A new energy vehicle charging and discharging collaborative optimization method based on vehicle-network interaction

Abstract

The present application relates to the technical field of power distribution network management, and more particularly to a new energy vehicle charging and discharging collaborative optimization method based on vehicle-to-grid interaction, which, when the power distribution network receives a power dispatching signal and the total power of the power distribution network at the current time does not equal 0, acquires a trained graph structure model of the power distribution network, acquires a real-time power disturbance feature matrix corresponding to each power distribution node at the current time, and inputs the running disturbance feature matrix, the branch current disturbance feature matrix and the effective adjacent disturbance feature matrix at the current time into the trained graph structure model to output the corresponding model output result; the model output result is used to acquire the electrical position value of each power distribution node at the current time, and the charging and discharging of the power distribution nodes of the power distribution network is controlled, thus breaking through the scheduling limitation of'same node same strategy', and improving the power distribution network improvement efficiency of unit charging and discharging power on the premise of ensuring the vehicle travel power.

Description

Technical Field

[0001] This invention relates to the field of power distribution network management technology, and in particular to a method for coordinated optimization of charging and discharging of new energy vehicles based on vehicle-grid interaction. Background Technology

[0002] After new energy vehicles are connected to charging and discharging facilities, they can both charge from the grid as a load and supply power to the grid when bidirectional capability is available. With the increasing number of new energy vehicles, a large number of vehicles are being distributed across residential areas, commercial areas, industrial parks, and public stations, making power batteries a variable resource. To improve energy efficiency, the coordinated optimization of charging and discharging for new energy vehicles is becoming increasingly important for the operation of charging and discharging systems.

[0003] Existing methods for coordinated charging and discharging optimization of new energy vehicles based on vehicle-to-grid interaction treat different charging piles as energy storage units with the same electrical function, and schedule charging and discharging based on time-of-use pricing and total system load to reduce peak-valley differences and increase the absorption of new energy. However, the electrical distance, line impedance, branch margin, and surrounding loads between different charging piles and substations in the distribution network vary. For example, when discharging 20kW, vehicles at the end of the feeder may simultaneously reduce the current in multiple lines and increase the downstream voltage, while vehicles near the busbar may only slightly change the total load. When the voltage at the photovoltaic concentrator node is high, charging is more valuable than discharging for local vehicles. Therefore, existing methods that only schedule charging and discharging based on time-of-use pricing and remaining vehicle power are difficult to adapt to continuous changes in load, switch status, distributed power sources, and online vehicles, which may still lead to problems such as local congestion and voltage deviation, making it difficult to guarantee the operating efficiency of the distribution network. Summary of the Invention

[0004] In view of this, embodiments of the present invention provide a new energy vehicle charging and discharging collaborative optimization method based on vehicle-to-grid interaction, in order to solve the problem that existing new energy vehicle charging and discharging collaborative optimization methods use the same charging and discharging allocation strategy for new energy vehicles at different access locations, making it difficult to solve problems such as local congestion and voltage deviation.

[0005] This invention provides a method for coordinated optimization of charging and discharging of new energy vehicles based on vehicle-to-grid interaction, the method comprising the following steps: When the distribution network receives a power dispatch signal and the total power of the distribution network at the current moment is not equal to 0, the historical monitoring data of each distribution node, each charging pile under the distribution node, and each historical charging vehicle under the charging pile are obtained in a preset historical period. The graph structure model of the distribution network is trained to obtain a trained graph structure model. The input of the trained graph structure model is the running disturbance feature matrix, the power disturbance feature matrix, the branch current disturbance feature matrix, and the effective adjacent disturbance feature set. For any distribution node, the reliable charging and discharging power of each charging vehicle at the current time is analyzed to obtain the real-time power disturbance feature matrix corresponding to the distribution node; the operating disturbance feature matrix, branch current disturbance feature matrix and effective adjacent disturbance feature set at the current time are obtained, and combined with the real-time power disturbance feature matrix corresponding to the distribution node as the input of the trained graph structure model, and the model output result corresponding to the distribution node is output. Using the model output results, deviation analysis is performed on the active power loss of the distribution network at the current moment, as well as the voltage and branch current of any distribution node, to obtain the electrical location value of any distribution node at the current moment; the electrical location value of each distribution node at the current moment is obtained, and the charging and discharging regulation of the distribution nodes of the distribution network is performed.

[0006] The beneficial effects of the embodiments of the present invention compared with the prior art are as follows: In this invention, when the distribution network receives a power dispatch signal and the total power of the distribution network at the current moment is not equal to 0, a power change graph response model of the distribution network is first constructed based on the historical monitoring data of the distribution network. This model, also known as a trained graph structure model, is used to analyze the power propagation relationship between distribution nodes. Then, based on the trained graph structure model, the changes in the overall network state caused by charging and discharging are simulated for each distribution node. This is used to analyze the voltage and current changes of a single distribution node on other branches and other distribution nodes, as well as the overall network loss changes of the distribution network. Thus, the voltage, current, and network loss of each distribution node under the disturbance of a single distribution node are obtained. The predicted values ​​are compared with the real-time distribution network status to quantify the electrical location value of each distribution node under charging and discharging, which can accurately identify the regulation potential of sensitive areas such as feeder ends and photovoltaic centralized nodes. Finally, based on the electrical location value of each distribution node under charging and discharging, adaptive regulation of the distribution network is carried out. This allows for the priority selection of vehicles that contribute more to alleviating voltage deviation, reducing branch congestion, and reducing network losses during regulation. This avoids the blind reliance on electricity price and remaining power in traditional methods, and improves the distribution network improvement efficiency per unit charging and discharging power while ensuring the power supply for vehicle travel. This breaks through the scheduling limitations of "same node, same strategy". Detailed Implementation

[0007] The embodiments of this disclosure are described in detail below. The embodiments described below are exemplary and intended to explain this disclosure, and should not be construed as limiting this disclosure.

[0008] It should be noted that the terms "first," "second," etc., used in this disclosure are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this disclosure described herein can be implemented in orders other than those described herein. The implementations described in the following exemplary embodiments do not represent all implementations consistent with this disclosure. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this disclosure.

[0009] To illustrate the technical solution of the present invention, specific embodiments are described below.

