CN-122020133-A - Power distribution network electric automobile bearing capacity evaluation method and system

CN122020133ACN 122020133 ACN122020133 ACN 122020133ACN-122020133-A

Abstract

The invention discloses a method and a system for evaluating the bearing capacity of an electric automobile of a power distribution network. Aiming at a target power distribution network, analyzing interest points and electric vehicle parameters, and simulating charging behaviors by using an intelligent body-cellular automaton model to generate diversified charging scenes. The intelligent ammeter is used for collecting actual operation data, a DistFlow model is used for calculating node power of each charging point, the model is solved in different time intervals, a unscented Kalman filtering algorithm is used for processing a power flow relation, and whether voltage and line load meet constraint conditions or not is checked to judge operation feasibility. Then, an optimized random forest model is established, key features are extracted and trained, mapping between the features and scene feasibility is established, the capacity of the power distribution network for bearing the charging load of the electric automobile is evaluated, dependence on a large number of simulation scenes can be reduced, calculation complexity is reduced, uncertainty factors are better processed, and the requirement of real-time scheduling of the power grid is met.

Inventors

GUO NING
Yang Yunteng
ZHANG CHI
XIE WENQIANG
XUE ZHITONG
GUO JIAHAO
WU FAN
LU XIAOXING
GUO ZIRAN
CHEN YE
YUAN YUBO
CHEN JINMING
ZHOU ZHIJUN
GUO YAJUAN
LU QINGNING
JIA MENGMENG
SONG SHUANG

Assignees

江苏省电力试验研究院有限公司
国网江苏省电力有限公司电力科学研究院

Dates

Publication Date: 20260512
Application Date: 20260413

Claims (10)

