based on the long short-term memory network, the weights of electric vehicle demand data and historical data are dynamically adjusted, and the model parameters of the long short-term memory network are adjusted and optimized. The historical data weight α t The specific adjustment calculation rules are as follows:

[0056] α t=α0+(α f -α0)·(1-e -kt );

[0057] Among them, α t represents the weight of historical data at time step t; t represents the time step; α0 represents the weight of initial historical data; α f It represents the weight of historical data at the final time; k represents the rate parameter of the adjustment function change; e represents the base of the natural logarithm.

[0058] In the process of analyzing the electric vehicle demand data, a specific data analysis algorithm was used. The algorithm uses the long short-term memory network to dynamically adjust the weights of the electric vehicle demand data and historical data, and then optimizes the model parameters of the long short-term memory network. By dynamically adjusting the weights, the model can adapt to these changes in a timely manner when facing the changing operation of the power grid. On the one hand, the model will not rely too much on historical data and can learn new data patterns more quickly, thereby improving the training effect and performance of the model. On the other hand, this adjustment method enables the model to better adapt to the changing environment and data, enhancing the robustness and reliability of the model. By accurately identifying the key characteristics of the power grid operation, more accurate electric vehicle prediction data is finally obtained, providing strong support for the optimized operation and dispatch of the power grid.

[0059] Specifically, the long short-term memory network includes:

[0060] S100, determine the number of LSTM layers and the number of hidden units;

[0061] S200, inputting electric vehicle demand data into the LSTM network and determining the time step required for prediction;

[0062] S300, use electric vehicle demand data and historical data to train the LSTM network, dynamically adjust the weights of electric vehicle demand data and historical data, optimize model parameters through back propagation algorithm, and introduce maximum time step limit historical data to optimize model learning;

[0063] By introducing a maximum time step limit to truncate historical data, it is used to reduce the over-reliance of the long short-term memory network on long-term historical data. If the original historical data is a sequence of length N x = [x1, x2, ..., x N ], the maximum number of time steps to be retained is T, then:

[0064]

[0065] Among them, x' represents the truncated sequence; x N-T+1:N Indicates that the sequence x is intercepted and the historical data between N-T+1 and N is retained; x1:N It means capturing historical data from the first historical data of sequence x to the Nth historical data.

[0066] By introducing a maximum time step limit to truncate historical data, it is used to reduce the excessive dependence of the LSTM network on long-term historical data, so that the LSTM network can focus more on processing recent data, learning and adapting to recent information, and to a certain extent control the dependence of the LSTM network on historical data.

[0067] S400, using the trained LSTM network, inputting the current electric vehicle demand data, and predicting the electric vehicle prediction data for the next time step.

[0068] In the LSTM network, at time step t, the electric vehicle demand data of LSTM is x t , the hidden state of the previous time step t-1 is h t-1 , then the specific function model of the LSTM network is:

[0069] Forget Gate:

[0070] f t =σ(W f ·[h t-1 ,x t ]+b f );

[0071] Where σ represents the sigmoid function; W f represents the forget gate weight matrix; b f represents the bias of the forget gate; f t represents the output of the forget gate;

[0072] Input Gate:

[0073] i t =σ(W i ·[h t-1 ,x t ]+b i );

[0074] Calculate candidate cell states;

[0075]

[0076] Among them, i t represents the output of the input gate; represents the candidate cell state; W i represents the weight matrix of the input gate; b i represents the bias term of the input gate; W C represents the weight matrix used to generate candidate cell states; b Crepresents the bias term used to generate candidate cell states; tanh represents the hyperbolic tangent function;

[0077] Update cell status:

[0078]

[0079] Among them, C t Indicates cell status; C t-1 represents the cell state at the previous time step;

[0080] Output Gate:

[0081] o t =σ(W o ·[h t-1 ,x t ]+b o );

[0082] h t =o t *tanh(C t );

[0083] Among them, h t Indicates hidden state; o t Represents the output of the output gate; W o represents the weight matrix of the output gate; represents the bias of the output gate; tanh(C t ) represents the cell state C t The hyperbolic tangent function is applied to each element in .