[0010] Specifically, this invention provides a method for coordinated optimization of charging and discharging of new energy vehicles based on vehicle-to-grid interaction, which may include: Step S101: When the distribution network receives a power dispatch signal and the total power of the distribution network at the current moment is not equal to 0, acquire the historical monitoring data of each distribution node, each charging pile under the distribution node, and each historical charging vehicle under the charging pile in the preset historical time period, train the graph structure model of the distribution network, and obtain the trained graph structure model. The input of the trained graph structure model is the operating disturbance feature matrix, the power disturbance feature matrix, the branch current disturbance feature matrix, and the effective adjacent disturbance feature matrix.

[0011] Multiple charging stations exist within the same distribution area. The distribution nodes of the distribution network are distributed near the main trunk, in the middle of the branches, and at the end of the feeders. The power changes of different distribution nodes have different impacts on other distribution nodes. Existing methods use the same charging and discharging strategy for new energy vehicles under different distribution nodes. When situations such as low voltage at the end of the feeder, high voltage at the photovoltaic central node, or local branches approaching the current carrying limit occur, existing methods are difficult to effectively solve the local distribution network problems such as voltage deviation, branch congestion, and network loss caused by these situations. Therefore, in this embodiment of the invention, the impact of power changes of different distribution nodes on the distribution network is considered, and the bidirectional electrical position value of each distribution node in the current operating state is evaluated. This is used to schedule charging and discharging, thereby improving the effect of unit charging and discharging power on improving local distribution network problems.

[0012] Considering that power dispatching in distribution networks typically has a fixed dispatching cycle, when power dispatching is required, the task must be completed within the corresponding dispatching cycle. According to the technical specifications of power demand response systems, in real-time demand response operations, the delay in transmitting the demand response signal from the master station to the demand response terminal should not exceed 1 second, and the time for the demand response terminal to receive the execution instruction and complete the response should not exceed 30 seconds. Therefore, in this embodiment of the invention, the dispatching cycle is 30 seconds long, with the first moment of the dispatching cycle as the start time of dispatching analysis and the last moment as the time to be dispatched, i.e., the moment of power regulation. For example, if the dispatching cycle is 5 minutes, the start time of the current dispatching cycle is 10:00, and the corresponding time to be dispatched is 10:05. Therefore, based on the target change in net interactive power of vehicles issued by the superior dispatching platform or distribution network operation control platform, the power dispatching signal is received. When the distribution network receives the power dispatching signal, the corresponding moment belongs to the first moment of the dispatching cycle, i.e., the current moment, thereby obtaining the total regulated power of the distribution network at the current moment. It then determines whether power regulation is needed within the current scheduling cycle. The specific determination method is as follows: This indicates a need to increase the charging power of new energy vehicles; This indicates a need to increase the discharge power of new energy vehicles; This indicates that vehicle power control will not be implemented during the current scheduling cycle.

[0013] Furthermore in First, historical monitoring data for each distribution node of the power distribution network, each charging pile under the distribution node, and each historically charged vehicle under the charging pile are obtained within a preset historical time period. The specific method for obtaining this data is as follows: Based on the current distribution network topology, it is possible to determine the multiple distribution nodes included in the distribution network, the multiple charging piles corresponding to a single distribution node, the one or more charging interfaces included under a single charging pile, and that a single charging interface can be bound to at most one new energy vehicle at the same time. Then, the active power, reactive power, voltage, active power of distributed power sources (referring to power generation equipment connected to the distribution node, including distributed photovoltaic, small wind turbines, etc.) of each distribution node can be obtained using the wireless data transmission terminal or data transmission unit (DTU) of the distribution network terminal, as well as the current and active power loss of each branch in the distribution network.

[0014] For each charging pile under each power distribution node, the rated bidirectional power (rated charging power and rated discharging power) of the charging pile is collected by the controller inside the power distribution node. The rated bidirectional power only needs to be collected once when connected. The rated bidirectional power usually does not change when the setting parameters are not adjusted. Therefore, it only needs to be updated and collected again when the equipment parameters change. The power value of each charging interface is collected by the bidirectional energy meter inside the charging pile. The power value corresponding to the charging direction is positive and the power value corresponding to the discharging direction is negative. For each new energy vehicle bound to a charging port, the vehicle battery management system collects the authorized vehicle's battery capacity, SOC value (remaining battery charge status), maximum allowable SOC value (if the user has set a target SOC value, the user's target SOC value will be used as the maximum allowable SOC value; if the user has not set one but the vehicle BMS provides a safe maximum allowable SOC value, that value will be used as the maximum allowable SOC value; if neither is provided, the default maximum allowable SOC value is 100%), and minimum SOC value required for leaving the site (set by the user; for example, if the user requires the minimum SOC value for leaving the site to be no less than 60%, the minimum SOC value required for leaving the site will be 60%; if the user has not set one, the user's target SOC value will be used).

[0015] It should be noted that the distribution network topology connection relationship is updated once when the switch connection status between distribution nodes changes. Therefore, except for the distribution network topology connection relationship and the rated bidirectional power of the charging pile, all monitoring data are collected every 5 seconds.

[0016] Then, based on the above acquisition method, all monitoring data of the distribution network at each sampling time can be obtained. In this embodiment of the invention, the monitoring data of the most recent 90 days are collected as the historical monitoring data of the distribution network, which is used to construct the power change graph response model of the distribution network, that is, the trained graph structure model, to analyze the power propagation relationship between distribution nodes. However, in order to reduce the impact of the data range of different physical quantities on the training of the graph structure model, a robust scaling transformation based on interquartile range is performed on the historical monitoring data in each dimension to eliminate the influence of dimensions and magnitude.

[0017] in, This represents the transformed numerical value. This represents the initial value (the collected monitoring data). This represents the median of all historical monitoring data for the corresponding dimension. Used to convert interquartile ranges to a scale comparable to the standard deviation, setting , This represents the interquartile range of historical monitoring data in the corresponding dimension. This indicates the minimum resolution of the corresponding metering device, used to avoid situations where historical monitoring data in the corresponding dimension is too concentrated, resulting in a denominator of 0. This represents the function that takes the maximum value.

[0018] It should be noted that, for the power values ​​of charging piles, in order to preserve the positive or negative nature of their charging and discharging data, a scaling transformation is first performed based on the absolute value of the collected power values. Then, the result after scaling based on the absolute value is multiplied by 1 to obtain the final transformation result. The result of scaling based on the absolute value is multiplied by -1 to obtain the final transformation result.

