CN114465256A - Multi-node electric vehicle charging load joint countermeasure generation interval prediction method - Google Patents
Multi-node electric vehicle charging load joint countermeasure generation interval prediction method Download PDFInfo
- Publication number
- CN114465256A CN114465256A CN202210079212.8A CN202210079212A CN114465256A CN 114465256 A CN114465256 A CN 114465256A CN 202210079212 A CN202210079212 A CN 202210079212A CN 114465256 A CN114465256 A CN 114465256A
- Authority
- CN
- China
- Prior art keywords
- node
- day
- charging
- correlation
- scene
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 64
- 230000008569 process Effects 0.000 claims description 16
- 239000000523 sample Substances 0.000 claims description 15
- 238000005070 sampling Methods 0.000 claims description 8
- 238000010219 correlation analysis Methods 0.000 claims description 7
- 239000011159 matrix material Substances 0.000 claims description 7
- 238000013507 mapping Methods 0.000 claims description 4
- 238000004364 calculation method Methods 0.000 claims description 3
- 239000012468 concentrated sample Substances 0.000 claims description 3
- 239000000126 substance Substances 0.000 claims description 3
- 238000005457 optimization Methods 0.000 claims description 2
- 230000002349 favourable effect Effects 0.000 abstract description 2
- 238000012216 screening Methods 0.000 abstract 1
- 238000004458 analytical method Methods 0.000 description 15
- 238000011160 research Methods 0.000 description 8
- 238000011156 evaluation Methods 0.000 description 7
- 230000006870 function Effects 0.000 description 7
- 230000000694 effects Effects 0.000 description 6
- 238000012549 training Methods 0.000 description 6
- 230000006399 behavior Effects 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 108090000672 Annexin A5 Proteins 0.000 description 3
- 101001095088 Homo sapiens Melanoma antigen preferentially expressed in tumors Proteins 0.000 description 3
- 102100037020 Melanoma antigen preferentially expressed in tumors Human genes 0.000 description 3
- 230000002596 correlated effect Effects 0.000 description 3
- 230000000875 corresponding effect Effects 0.000 description 3
- 230000001419 dependent effect Effects 0.000 description 3
- 238000005259 measurement Methods 0.000 description 3
- 238000012360 testing method Methods 0.000 description 3
- 238000002474 experimental method Methods 0.000 description 2
- 238000000342 Monte Carlo simulation Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 230000001186 cumulative effect Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000005315 distribution function Methods 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000035699 permeability Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000012800 visualization Methods 0.000 description 1
Images
Classifications
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J3/00—Circuit arrangements for ac mains or ac distribution networks
- H02J3/28—Arrangements for balancing of the load in a network by storage of energy
- H02J3/32—Arrangements for balancing of the load in a network by storage of energy using batteries with converting means
- H02J3/322—Arrangements for balancing of the load in a network by storage of energy using batteries with converting means the battery being on-board an electric or hybrid vehicle, e.g. vehicle to grid arrangements [V2G], power aggregation, use of the battery for network load balancing, coordinated or cooperative battery charging
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J3/00—Circuit arrangements for ac mains or ac distribution networks
- H02J3/003—Load forecast, e.g. methods or systems for forecasting future load demand
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J2203/00—Indexing scheme relating to details of circuit arrangements for AC mains or AC distribution networks
- H02J2203/20—Simulating, e g planning, reliability check, modelling or computer assisted design [CAD]
Landscapes
- Engineering & Computer Science (AREA)
- Power Engineering (AREA)
- Charge And Discharge Circuits For Batteries Or The Like (AREA)
- Remote Monitoring And Control Of Power-Distribution Networks (AREA)
- Electric Propulsion And Braking For Vehicles (AREA)
Abstract
The invention discloses a multi-node electric vehicle charging load joint countermeasure generation interval prediction method, which comprises the steps of analyzing the time-space correlation among multi-node charging loads in a combined charging scene of a day to be predicted and a historical day, and determining an original multi-node multi-correlation day combined charging scene set for describing the charging behavior of a multi-node electric vehicle; generating strong randomness of space-time distribution of the confrontation network depicting charging load by utilizing the gradient punishment Wasserstein, and generating a large number of combined charging scenes which have similar probability distribution with the original scene set but have difference in time sequence distribution in confrontation; screening out a combined scene set strongly related to the days to be predicted by adopting a weighted 2-D correlation coefficient according to the generated multi-node multi-correlation day combined charging scene set; obtaining a multi-node charging load interval prediction result according to the combination scene set related to the intensity of day to be predicted; the method can effectively predict the charging load time-space distribution of the electric automobile in the distribution network space, and is more favorable for improving the stability and the economical efficiency of the operation of the distribution network.