[0084] Specifically, based on the electric vehicle demand data and the electric vehicle prediction data, a dynamic partitioning algorithm is used to divide the entire service area into multiple intervals, including:

[0085] Using the Apriori algorithm to perform correlation analysis on the electric vehicle demand data and the electric vehicle forecast data to identify potential risks;

[0086] The Apriori algorithm includes item set generation and pruning steps:

[0087] Itemset generation includes:

[0088] Set a minimum support threshold. Support refers to the frequency of an item set appearing in all transactions. If the minimum support is set to 0.2, then only item sets with a frequency greater than or equal to 0.2 in the data set will be considered frequent item sets.

[0089] Generate candidate 1-item sets: First, scan the data for the first time to find all single items (i.e., 1-item sets), count the number of times they appear in the data set, and calculate the support of each 1-item set. Filter out 1-item sets whose support is greater than or equal to the minimum support threshold as frequent 1-item sets.

[0090] Generate higher-order candidate item sets: Through the connection operation, the frequent k-1 item set is connected to itself to generate a candidate k item set.

[0091] Pruning step: Prune the generated candidate k-item sets and remove the itemsets that do not meet the conditions. According to the properties of the Apriori algorithm, all subsets of a frequent itemset must also be frequent. Therefore, if a k-1 item subset of a candidate k-item set is not frequent, then this candidate k-item set cannot be frequent and will be deleted from the candidate set.

[0092] Repeat the above steps: Repeat the process of generating candidate item sets and pruning until no more frequent item sets can be generated. The final frequent item set is the potential association pattern in the data.

[0093] The Apriori algorithm is used to conduct in-depth analysis of electric vehicle demand data and electric vehicle forecast data, and to find potential associations between different variables in the data through frequent item set mining. For example, there is a correlation between high grid demand in a specific time period and specific weather conditions or specific traffic flow. Identifying these associations helps to discover potential risk factors. If the grid demand increases sharply under certain weather conditions, and the current grid supply capacity may not be able to meet it, this constitutes a potential risk.

[0094] Intelligent classification of potential risks based on clustering algorithms;

[0095] The service area is divided into multiple intervals according to the classification results. Through this data-based dynamic partitioning method, the power grid service area can be managed more scientifically and accurately, the stability, reliability and efficiency of the power grid can be improved, and various potential risks can be better dealt with.

[0096] Specifically, the clustering algorithm includes:

[0097] Using the charging station location information and electric vehicle charging demand data, a charging station attraction graph network is generated; the nodes in the network represent charging stations, and the edges between the nodes represent the attraction relationship between charging stations;

[0098] A hierarchical graph clustering algorithm is used to extract and divide areas with potential risk constraints and electric vehicle charging density constraints by selecting similar and active nodes and combining edge weights and geographic distance constraints;

[0099] Compare the intervals with potential risk constraints and electric vehicle charging density constraints with the charging station attraction graph network, filter out the intervals with high potential risks in the clustering results, and accurately recommend all intervals where electric vehicles can be arranged for charging;

[0100] And set the priority order according to the distance between the electric vehicle and the charging station. Improve the efficiency and safety of electric vehicle charging.

[0101] Specifically, the charging station attraction graph network uses the locations of charging stations as nodes in the graph, and uses the charging density of electric vehicles in each charging station to obtain the node access intensity as the node feature; in order to represent the correlation between nodes, a hierarchical weighted edge graph is designed and constructed, and the neighbor nodes are divided into direct neighbors and indirect neighbors according to geographical isolation. The adjacency matrix of the edge graph determines the edge weights of the two types of neighbors through the similarity of potential risks.

[0102] Among them, when the two charging station locations represented by nodes pi and pj have a common electric vehicle charging, node pj is the direct neighbor of node pi; when the two charging station locations represented by nodes pi and pl do not have a shared electric vehicle charging, but nodes pi and pl have the same direct neighbor pj, node pl is the indirect neighbor of node pi.

[0103] Firstly, the charging station location information and electric vehicle charging demand data are used to generate a charging station attraction graph network, which is used to determine the location of each charging station and the charging demand of electric vehicles in the service area.