[0019] Thus, we can obtain historical monitoring data for each distribution node of the power distribution network, each charging pile under the distribution node, and each historically charged vehicle under the charging pile within a preset historical period (the most recent 90 days), and the data used in subsequent calculations are all dimensionless data.

[0020] A power distribution network is a holistic system, meaning that distribution nodes do not operate independently but are interconnected. The power changes of distribution nodes at different locations have varying impacts on the overall distribution network. For example, distribution nodes at the ends of branches and feeders have a greater impact on the local voltage state of the distribution network, while the impact of distribution nodes near the main line is more likely to be distributed across multiple branches. Therefore, based on the distribution network topology, historical monitoring data from each distribution node, each charging pile under a distribution node, and each historically charging vehicle under a charging pile are used to train a graph structure model of the distribution network within a preset historical time period. This trained graph structure model is then used to analyze the impact of power changes at each distribution node on other distribution nodes. The process of training the graph structure model of the distribution network is as follows: Based on the distribution and connection relationships of the various distribution nodes in the distribution network, the distribution network is represented as a diagram. ,in, Represents the set of distribution nodes. This represents the set of edges formed by lines and switches. Since switches in a line may be open or closed, resulting in different actual topological connections at different monitoring times, but in order to facilitate learning the propagation relationship of charging and discharging power from historical data, the connection relationship is still retained even if a line switch is open at certain monitoring times.

[0021] Since the propagation relationship of power can only be effectively reflected when the power changes, the power change of each power distribution node at each historical moment within a preset historical period is obtained based on the power value of each charging interface of each charging pile under each power distribution node at each historical moment within a preset historical period. For any power distribution node, taking any historical moment within the preset historical period as an example, the power values ​​of each charging interface of each charging pile under the power distribution node at any historical moment are accumulated to obtain the total power value of the power distribution node at any historical moment. The difference between the total power value at any historical moment and the total power value at the previous moment is taken as the power change of the power distribution node at any historical moment.

[0022] In one embodiment, taking the u-th distribution node and the i-th historical time as an example, the power change of the u-th distribution node at the i-th historical time is:

[0023] in, This represents the power change of the u-th distribution node at the i-th historical moment. This indicates the number of charging interfaces contained in the u-th power distribution node. Let represent the power value of the v-th charging interface under the u-th distribution node at the i-th historical moment. This represents the power value of the v-th charging interface under the u-th distribution node at the (i-1)-th historical moment.

[0024] It should be noted that, This represents the cumulative power value of all charging interfaces under the u-th distribution node at the i-th historical moment. This represents the cumulative power value of all charging interfaces under the u-th distribution node at the (i-1)th historical moment, corresponding to This represents the difference between the cumulative power values ​​of two adjacent historical moments. For the first historical moment, since there is no previous moment as a reference, it is assumed that its power has not changed. .

[0025] Similarly, we can obtain the power change of the u-th distribution node at each historical moment within a preset historical period, take the absolute value of each power change, and denote the absolute value of the power change at the i-th historical moment as . The median of the absolute values ​​of all power changes is taken as the power change threshold. ,like If the i-th historical moment is considered the main power change moment of the u-th distribution node, then the power change amount of each distribution node at each historical moment within the preset historical period and the corresponding power change threshold of the distribution node are obtained. The main power change moment of the corresponding distribution node is then obtained using the power change threshold of each distribution node. If any historical moment within the preset historical period is the main power change moment of at least one distribution node, then that historical moment is recorded as a moment of significant power change. Thus, multiple moments of significant power change within the preset historical period can be obtained, and these moments are used as sample moments. The graph structure model of the distribution network is trained using sample data from these sample moments. During training, the sample data is arranged according to the order of the sample moments, with every 40 sample data points forming a sample dataset for sequential model training to update the model parameters. If there are fewer than 40 sample data points, all sample data points are used as a single sample dataset for training.

[0026] The process of constructing sample data for a given time point is as follows: For any moment y with a significant power change within a preset historical time period, the previous moment of that moment with a significant power change is taken as the disturbance moment y-1. Taking any distribution node as an example, for any historical charging vehicle at that time point, based on the SOC value, minimum SOC value required for departure, and upper limit of allowable SOC value of that historical charging vehicle at that time point, the dischargeable energy and dischargeable energy of that historical charging vehicle at that time point are obtained respectively.

[0027]

[0028] in, This represents the dischargeable energy of any historically charged vehicle at the time of the disturbance. This represents the rechargeable energy of any historical charging vehicle at the time of the disturbance. This indicates the battery capacity of any of the historically charged vehicles. This represents the function that takes the maximum value, and 0 represents a constant. This represents the SOC value of any historically charged vehicle at the time of the disturbance. This represents the minimum SOC value required for any historically charged vehicle to leave the site at the time of the disturbance. This indicates the maximum allowable SOC value for any historically charged vehicle. This indicates that the discharge energy is 0 when the SOC value reaches the minimum requirement. This indicates that the rechargeable energy is 0 when the SOC value reaches its upper limit.

[0029] By combining the dischargeable energy and rechargeable energy of any historical charging vehicle at the time of the disturbance, as well as the probability that any historical charging vehicle remains until the time of scheduling, the maximum dischargeable power and maximum rechargeable power of any historical charging vehicle are obtained respectively:

[0030]

[0031] in, This indicates the maximum discharge power of any historically charged vehicle. This indicates the maximum rechargeable power of any of the historically charged vehicles. This represents the function that takes the minimum value. This represents the rated discharge power limited by the vehicle and its charging station (the minimum of the maximum allowed discharge power of the vehicle and the rated power of the charging station in the discharge direction). This represents the rated charging power limited by the vehicle and its charging station (the minimum of the maximum allowed charging power of the vehicle and the rated power of the charging station in the charging direction). This represents the time interval between the disturbance time and the time to be scheduled, where the time to be scheduled refers to the last moment of the scheduling cycle. This represents the probability that any historical charging vehicle will remain until the scheduled time. Since the arrival and departure of historical charging vehicles are predetermined, the probability of remaining until the scheduled time is not needed. Therefore, by default, each historical charging vehicle has a probability of [missing information]. The value is a constant 1.