Description
Technical Field
The invention belongs to the technical field of electric vehicle charging load time-space distribution prediction, and particularly relates to a multi-node electric vehicle charging load joint countermeasure generation interval prediction method.
Background
Under the high-proportion Electric Vehicle (EV) permeability scene, a large number of charging loads with strong time-space uncertainty are connected to a power grid, which may cause the problems of low node voltage, line blockage and the like to the power grid, and bring great challenges to the safe and stable operation of a power distribution network. Therefore, after the EV is connected into the power grid, the charging load of each node in the distribution network space needs to be accurately predicted, and the time-space distribution of the charging load is described. Therefore, an important reference basis is provided for arranging a reasonable scheduling plan, and the negative influence of the charging load on the operation of the power distribution network is reduced.
At present, the prediction method for EV charging load time-space distribution mainly includes two methods: firstly, acquiring time-space distribution of charging load (namely a physical model driving method) by establishing an EV charging load physical model; the other method is to adopt a historical charging load data-driven artificial intelligence algorithm to predict the charging load (namely a data-driven method). The existing EV charging load time-space distribution prediction is driven by a physical model. In part of researches, charging load is calculated by adopting a Monte Carlo method after EV travel time and daily travel mileage data are analyzed. In addition, studies are carried out to predict the movement randomness of different types of EVs in different areas and time periods, the driving willingness of the vehicle owners and the influence of the electricity price on the space-time distribution of the charging load, but the positions and the charging cycles of the EVs are fixed, and the driving process of the EVs is not considered. In response to this problem, the influence of the traffic network and the distribution network on the time-space distribution of the charging load can be taken into account. But this process does not take into account the effects of traffic conditions on the EV travel path. In contrast, in some researches, traffic network modeling and trip chain theory are adopted to simulate the dynamic driving process of the electric automobile. A large number of research results are obtained in the research, however, more variables are involved in modeling, and more assumed conditions need to be set subjectively, so that the objectivity of EV charging load time-space distribution in the model is poor.
Compared with a physical model, the charging load prediction method based on data driving has the advantages that: the historical charging load data can be comprehensively utilized, and a large amount of model parameters are not required to be set. A prediction model is established by utilizing historical traffic data and weather data, and the charging requirement of the electric automobile is predicted. And a probability prediction method of the charging load of the electric vehicle aiming at different geographical areas can be provided based on the actual load of the electric vehicle. In addition, the deep learning method based on data driving achieves better effects in the field of EV charging load prediction. However, the inter-multinode EV charging load spatial correlation within the distribution network space was not considered in these studies. Moreover, the existing research shows that the EV charging load certainty prediction result is difficult to effectively reflect the risk of the strong time-space uncertainty of the charging load on the power distribution network; compared with a deterministic prediction result, the charging load interval prediction result can more effectively depict the strong randomness of the charging load.
Disclosure of Invention
The invention aims to provide a multi-node electric vehicle charging load joint countermeasure generation interval prediction method, and solves the problems that in the prior art, a space-space distribution prediction model of electric vehicle charging loads is poor in objectivity, and the spatial correlation of the multi-node electric vehicle charging loads in a distribution network space is not considered.
The technical scheme adopted by the invention is that the multi-node electric automobile charging load joint countermeasure generation interval prediction method is implemented according to the following steps:
and 4, obtaining a multi-node charging load interval prediction result and a certainty prediction result according to the last day data of the day-related combined scene set to be predicted.
The invention is also characterized in that:
the specific process of the step 1 is as follows: mapping the historical charging load data of the electric automobile into an IEEE33 node power distribution network system, numbering space nodes of a charging scene in the IEEE33 node power distribution network system by 1, … and 32 to obtain the historical charging load data of the electric automobile corresponding to each node, and defining a multi-node combined charging scene to be predicted on day to be expressed as a matrix DntThe multi-node combined charging scene of the historical days is expressed as a matrix (D-i)ntD is calculated according to all historical data of the charging load of the electric automobilentAnd (D-i)ntTime-space correlation between charging loads in two matrices
The calculation formula is as follows:
in equation (1), n denotes a spatial node number in the combined charging scenario, t denotes a sampling time point of the spatial charging load in the combined charging scenario, and n is 1 and 2, respectively…,32 and t ═ 1,2, …, 24; and is
and obtaining a multi-node combined charging scene of the extremely strong correlation day of the day to be predicted according to correlation analysis, and constructing an original multi-node multi-correlation day combined charging scene set by arranging the multi-node combined charging scene of the extremely strong correlation day and the multi-node combined charging scene of the day to be predicted according to a time sequence.