[0104] The embodiment of the present invention is based on a Bayesian algorithm of gravity attraction. It estimates the visit popularity of charging stations by incorporating the activity of electric vehicle traffic in space and time and the distance of charging station locations, and constructs a hierarchical edge graph for charging station location nodes to represent the geographical distance and functional correlation between charging station locations. This can optimize charging stations with high potential risks for electric vehicle charging and high electric vehicle charging density.

[0105] Specifically, the node access intensity is estimated by the node spatial attraction and node temporal attraction; the node spatial attraction is expressed as the spatial attraction of node pi to each electric vehicle; the node temporal attraction represents the number of electric vehicles charged at a node pi recorded through the vehicle-grid interaction platform within a certain time period t.

[0106] The embodiment of the present invention formulates a priority scheduling strategy for users based on the electric vehicle demand data and forecast data within the interval, combined with the user's urgency and real-time traffic flow data, to achieve efficient scheduling of electric vehicles, meet the needs of different users, and at the same time improve the operating efficiency and service quality of the entire vehicle-grid interactive system.

[0107] In a specific embodiment of the present invention, a vehicle-grid interactive scheduling system based on dynamic partitioning is provided. Figure 2 and Figure 3 As shown, including:

[0108] A data collection module and a vehicle-network interaction platform, data transmission is performed between the data collection module and the vehicle-network interaction platform;

[0109] The data acquisition module is used to collect electric vehicle demand data, and the electric vehicle demand data includes: traffic flow, electric vehicle charging demand and power grid load data;

[0110] The vehicle-grid interaction platform comprises: a data analysis unit, a dynamic partitioning unit and an interactive scheduling unit connected in sequence;

[0111] A data analysis unit, used to dynamically adjust the weights of electric vehicle demand data and historical data based on a long short-term memory network to obtain electric vehicle forecast data;

[0112] A dynamic partitioning unit, for dividing the entire service area into a plurality of intervals using a dynamic partitioning algorithm based on the electric vehicle demand data and the electric vehicle prediction data, setting interval boundary adjustment rules, and dynamically adjusting interval boundaries when specific conditions are monitored;

[0113] The interactive scheduling unit is used for the interactive scheduling algorithm to dynamically adjust the scheduling strategy based on the electric vehicle demand data and electric vehicle prediction data within the interval, and formulate a priority scheduling strategy for the user in combination with real-time traffic flow data according to the electric vehicle charging demand and the user's urgency.

[0114] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0115] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A vehicle-grid interactive scheduling method based on dynamic partitioning, characterized in that: include: Collecting electric vehicle demand data, the electric vehicle demand data including: traffic flow, electric vehicle charging demand and power grid load data; Building a vehicle-grid interactive platform, the vehicle-grid interactive platform integrating a data analysis algorithm, a dynamic partitioning algorithm, and an interactive scheduling algorithm; The data analysis algorithm dynamically adjusts the weights of electric vehicle demand data and historical data based on a long short-term memory network to obtain electric vehicle forecast data; Based on the electric vehicle demand data and the electric vehicle prediction data, a dynamic partitioning algorithm is used to divide the entire service area into multiple intervals, and a boundary adjustment rule of the interval is set. When a specific condition is monitored, the boundary of the interval is dynamically adjusted; The interactive scheduling algorithm dynamically adjusts the scheduling strategy based on the electric vehicle demand data and electric vehicle forecast data within the interval, and formulates a priority scheduling strategy for users based on the electric vehicle charging demand and the user's urgency, combined with real-time traffic flow data.

2. The vehicle-grid interactive scheduling method based on dynamic partitioning according to claim 1 is characterized in that: Based on the long short-term memory network, the weights of electric vehicle demand data and historical data are dynamically adjusted, and the model parameters of the long short-term memory network are adjusted and optimized. The historical data weight α t The specific adjustment calculation rules are as follows: α t ＝α0+(α f -α0)·(1-e -kt )； Among them, α t represents the weight of historical data at time step t; t represents the time step; α0 represents the weight of initial historical data; α f It represents the weight of historical data at the final time; k represents the rate parameter of the adjustment function change; e represents the base of the natural logarithm.