1. The method for evaluating the bearing capacity of the electric automobile of the power distribution network is characterized by comprising the following steps of: aiming at a target power distribution network, analyzing interest points and electric vehicle parameters in a target area, and simulating charging behaviors by using an intelligent body-cellular automaton model so as to generate diversified electric vehicle charging scenes; Collecting actual running data of a power grid by using the electric automobile charging scene, representing a power flow relation of a target power distribution network by using a DistFlow model, calculating node power of each electric automobile charging connection point, solving a DistFlow model in different time intervals, processing the power flow relation of the target power distribution network by using a unscented Kalman filtering algorithm based on a solving result, defining and solving a feasibility problem, and judging the running feasibility in each scene by checking whether a voltage level and a line load meet constraint conditions; Establishing an optimized random forest model, extracting key features of the electric vehicle charging scene, training the key features by adopting the optimized random forest model, establishing a mapping between the features and scene feasibility, and evaluating the bearing capacity of the power distribution network to the electric vehicle charging load.
2. The method for evaluating the bearing capacity of an electric vehicle in a power distribution network according to claim 1, wherein the analyzing interest points and electric vehicle parameters in a target area for a target power distribution network and simulating charging behaviors by using an agent-cellular automaton model to generate diversified electric vehicle charging scenes comprises: aiming at a target power distribution network, analyzing the distribution situation of interest points in a target power distribution network area so as to determine the types of the divided areas and determine the boundary range of the electric vehicles in a concentrated way; Randomly selecting samples based on historical data to obtain basic parameters of the electric automobile, abstracting the basic parameters into an intelligent body, setting attribute and behavior rules, and constructing a cellular automaton environment to obtain an intelligent body-cellular automaton model; and simulating charging behaviors by using the intelligent body-cellular automaton model so as to generate diversified electric automobile charging scenes.
3. The method for evaluating the bearing capacity of an electric vehicle in a power distribution network according to claim 2, wherein the analyzing, for a target power distribution network, a distribution situation of interest points in a target power distribution network area to divide types of areas, and determining a boundary range where electric vehicles appear in a centralized manner includes: dividing a target power distribution network area according to a custom scale to cover a living area, a business area and a public service area; and calculating the frequency density ratio of different functional areas according to the number and the weight of the interest points, and if the frequency density ratio of a certain class of interest points exceeds a threshold value, defining the area as a corresponding functional area, wherein the functional area comprises a residential area, a business area or a public service area.
4. The method for evaluating the bearing capacity of the electric automobile of the power distribution network according to claim 2, wherein the randomly selecting samples based on the historical data to obtain the basic parameters of the electric automobile, abstracting the basic parameters into an agent, setting attributes and behavior rules, and constructing a cellular automaton environment to obtain an agent-cellular automaton model comprises: Abstracting each electric automobile into an intelligent agent, and defining the electric automobile as a basis and dynamic attributes of the intelligent agent, wherein the basis attributes comprise a starting point, an ending point, an initial SOC, a battery capacity and a rated endurance mileage; Simulating the selection of the departure place and the destination of the vehicle through historical data; Making a driving rule and a charging decision rule of the intelligent agent, wherein when the residual electric quantity is smaller than a minimum limit value, if the residual electric quantity cannot reach a terminal point, the charging of a nearest charging station is selected, otherwise, the charging is carried out after the terminal point is reached; Dividing a road network into cells, defining a neighborhood relation of the cells, supporting the movement of an intelligent agent among the cells by adopting an extended Moore neighborhood structure, and recording the current number of vehicles and the congestion degree of each cell; And setting behavior parameters of the electric automobile according to the characteristics of the functional areas, wherein the behavior parameters comprise maximum speed, random deceleration probability and energy consumption in different areas, so as to obtain an intelligent body-cellular automaton model.
5. The method for evaluating the bearing capacity of an electric vehicle in a power distribution network according to claim 2, wherein the simulating the charging behavior by using the intelligent agent-cellular automaton model to generate diversified electric vehicle charging scenes comprises: A-based use of an agent-cellular automaton model The shortest path planning method of the algorithm determines a driving route for each agent; adjusting the travelling speed of the vehicle according to the speed limit of the area, and updating the position of the vehicle at a certain time; updating the residual electric quantity of the electric automobile according to the driving distance and the corresponding energy consumption, and when the residual electric quantity is detected to be lower than the set minimum limit value, guiding the automobile to a nearest charging station for charging; The method comprises the steps of counting the number of electric vehicles in each cell and charging requirements thereof, recording charging load change conditions at different time points, and analyzing and obtaining charging requirement distribution conditions of the electric vehicles in different time periods and different functional areas based on a space-time distribution evolution model of the charging requirements so as to obtain an electric vehicle charging scene.
6. The method for evaluating the bearing capacity of an electric vehicle in a power distribution network according to claim 1, wherein the step of using DistFlow model to characterize the trend relationship of the target power distribution network comprises the steps of: obtaining a topology model N= (V, E) of a target power distribution network, wherein V is a node set, In each cell, the electric vehicle charges by selecting a charging station closest to the electric vehicle, and the load demand of the charging station is accumulated by the charging requests of all electric vehicles in the cell and mapped to each node of a power distribution network; Based on advanced measurement technology, the intelligent ammeter is used for measuring actual operation data of the power grid, including active and reactive power injection of nodes and amplitude of node voltage, through The expression DistFlow model, wherein, P and q are respectively the active and reactive power injection of the node, v is the square of the voltage amplitude of the node, P, Q is the active and reactive power flow of the line; And Respectively the lines Constraints include active/reactive power flow and voltage limits on the line, and DistFlow model is written as: 。