[0032] Therefore, it is possible to obtain the maximum rechargeable power and maximum dischargeable power of each historical charging vehicle at any distribution node at the time of the disturbance. The maximum rechargeable power of all historical charging vehicles is accumulated to obtain the accumulated maximum rechargeable power value, and the maximum dischargeable power of all historical charging vehicles is accumulated to obtain the accumulated maximum dischargeable power value. Simultaneously, the voltage U, active power P, reactive power Q, distributed power source active power PG, and total power W of any distribution node at the time of the disturbance are obtained, and these, along with the accumulated maximum rechargeable power value D and the accumulated maximum dischargeable power value C, form the operating characteristic vector of any distribution node at the time of the disturbance. The operating characteristic vector of the u-th distribution node at the time of the disturbance is denoted as... ,in, , This represents the power value of the v-th charging interface under the u-th distribution node at disturbance time y-1. Similarly, the operating feature vectors of each distribution node at the disturbance time are obtained to form the operating disturbance feature matrix at any time of significant power change. N represents the number of distribution nodes.

[0033] The power change at each distribution node at the disturbance time is decomposed into charging power change and discharging power change to form the corresponding power disturbance feature vector of the distribution node. For any distribution node, the power change at the disturbance time is denoted as... The change in power Break it down into changes in charging power and change in discharge power And form the power disturbance feature vector of any distribution node at the time of disturbance. ,in, Let denote the maximum value function, and let be the power disturbance eigenvector of the u-th distribution node at the disturbance time. Similarly, the power disturbance feature vector of each distribution node at the time of disturbance is obtained, and then the power disturbance feature vectors of all distribution nodes are used to form the power disturbance feature matrix at any time of significant power change. .

[0034] The branch current disturbance characteristic matrix is ​​formed by obtaining the current of each branch of the distribution network at the time of the disturbance, and constructing the branch current disturbance characteristic matrix at any time of significant power change. ,in, Let nl represent the current of the first branch at disturbance time y-1, and nl represent the number of branches in the distribution network. A switching feature vector is formed by obtaining the switching feature values ​​of any distribution node and each of its adjacent distribution nodes at the disturbance time. In this context, an adjacent distribution node refers to any other distribution node in the distribution network that has a direct connection to any other distribution node. This indicates the number of adjacent distribution nodes for any given distribution node. Let represent the switching characteristic value of the u-th distribution node and its first adjacent distribution node. When the u-th distribution node and its first adjacent distribution node are in a closed connection state, the corresponding switching characteristic value is 1; conversely, when the u-th distribution node and its first adjacent distribution node are in an open connection state, the corresponding switching characteristic value is 0. Similarly, obtain the switching characteristic vector of each distribution node at the time of disturbance, and form the effective adjacent disturbance characteristic set of all distribution nodes at any time of significant power change. .

[0035] Furthermore, the voltage change at each distribution node at any given moment of significant power change is used to construct a voltage change matrix for that moment of significant power change. Simultaneously, the current change of each branch at any moment of significant power change is obtained to form a branch current change matrix. And obtain the change in active power loss of the distribution network at any time of significant power change. ,in, This represents the active power loss of the distribution network at any moment of significant power change. This represents the active power loss of the distribution network at the time preceding any significant change in power.

[0036] A training sample is formed by combining the operating disturbance feature matrix, power disturbance feature matrix, branch current disturbance feature matrix, effective adjacent disturbance feature set, voltage change matrix, branch current change matrix, and active power loss change at any moment of significant power change. Similarly, training samples are obtained for each moment of significant power change within a preset historical time period.

[0037] Finally, the graph structure model of the distribution network is trained using the operating disturbance feature matrix, power disturbance feature matrix, branch current disturbance feature matrix, and effective adjacent disturbance feature set as inputs, and the voltage change matrix, branch current change matrix, and active power loss change matrix as outputs. The trained graph structure model is obtained by using all training samples. The training of the graph structure model is known to be an existing technique, and is only briefly described here: First, initialize the hidden features of the power distribution nodes:

[0038] in, Let represent the initial hidden feature vector of the u-th distribution node. and This represents the feature mapping matrix to be trained. This represents the operating characteristic vector of the u-th distribution node. Let represent the power disturbance characteristic vector of the u-th distribution node. This represents the bias vector to be trained. This represents the hyperbolic tangent activation function. Using the hyperbolic tangent function can preserve the positive and negative features of the data after scaling during the data preparation stage.

[0039] Then, for each valid branch ( ), calculate the association score between two connected distribution nodes at each level:

[0040] in, Indicates the first The association score of the edge between the u-th distribution node and the b-th distribution node in layer 1. , , , and The parameter matrix and vector obtained through training, and They represent the first Hidden features of the u-th and b-th distribution nodes at the lower level.

[0041] The higher the correlation score between distribution nodes corresponding to effective branches, the higher the weight of their transmitted information should be. Therefore, for each distribution node, the weight of the transmitted information of its effective adjacent distribution nodes to that distribution node is calculated:

[0042] in, This represents the weight of the information transmitted from the b-th distribution node to the u-th distribution node. Indicates the first The association score of the edge between the u-th distribution node and the b-th distribution node in layer 1. This represents the set of all valid neighboring nodes of the u-th distribution node. Indicates the first The association score of the edge between the u-th distribution node and the j-th distribution node in the layer.

[0043] Therefore, the adjacent messages received by each distribution node are:

[0044] in, This represents the neighbor message received by the u-th distribution node. This represents the set of all valid neighboring nodes of the u-th distribution node. This represents the weight of the information transmitted from the b-th distribution node to the u-th distribution node. This represents the hidden features of the b-th distribution node. This is the parameter matrix to be trained. If the set of effective neighbor nodes is empty, then set it directly. It is a zero vector with the same dimension as the hidden features of the node (no need to calculate the weight step and this step again), that is, the distribution node is updated only according to its own operating features.

[0045] Update the state of each distribution node based on its own hidden features and received neighbor messages:

[0046] in, This represents the next-level state of the u-th distribution node. Indicates the first The hidden features of the u-th distribution node at the lower level. This represents the neighbor message received by the u-th distribution node. and The parameter matrix to be trained, This represents the vector to be trained.

[0047] The graph propagation layer number L is set to the maximum value of the shortest path length between any two distribution nodes at the current moment, which is calculated based on the topology of the distribution network at the current moment. After completing L-layer graph propagation, the voltage change of the distribution nodes is predicted based on the final hidden features and branch current characteristics of the distribution nodes (only the hidden features of the distribution nodes are needed). (Obtained through training), branch current change (requires vector concatenation of the hidden features of the distribution nodes at both ends of the branch and the branch current, followed by activation using the hyperbolic tangent function). The parameters used in the prediction calculation process are all obtained through training. The parameters are calculated based on the response characteristics of the branches and the change in active power loss of the distribution network (the average value of the hidden characteristics of all distribution nodes and the average value of the response characteristics of the effective branches are spliced ​​together to generate the full-map response characteristics, and the full-map response characteristics are calculated).