3. The vehicle-grid interactive scheduling method based on dynamic partitioning according to claim 2 is characterized in that: The long short-term memory network comprises: S100, determine the number of LSTM layers and the number of hidden units; S200, inputting electric vehicle demand data into the LSTM network and determining the time step required for prediction; S300, use electric vehicle demand data and historical data to train the LSTM network, dynamically adjust the weights of electric vehicle demand data and historical data, optimize model parameters through back propagation algorithm, and introduce maximum time step limit historical data to optimize model learning; S400, using the trained LSTM network, inputting the current electric vehicle demand data, and predicting the electric vehicle prediction data for the next time step.

4. The vehicle-grid interactive scheduling method based on dynamic partitioning according to claim 1 is characterized in that: Based on the electric vehicle demand data and the electric vehicle forecast data, a dynamic partitioning algorithm is used to divide the entire service area into multiple intervals, including: Using the Apriori algorithm to perform correlation analysis on the electric vehicle demand data and the electric vehicle forecast data to identify potential risks; Intelligent classification of potential risks based on clustering algorithms; The service area is divided into a plurality of sections according to the classification result.

5. The vehicle-grid interactive scheduling method based on dynamic partitioning according to claim 4 is characterized in that: The clustering algorithm includes: Generate a charging station attraction graph network using charging station location information and electric vehicle charging demand data; A hierarchical graph clustering algorithm is used to extract and divide areas with potential risk constraints and electric vehicle charging density constraints by selecting similar and active nodes and combining edge weights and geographic distance constraints; Compare the intervals with potential risk constraints and electric vehicle charging density constraints with the charging station attraction graph network, filter out the intervals with high potential risks in the clustering results, and accurately recommend all intervals where electric vehicles can be arranged for charging; And set the priority order according to the distance between electric vehicles and charging stations.

6. The vehicle-grid interactive scheduling method based on dynamic partitioning according to claim 5 is characterized in that: The charging station attraction graph network uses the charging station locations as nodes in the graph and uses the electric vehicle charging density in each charging station to obtain the node access intensity as the node feature; In order to represent the correlation between nodes, a hierarchical weighted edge graph is designed and constructed. Neighbor nodes are divided into direct neighbors and indirect neighbors according to geographical isolation. The adjacency matrix of the edge graph determines the edge weights of the two types of neighbors by the similarity of potential risks.

7. The vehicle-grid interactive scheduling method based on dynamic partitioning according to claim 6 is characterized in that: The node access intensity is estimated by the node spatial attraction and node temporal attraction; the node spatial attraction is expressed as the spatial attraction of node pi to each electric vehicle; the node temporal attraction represents the number of electric vehicles charged at a node pi recorded through the vehicle-network interaction platform within a certain time period t.

8. A vehicle-grid interactive scheduling system based on dynamic partitioning, applying a vehicle-grid interactive scheduling method based on dynamic partitioning according to any one of claims 1 to 7, characterized in that: include: A data collection module and a vehicle-network interaction platform, data transmission is performed between the data collection module and the vehicle-network interaction platform; The data acquisition module is used to collect electric vehicle demand data, and the electric vehicle demand data includes: traffic flow, electric vehicle charging demand and power grid load data; The vehicle-grid interaction platform comprises: a data analysis unit, a dynamic partitioning unit and an interactive scheduling unit connected in sequence; A data analysis unit, used to dynamically adjust the weights of electric vehicle demand data and historical data based on a long short-term memory network to obtain electric vehicle forecast data; A dynamic partitioning unit, for dividing the entire service area into a plurality of intervals using a dynamic partitioning algorithm based on the electric vehicle demand data and the electric vehicle prediction data, setting interval boundary adjustment rules, and dynamically adjusting interval boundaries when specific conditions are monitored; The interactive scheduling unit is used for the interactive scheduling algorithm to dynamically adjust the scheduling strategy based on the electric vehicle demand data and electric vehicle prediction data within the interval, and formulate a priority scheduling strategy for the user in combination with real-time traffic flow data according to the electric vehicle charging demand and the user's urgency.

Images