7. The method for evaluating the bearing capacity of an electric vehicle in a power distribution network according to claim 1, wherein the calculating the node power of each charging connection point of the electric vehicle and solving DistFlow models in different time intervals includes: the candidate access position of the electric automobile is as follows Scene of Indicating electric automobile resources accessed at position L and net injection or consumption of each node in period T The net power curve is ; For each electric automobile scene psi, the scene psi contains the position, the quantity and the charging behaviors, and the scene psi is obtained by A model is obtained for a corresponding time step t, wherein, The squares of the active power flow, the reactive power flow and the voltage amplitude of the line in the corresponding time step t are respectively; Assuming that the electric automobile works in a constant power charging mode, and maintains a fixed power factor eta through simple reactive power control, and combines a baseline load { , Charge load of electric automobile { and } By (E) } passing 、 Computing node net power injection 、 Wherein the matrix Is an adjacent matrix from the position of the electric automobile to the bus, namely when the jth electric automobile is positioned on the bus i Otherwise 。
8. The method for evaluating the bearing capacity of the electric automobile of the power distribution network according to claim 1, wherein the step of processing the trend relationship of the target power distribution network by using the unscented kalman filter algorithm based on the solved result, defining and solving the feasibility problem, and judging the operation feasibility in each scene by checking whether the voltage level and the line load meet the constraint conditions comprises the following steps: Setting a state and an observation variable of power flow calculation of the power distribution network based on a DistFlow model corresponding to a solving result, initializing an empty load state and an uncertainty matrix thereof, and configuring UKF algorithm parameters according to power distribution network characteristics; Generating Sigma points by using the current state and covariance to represent current state distribution, and predicting future state changes of the Sigma points by DistFlow model to obtain a predicted state mean value and covariance to form a predicted result; converting the prediction result into an observation space, calculating an observation mean value and a covariance, determining a Kalman gain, and updating system state estimation by using actual measurement data; defining a feasibility problem to check whether the voltage and line load constraints meet the conditions, setting upper and lower voltage limits for each node in the time step, and assigning active and reactive power flows and upper apparent power limits for each line; The method comprises the steps of constructing a scene, namely, dividing the scene into two cases of feasibility and infeasibility, and evaluating the overall feasibility of the scene according to the upper limit of the number of times of infeasible solutions allowed to appear in a plurality of time steps, wherein if all the time steps meet voltage and line load constraints, the scene is considered to be feasible, and otherwise, the scene is considered to be infeasible.
9. The method for evaluating the bearing capacity of the electric vehicle of the power distribution network according to claim 1, wherein the steps of establishing an optimized random forest model, extracting key features of the charging scene of the electric vehicle, training the key features by adopting the optimized random forest model, establishing a mapping between the features and the scene feasibility, and evaluating the bearing capacity of the power distribution network to the charging load of the electric vehicle comprise: Extracting quantitative features which obviously influence feasibility from the electric vehicle charging scene, wherein the quantitative features comprise the number and the duty ratio of the electric vehicles, the space-time distribution of charging demands, access positions and net power injection of power distribution network nodes to form feature vectors; determining the real feasibility of the scenes by simulating the voltage and line load constraint meeting conditions of each electric automobile charging scene in a set time period, setting a tag value for each scene according to the real feasibility, and constructing an initial training set; Repeatedly and randomly selecting samples from the initial training set by using a resampling method as training subsets of decision trees, and applying an optimal segmentation strategy to randomly selected variables until the variables cannot be split again for each node to form a random forest structure consisting of a plurality of decision trees, and carrying out regression prediction on each decision tree in the random forest by using the rest samples to verify the performance of the model; Optimizing super parameters of an AdaBoost model through grid search, iteratively optimizing the performance of the whole training tree model based on an AdaBoost idea, weighting each decision tree to train and strengthen the performance of a learner, adjusting the weight of training data according to the prediction error of a previous round of decision tree and retraining a new decision tree after each iteration, and verifying the model prediction accuracy by adopting average absolute error and root mean square error as evaluation indexes; And carrying out weighted summation on the prediction results of all the decision trees to obtain a final prediction result so as to judge whether the voltage/line load constraint of the target area is met or not, and further determining the feasibility of the scene, wherein the node power injection and the voltage maximum value under the constraint condition are obtained by analyzing the characteristic distribution rule of the feasible scene, and the bearing capacity of the charging load of the electric automobile of the power distribution network is determined.
10. Distribution network electric automobile bearing capacity evaluation system, its characterized in that includes: the scene generation unit is used for analyzing interest points and electric vehicle parameters in a target area aiming at a target power distribution network, and simulating charging behaviors by using an intelligent body-cellular automaton model so as to generate diversified electric vehicle charging scenes; The judging unit is used for collecting actual running data of the power grid by using the electric automobile charging scene, characterizing the power flow relation of the target power distribution network by using a DistFlow model, calculating the node power of each electric automobile charging connection point, solving a DistFlow model in different time intervals, processing the power flow relation of the target power distribution network by using an unscented Kalman filtering algorithm based on the solving result, defining and solving the feasibility problem, and judging the running feasibility in each scene by checking whether the voltage level and the line load meet constraint conditions or not; The prediction unit is used for establishing an optimized random forest model, extracting key features of the electric vehicle charging scene, training the key features by adopting the optimized random forest model, establishing a mapping between the features and scene feasibility, and evaluating the bearing capacity of the power distribution network to the electric vehicle charging load.