[0048] The predicted values ​​of voltage change, branch current change, and active power loss change of each distribution node output by the graph structure model are compared with the corresponding measured monitoring targets. The training loss of the three monitoring targets is calculated (Here, Huber loss is used to calculate the error). The training losses are summarized to obtain the total training loss of the model. The gradient of the model parameters is calculated in reverse based on the total training loss of the model, and the model parameters are updated until the validation set loss no longer decreases. The trained graph structure model is then obtained as the power change graph response model of the distribution network.

[0049] Step S102: For any distribution node, analyze the reliable charging and discharging power of each charging vehicle at any distribution node at the current time to obtain the real-time power disturbance feature matrix corresponding to any distribution node; obtain the operating disturbance feature matrix, branch current disturbance feature matrix and effective adjacent disturbance feature set at the current time, and use the real-time power disturbance feature matrix corresponding to any distribution node as the input of the trained graph structure model to output the model output result corresponding to any distribution node.

[0050] The electrical location value of each distribution node is influenced not only by other distribution nodes but also by the current load and charging / discharging direction. For example, discharging vehicles at the end of a feeder during peak evening charging hours may significantly improve local low voltage, but continuing to discharge during midday when photovoltaic power is strong may exacerbate local high voltage. Therefore, it is necessary to analyze the impact of each distribution node's charging or discharging on the overall network voltage, branch load, and network losses by considering the current state of each distribution node.

[0051] Since charging vehicles are not always connected to the charging port and may leave at any time, the allocation of charging and discharging power must not only improve energy utilization efficiency but also ensure the vehicle's available power for the entire journey. This means that the vehicle's rated power does not equal the power it can reliably operate at during its entire stay. Therefore, for any given power distribution node, a reliable charging and discharging power analysis is performed on each charging vehicle at that node at the current moment, resulting in the real-time power disturbance characteristic matrix corresponding to that node. Let any charging vehicle at any distribution node at the current time be the target vehicle. Considering that the dwell time of a vehicle at the charging interface is often affected by usage demand, such as the relatively low likelihood of vehicles staying during peak hours like commuting and the potentially longer dwell time during rest periods like nighttime, the likelihood of the target vehicle maintaining a connection at the scheduled time is estimated by combining the dwell time of historical charging vehicles. This allows vehicles with a higher probability of maintaining a connection to be prioritized for scheduling. Furthermore, the charging access times of all historical charging vehicles at any distribution node and the charging access time of the target vehicle are obtained. The charging access time only refers to a specific time (e.g., 9:00 AM, 10:00 AM), without distinguishing between specific dates. The K-means clustering algorithm is used to cluster all charging access times, where the K value is determined by the elbow method. Historical charging vehicles in the cluster containing the target vehicle are considered as vehicles from the same time period.

[0052] Calculate the dwell time of each vehicle from the moment of charging connection to the moment of departure during the same period. Based on the sum of the charging access time and each dwell time of the target vehicle, the estimated departure time of the target vehicle is obtained. The number of vehicles with estimated departure times later than the current time corresponding to the same period is counted and recorded as the first number. The number of vehicles with estimated departure times later than the waiting-to-be-scheduled time corresponding to the same period is counted and recorded as the second number. The ratio of the first number to the second number is taken as the probability that the target vehicle will remain from the current time to the waiting-to-be-scheduled time. The probability that the target vehicle will remain from the current time to the waiting-to-be-scheduled time is then recorded. , Indicates the first quantity. Indicates the second quantity.

[0053] Vehicle charging and discharging scheduling must be conducted within the allowable range of the vehicle battery. Therefore, using the aforementioned calculation formulas for maximum discharge power and maximum chargeable power, based on the target vehicle's current SOC value, the minimum SOC value required for departure, and the upper limit of the allowable SOC value, and considering the probability that the target vehicle will remain until the scheduled time, the maximum discharge power and maximum chargeable power of the target vehicle at the current moment are obtained. Similarly, the maximum discharge power and maximum chargeable power of each charging vehicle at any distribution node at the current moment are obtained. Then, based on the maximum discharge power and maximum chargeable power of each charging vehicle at any distribution node at the current moment, and the active power of any distribution node at each historical moment within a preset historical period, the test power change amplitude of any distribution node at the current moment is obtained. Based on the active power of any distribution node at each historical moment within a preset historical period, calculate the change in active power of any distribution node at each historical moment, and record the change in active power of the u-th distribution node at the i-th historical moment. Based on the absolute value of each change in active power, the median of the absolute values ​​is taken as the typical power change amplitude for any given distribution node. To characterize the feasibility of charging vehicles at the current moment, the maximum discharge power and maximum charge power of each charging vehicle at any power distribution node at the current moment are further combined to calculate the amplitude of the test power change of any power distribution node at the current moment:

[0054] in, This represents the magnitude of the test power change at any given distribution node at the current moment. This represents the function that takes the minimum value. This represents the typical power variation amplitude of any of the aforementioned distribution nodes. This indicates the number of charging vehicles at any given power distribution node at the current moment. This represents the function that takes the maximum value. This represents the maximum discharge power of the e-th charging vehicle at any of the aforementioned power distribution nodes at the current moment. This represents the maximum discharge power of the e-th charging vehicle at any of the aforementioned power distribution nodes at the current moment. Used to ensure that power changes can be performed.

[0055] If the total power of the distribution network regulation at the current moment is greater than 0, it is necessary to analyze the electrical position value of any distribution node in the charging direction. In this case, the power changes of other distribution nodes need to be set to 0. Therefore, based on the measured power change amplitude of any distribution node at the current moment, the power disturbance feature vector of any distribution node at the current moment is set. , This represents the amplitude of the test power change of any given distribution node at the current moment, while simultaneously setting the power disturbance feature vectors of other distribution nodes at the current moment. The power disturbance feature vectors of all distribution nodes at the current moment are used to form the real-time power disturbance feature matrix corresponding to any distribution node. .

[0056] If the total power of the distribution network regulation at the current moment is less than 0, it is necessary to analyze the electrical position value of any distribution node in the discharge direction. Similarly, it is necessary to set the power change of other distribution nodes to 0. Therefore, based on the measured power change amplitude of any distribution node at the current moment, the power disturbance feature vector of any distribution node at the current moment is set. Simultaneously, the power disturbance feature vectors of other distribution nodes at the current moment are set. The power disturbance feature vectors of all distribution nodes at the current moment are used to form the real-time power disturbance feature matrix corresponding to any distribution node. .

[0057] Simultaneously, following the methods described above for obtaining each training sample, the operational perturbation feature matrix at the current moment is obtained. Branch current disturbance characteristic matrix and the set of effective adjacency perturbation features The real-time power disturbance feature matrix corresponding to any of the power distribution nodes at the current moment. This forms new input data, which is then input into the trained graph structure model, corresponding to the output voltage change matrix. Branch current change matrix and changes in active power loss .

[0058] Step S103: Using the model output results, perform deviation analysis on the active power loss of the distribution network at the current moment and the voltage and branch current of any distribution node to obtain the electrical location value of any distribution node at the current moment; obtain the electrical location value of each distribution node at the current moment, and perform charging and discharging regulation on the distribution nodes of the distribution network.

[0059] The model output includes a voltage change matrix, a branch current change matrix, and an active power loss change matrix. Therefore, by using the voltage of each distribution node at the current moment and the voltage change matrix in the model output, the model can be optimized. The calculation formula is used to obtain the predicted voltage value for each distribution node, where, This represents the predicted voltage value for the u-th distribution node. This represents the predicted voltage change corresponding to the u-th distribution node in the voltage change matrix of the model output. This represents the voltage of the u-th distribution node at the current moment. Similarly, based on the current of each branch at the current moment and the branch current change matrix in the model output, the predicted current value of each branch is obtained; based on the active power loss of the distribution network at the current moment and the change in active power loss in the model output, the predicted active power loss value of the distribution network is obtained.

[0060] Since the closer the voltage is to the rated voltage, the more stable the distribution network operation; the smaller the branch current, the smaller the distribution network load; and the smaller the network loss, the better the distribution network condition, the electrical location value of any distribution node at the current moment is calculated based on the differences between the voltage and the predicted voltage value of each distribution node, the differences between the current and the predicted current value of each branch, and the differences between the active power loss and the predicted active power loss value of the distribution network.

[0061] in, This represents the electrical location value of any distribution node at the current moment, where N represents the number of distribution nodes. This represents the voltage of the j-th distribution node at the current moment. This represents the rated voltage of the j-th distribution node. This represents the distance from the rated voltage of the j-th distribution node to the allowable voltage boundary (set to 0.07 × rated voltage according to current standards). This represents the predicted voltage value of the j-th distribution node at the current moment. This indicates the total number of branches in the distribution network at the current moment. Indicates the next time step. The current in each branch, Indicates the first The historical maximum current value of each branch Indicates the next time step. Predicted current values ​​for each branch. This represents the predicted active power loss of the distribution network at the current moment. This represents the active power loss of the distribution network at the current moment. This represents the reference active power loss of the distribution network under rated operating conditions. It is a very small constant used to prevent the denominator from being 0.

[0062] because There is a need to increase the charging power of new energy vehicles; Therefore, it is necessary to increase the discharge power of new energy vehicles. Thus, if the total power of the distribution network at the current moment is greater than 0, the power disturbance feature vector of any distribution node in the charging direction at the current moment will be... Power disturbance feature vectors of other distribution nodes at the current moment The real-time power disturbance feature matrix corresponding to any of the power distribution nodes is formed. The system utilizes a trained graph structure model to obtain the model output in the charging direction. Then, based on the aforementioned method for calculating electrical position value, it obtains the electrical position value in the charging direction. Similarly, it obtains the electrical position value of each distribution node in the charging direction at the current moment and allocates power to each distribution node sequentially in descending order of electrical position value. If two distribution nodes have the same electrical position value, the distribution node with the larger cumulative maximum rechargeable power of the charging vehicles is prioritized to reduce the number of distribution nodes that need to be controlled simultaneously. Within each distribution node, power is allocated sequentially in descending order of the maximum rechargeable power of each charging vehicle. If the remaining total power to be allocated is less than the maximum rechargeable power of the charging vehicle, the remaining total power to be allocated is entirely allocated to that charging vehicle.

[0063] Similarly, if the total power of the distribution network at the current moment is less than 0, then the power disturbance feature vector of any distribution node in the discharge direction at the current moment will be... Power disturbance feature vectors of other distribution nodes at the current moment The real-time power disturbance feature matrix corresponding to any of the power distribution nodes is formed. The system utilizes a trained graph structure model to obtain the model output in the discharge direction. Then, based on the aforementioned method for calculating electrical position value, it obtains the electrical position value in the discharge direction. Similarly, it obtains the electrical position value of each distribution node in the discharge direction at the current moment and allocates power to each distribution node sequentially in descending order of electrical position value. If two distribution nodes have the same electrical position value, the distribution node with the larger cumulative maximum discharge power of the charging vehicles is prioritized to reduce the number of distribution nodes that need to be controlled simultaneously. Within each distribution node, power is allocated sequentially in descending order of the maximum discharge power of each charging vehicle. If the remaining total power to be allocated is less than the maximum discharge power of the charging vehicle, the remaining total power to be allocated is entirely allocated to that charging vehicle.

[0064] 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 the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A method for coordinated optimization of charging and discharging of new energy vehicles based on vehicle-to-grid interaction, characterized in that, The method includes: When the distribution network receives a power dispatch signal and the total power of the distribution network at the current moment is not equal to 0, the historical monitoring data of each distribution node, each charging pile under the distribution node, and each historical charging vehicle under the charging pile are obtained in a preset historical period. The graph structure model of the distribution network is trained to obtain a trained graph structure model. The input of the trained graph structure model is the running disturbance feature matrix, the power disturbance feature matrix, the branch current disturbance feature matrix, and the effective adjacent disturbance feature set. For any distribution node, the reliable charging and discharging power of each charging vehicle at the current time is analyzed to obtain the real-time power disturbance feature matrix corresponding to the distribution node; the operating disturbance feature matrix, branch current disturbance feature matrix and effective adjacent disturbance feature set at the current time are obtained, and combined with the real-time power disturbance feature matrix corresponding to the distribution node as the input of the trained graph structure model, and the model output result corresponding to the distribution node is output. Using the model output results, deviation analysis is performed on the active power loss of the distribution network at the current moment, as well as the voltage and branch current of any distribution node, to obtain the electrical location value of any distribution node at the current moment; the electrical location value of each distribution node at the current moment is obtained, and the charging and discharging regulation of the distribution nodes of the distribution network is performed.