Description

Power distribution network electric automobile bearing capacity evaluation method and system Technical Field The invention relates to the technical field of smart grids, in particular to a method and a system for evaluating the bearing capacity of an electric car of a power distribution network. Background With the continuous increase of the quantity of electric vehicles in the world, the behavior that the vehicles are connected into a power distribution network for charging gradually becomes a key factor affecting the stability and the electric energy quality of the power grid. Particularly, the high-density charging demands intensively appear in a specific area, so that not only are peak-to-valley fluctuation of the power grid load aggravated, but also the technical problems of voltage out-of-limit, line overload and the like can be caused. Therefore, the load bearing capacity of the power distribution network to the electric automobile is accurately estimated, and the method is very important for ensuring the safe operation of the power distribution network and improving the service quality. However, the conventional evaluation method is generally based on static scene analysis, and the influence of user behavior and traffic conditions on the charging mode of the electric vehicle is not fully considered, thereby limiting the accuracy and practicality of the evaluation result. In practical application, the travel habit and charging selection of the electric automobile user show remarkable space and time diversity, which directly relates to the charging demand distribution in different areas. The traditional deterministic simulation method is difficult to reflect the diversity and dynamic change due to the adoption of fixed parameter setting, so that deviation exists between a predicted result and an actual situation. In addition, the traditional method adopting static tide calculation cannot effectively capture the time-space evolution characteristics in the charging behavior of the electric automobile, and ignores the influence of factors such as traffic flow, congestion and the like on the charging demand, so that the power grid planning and scheduling decision lacks sufficient flexibility and foresight. Therefore, it is necessary to design a new method, by applying artificial intelligence technology, useful information contained in historical data can be deeply mined, and a complex mapping relation between multidimensional features and power grid operation parameters is established, so that more accurate bearing capacity assessment is realized. The method not only can reduce the dependence on a large number of simulation scenes and reduce the calculation complexity, but also can better process uncertainty factors, meet the requirement of real-time scheduling of the power grid, and provide more scientific and reasonable planning suggestions. Disclosure of Invention The invention aims to overcome the defects of the prior art and provides a method and a system for evaluating the bearing capacity of an electric automobile of a power distribution network. In order to achieve the purpose, the invention adopts the following technical scheme that the method for evaluating the bearing capacity of the electric automobile of the power distribution network comprises the following steps: aiming at a target power distribution network, analyzing interest points and electric vehicle parameters in a target area, and simulating charging behaviors by using an intelligent body-cellular automaton model so as to generate diversified electric vehicle charging scenes; Collecting actual running data of a power grid by using the electric automobile charging scene, representing a power flow relation of a target power distribution network by using a DistFlow model, calculating node power of each electric automobile charging connection point, solving a DistFlow model in different time intervals, processing the power flow relation of the target power distribution network by using a unscented Kalman filtering algorithm based on a solving result, defining and solving a feasibility problem, and judging the running feasibility in each scene by checking whether a voltage level and a line load meet constraint conditions; Establishing an optimized random forest model, extracting key features of the electric vehicle charging scene, training the key features by adopting the optimized random forest model, establishing a mapping between the features and scene feasibility, and evaluating the bearing capacity of the power distribution network to the electric vehicle charging load. The invention also provides a system for evaluating the bearing capacity of the electric automobile of the power distribution network, which comprises the following steps: the scene generation unit is used for analyzing interest points and electric vehicle parameters in a target area aiming at a target power distribution network, and simulating charging behaviors by u