2. The method for coordinated optimization of charging and discharging of new energy vehicles based on vehicle-to-grid interaction according to claim 1, characterized in that, The process involves acquiring historical monitoring data for each distribution node of the distribution network, each charging pile under the distribution node, and each historically charging vehicle under the charging pile within a preset historical time period. This data is then used to train a graph structure model of the distribution network, resulting in a trained graph structure model, including: For any power distribution node, based on the power value of each charging interface of each charging pile under that power distribution node at each historical moment within a preset historical period, the power change at each historical moment is obtained. The median of the absolute values ​​of all power changes is taken as the power change threshold. If the absolute value of the power change at any historical moment is greater than the power change threshold, then that historical moment is recorded as the main power change moment of that power distribution node. Based on the main power change moments of each power distribution node, if any historical moment within the preset historical period is the main power change moment of at least one power distribution node, then that historical moment is recorded as the moment of significant power change. For any moment of significant power change within a preset historical time period, the previous moment of that moment of significant power change is taken as the disturbance moment. The operating feature vector of each distribution node at the disturbance moment is obtained to form the operating disturbance feature matrix of that moment of significant power change. The power change of each distribution node at the disturbance moment is decomposed into the charging power change and the discharging power change to form the power disturbance feature vector of the corresponding distribution node. The power disturbance feature vectors of all distribution nodes are combined to form the power disturbance feature matrix of that moment of significant power change. The current of each branch of the distribution network at the time of the disturbance is obtained to form the branch current disturbance feature matrix at any time of significant power change; the switching feature values ​​of any distribution node and each of its adjacent distribution nodes at the time of the disturbance are obtained to form a switching feature vector, and the switching feature vectors of all distribution nodes are used to form the effective adjacent disturbance feature set at any time of significant power change. Obtain the voltage change and branch current change of each distribution node at any time when the power changes significantly, form a voltage change matrix and a branch current change matrix at any time when the power changes significantly, and obtain the active power loss change of the distribution network at any time when the power changes significantly. A training sample is formed by combining the operating disturbance feature matrix, power disturbance feature matrix, branch current disturbance feature matrix, effective adjacent disturbance feature set, voltage change matrix, branch current change matrix, and active power loss change at any moment of significant power change. The training samples corresponding to each moment of significant power change within a preset historical period are obtained. The operating disturbance feature matrix, power disturbance feature matrix, branch current disturbance feature matrix, and effective adjacent disturbance feature set are taken as inputs, and the voltage change matrix, branch current change matrix, and active power loss change are taken as outputs. The graph structure model of the distribution network is trained using all training samples to obtain the trained graph structure model.

3. The method for coordinated optimization of charging and discharging of new energy vehicles based on vehicle-to-grid interaction according to claim 2, characterized in that, The step of obtaining the power change of each distribution node at each historical moment within a preset historical period based on the power value of each charging interface of each charging pile under each distribution node at each historical moment within a preset historical period includes: For any power distribution node, taking any historical moment within a preset historical period as an example, the power value of each charging interface of each charging pile under the power distribution node is accumulated at any historical moment to obtain the total power value of the power distribution node at any historical moment. The difference between the total power value at any historical moment and the total power value at the previous moment is taken as the power change of the power distribution node at any historical moment.

4. The method for coordinated optimization of charging and discharging of new energy vehicles based on vehicle-to-grid interaction according to claim 3, characterized in that, The step of obtaining the operating feature vector of each distribution node at the time of the disturbance includes: For any power distribution node, obtain the maximum rechargeable power and maximum dischargeable power of each historical charging vehicle at the time of the disturbance, and obtain the cumulative value of the maximum rechargeable power and the cumulative value of the maximum dischargeable power. Obtain the voltage, active power, reactive power, active power of distributed power sources, and total power value of any distribution node at the time of the disturbance, and combine them with the accumulated value of the maximum rechargeable power and the accumulated value of the maximum dischargeable power to form the operating characteristic vector of any distribution node at the time of the disturbance.

5. The method for coordinated optimization of charging and discharging of new energy vehicles based on vehicle-to-grid interaction according to claim 2, characterized in that, The step of decomposing the power change of each distribution node at the time of the disturbance into a power disturbance feature vector consisting of a charging power change and a discharging power change, includes: For any distribution node, let the power change of that distribution node at the time of the disturbance be denoted as . The change in power Break it down into changes in charging power and change in discharge power And form the power disturbance feature vector of any of the distribution nodes. ,in, This represents the function that takes the maximum value.

6. The method for coordinated optimization of charging and discharging of new energy vehicles based on vehicle-to-grid interaction according to claim 4, characterized in that, The step of obtaining the maximum rechargeable power and maximum dischargeable power of each historical charging vehicle at any power distribution node at the time of the disturbance includes: For any historical charging vehicle at any distribution node at the time of the disturbance, the dischargeable energy and dischargeable energy of the historical charging vehicle at the time of the disturbance are obtained based on the SOC value of the historical charging vehicle at the time of the disturbance, the minimum SOC value required for departure, and the upper limit of the allowable SOC value: ； ； in, This represents the dischargeable energy of any historically charged vehicle at the time of the disturbance. This represents the rechargeable energy of any historical charging vehicle at the time of the disturbance. This indicates the battery capacity of any of the historically charged vehicles. This represents the function that takes the maximum value, and 0 represents a constant. This represents the SOC value of any historically charged vehicle at the time of the disturbance. This represents the minimum SOC value required for any historically charged vehicle to leave the site at the time of the disturbance. This indicates the maximum allowable SOC value for any historically charged vehicle. By combining the dischargeable energy and rechargeable energy of any historical charging vehicle at the time of the disturbance, as well as the probability that any historical charging vehicle remains until the time of scheduling, the maximum dischargeable power and maximum rechargeable power of any historical charging vehicle are obtained respectively: ； ； in, This indicates the maximum discharge power of any historically charged vehicle. This indicates the maximum rechargeable power of any of the historically charged vehicles. This represents the function that takes the minimum value. This represents the rated discharge power limited by both the historically charged vehicle and its charging station. This represents the rated charging power limited by both the historically charged vehicle and its charging station. This represents the time interval between the disturbance time and the time to be scheduled, where the time to be scheduled refers to the last moment of the scheduling cycle. The probability that any historical charging vehicle remains until the time to be scheduled is represented by a constant value of 1.

7. The method for coordinated optimization of charging and discharging of new energy vehicles based on vehicle-to-grid interaction according to claim 1, characterized in that, The analysis of the reliable charging and discharging power of each charging vehicle at any of the current power distribution nodes yields the real-time power disturbance feature matrix corresponding to any given power distribution node, including: Let any charging vehicle at any distribution node at the current time be the target vehicle. Obtain the charging access time of all historical charging vehicles at any distribution node and the charging access time of the target vehicle. Cluster all charging access times and take the historical charging vehicles in the cluster where the target vehicle is located as vehicles in the same time period. Calculate the dwell time of each vehicle from the charging access time to the departure time during the same period. Based on the sum of the charging access time of the target vehicle and each dwell time, obtain the estimated departure time of the target vehicle. Count the number of vehicles in the same period whose estimated departure time is later than the current time, and record it as the first quantity. Count the number of vehicles in the same period whose estimated departure time is later than the time to be scheduled, and record it as the second quantity. The ratio of the first quantity to the second quantity is used as the probability that the target vehicle will remain from the current time to the time to be scheduled. Based on the target vehicle's SOC value at the current moment, the minimum SOC value required for departure, and the upper limit of the allowed SOC value, and combined with the probability that the target vehicle will remain at the current moment until the time to be scheduled, the maximum discharge power and maximum chargeable power of the target vehicle at the current moment are obtained. Based on the maximum discharge power and maximum chargeable power of each charging vehicle at any power distribution node at the current time, and the active power of any power distribution node at each historical moment within a preset historical period, the test power change amplitude of any power distribution node at the current time is obtained. If the total power of the distribution network regulation at the current moment is greater than 0, then based on the amplitude of the measured power change, the power disturbance feature vector of any distribution node at the current moment is set. , This represents the amplitude of the test power change of any given distribution node at the current moment, while simultaneously setting the power disturbance feature vectors of other distribution nodes at the current moment. The power disturbance feature vectors of all distribution nodes at the current moment are used to form the real-time power disturbance feature matrix corresponding to any distribution node. If the total power of the distribution network regulation at the current moment is less than 0, then based on the amplitude of the measured power change, the power disturbance feature vector of any distribution node at the current moment is set. Simultaneously, the power disturbance feature vectors of other distribution nodes at the current moment are set. The power disturbance feature vectors of all distribution nodes at the current moment are used to form the real-time power disturbance feature matrix corresponding to any distribution node.

8. The method for coordinated optimization of charging and discharging of new energy vehicles based on vehicle-to-grid interaction according to claim 7, characterized in that, The step of obtaining the test power change amplitude of any distribution node at the current moment based on the maximum discharge power and maximum charge power of each charging vehicle at any distribution node at the current moment, and the active power of any distribution node at each historical moment within a preset historical period, includes: Based on the active power of any distribution node at each historical moment within a preset historical period, calculate the change in active power of any distribution node at each historical moment, and take the median of the absolute values ​​of each change in active power as the typical power change amplitude of any distribution node. Based on the maximum discharge power and maximum charge power of each charging vehicle at any given power distribution node at the current moment, and the typical power change amplitude of any given power distribution node, calculate the test power change amplitude of any given power distribution node at the current moment: ； in, This represents the magnitude of the test power change at any given distribution node at the current moment. This represents the function that takes the minimum value. This represents the typical power variation amplitude of any of the aforementioned distribution nodes. This indicates the number of charging vehicles at any given power distribution node at the current moment. This represents the function that takes the maximum value. This represents the maximum discharge power of the e-th charging vehicle at any of the aforementioned power distribution nodes at the current moment. This represents the maximum discharge power of the e-th charging vehicle at any of the power distribution nodes at the current moment.

9. The method for coordinated optimization of charging and discharging of new energy vehicles based on vehicle-to-grid interaction according to claim 1, characterized in that, The model output results are used to perform deviation analysis on the active power loss of the distribution network at the current moment, as well as the voltage and branch current of any distribution node, to obtain the electrical location value of any distribution node at the current moment, including: The model output includes a voltage change matrix, a branch current change matrix, and an active power loss change matrix. Based on the model output, the predicted voltage change of each distribution node, the predicted current change of each branch, and the predicted active power loss change of the distribution network at the current moment are obtained. Then, by combining the voltage of any distribution node, the current of each branch, and the active power loss of the distribution network at the current moment, the predicted voltage value of each distribution node, the predicted current value of each branch, and the predicted active power loss value of the distribution network are obtained. Based on the differences between the voltage and predicted voltage values ​​of each distribution node at the current moment, the differences between the current and predicted current values ​​of each branch, and the differences between the active power loss and predicted active power loss values ​​of the distribution network, calculate the electrical location value of any distribution node at the current moment: ； in, This represents the electrical location value of any distribution node at the current moment, where N represents the number of distribution nodes. This represents the voltage of the j-th distribution node at the current moment. This represents the rated voltage of the j-th distribution node. This represents the distance from the rated voltage to the allowable voltage boundary of the j-th distribution node. This represents the predicted voltage value of the j-th distribution node at the current moment. This indicates the total number of branches in the distribution network at the current moment. Indicates the next time step. The current in each branch, Indicates the first The historical maximum current value of each branch Indicates the next time step. Predicted current values ​​for each branch. This represents the predicted active power loss of the distribution network at the current moment. This represents the active power loss of the distribution network at the current moment. This represents the reference active power loss of the distribution network under rated operating conditions. It is a very small constant used to prevent the denominator from being 0.

10. The method for coordinated optimization of charging and discharging of new energy vehicles based on vehicle-to-grid interaction according to claim 1, characterized in that, The step of obtaining the electrical location value of each distribution node at the current moment and performing charging and discharging control on the distribution nodes of the distribution network includes: The electrical position value includes the electrical position value in the charging direction and the electrical position value in the discharging direction. If the total power of the distribution network regulation at the current moment is greater than 0, the electrical position value of each distribution node in the charging direction at the current moment is obtained, and power is allocated to each distribution node in descending order of electrical position value. If the total power of the distribution network regulation at the current moment is less than 0, the electrical position value of each distribution node in the discharging direction at the current moment is obtained, and power is allocated to each distribution node in descending order of electrical position value.

Images