The specific process of the step 2 is as follows:
step 2.1, constructing a gradient penalty Wasserstein generation countermeasure network based on an original multi-node multi-correlation day combined charging scene set, optimizing a generator and a discriminator in the countermeasure network, and taking the generated network after optimization as a multi-node multi-correlation day combined charging scene generation model;
and 2.2, inputting the concentrated data of the original multi-node multi-correlation-day combined charging scene into a multi-node multi-correlation-day combined charging scene generation model, generating a same-dimension multi-node multi-correlation-day combined charging scene with the mass similar to the data of the original combined charging scene in probability distribution but different in time sequence distribution, and forming a multi-node multi-correlation-day combined charging scene set by the generated mass multi-node multi-correlation-day combined charging scenes.
Step 2.1 the concrete process of optimizing the generator and the discriminator in the countermeasure network is as follows:
using the Wasserstein distance instead of the JS divergence to describe the difference between the generated data and the true data distribution, applying the Wasserstein distance to the generated countermeasure network, expressed as:
wherein the content of the first and second substances,
as desired;
to generate a sample;
representing the result obtained by the discriminator; z is the noise vector input by the generator, and the probability distribution is Z-PZ(z); x is the characteristic vector of the original multi-node multi-correlation day combined charging scene concentrated sample, and X-PX(x)；
Adding a gradient penalty item in a discriminator loss function, wherein an objective function of a multi-node multi-correlation day combined charging scene generation model is as follows:
in the formula, lambda is a gradient penalty coefficient,
and charging load historical data for the electric automobile and generating a multi-node multi-correlation day combined charging scene data probability distribution-based linear sampling value.
The specific process of the step 3 is as follows:
calculating a historical day multi-node combined charging scene extremely strongly related to a multi-node combined charging scene on a day to be measured and generating a weighted 2-D correlation coefficient R of the jth scene in a multi-node multi-related day combined charging scene setjThe expression is:
in the formula:
2-D correlation coefficients of a history day multi-node combined charging scene D-i which is extremely strongly related to a multi-node combined charging scene on a day to be measured and a jth scene in a multi-node multi-related day combined charging scene set are represented;
and sequentially selecting the first M combined charging scenes from high to low to obtain the correlation coefficient to form a day-to-day associated combined scene set to be predicted.
The specific process of the step 4 is as follows: according to the combined scene of the day to be predicted, the charging load of each node in the last day scene is concentrated as the charging load of each node in the day to be predicted
Where n represents the node number and n ∈ [1,32 ]]，
And (3) calculating the charging load interval prediction result and the certainty prediction result of each node by adopting an equation (6):
wherein
Respectively representing the upper limit and the lower limit of the prediction result of the node n charging load interval at the time t;
and (4) representing the electric vehicle charging load certainty prediction result of the node n at the time t.
The invention has the beneficial effects that:
according to the multi-node electric vehicle charging load joint countermeasure generation interval prediction method, interval prediction of spatial correlation among multi-node electric vehicle charging loads is considered, each prediction index is better, the time-space distribution of the electric vehicle charging loads in the distribution network space can be predicted more effectively, and the stability and the economical efficiency of the operation of a distribution network are improved more favorably.
Drawings
FIG. 1 is a graph illustrating the correlation between the charging load of a multi-node combined charging scenario for a day to be predicted and multiple calendar days in accordance with an embodiment of the present invention;
FIG. 2 is a block diagram of a multi-node multi-dependent day joint charging scenario in accordance with an embodiment of the present invention;
FIG. 3 is a flow chart of a multi-node multi-dependent day joint charging scenario generation in an implementation of the present invention;
FIG. 4 is a graph illustrating probability distribution characteristics of data generated in a multi-node multi-dependent-day joint charging scenario set and data generated in an original multi-node multi-dependent-day joint charging scenario set in accordance with an embodiment of the present invention;
FIG. 5 is a graph illustrating statistical characteristic analysis of a multi-node multi-correlation-day joint charging scenario sample in accordance with an embodiment of the present invention;
FIG. 6 is a graph illustrating the spatial correlation between the charging loads of electric vehicles in a combined charging scenario in accordance with an embodiment of the present invention;
FIG. 7 is a timing sequence distribution analysis diagram of the charging loads of each node in the original multi-node multi-correlation-day combined charging scenario set and multi-node multi-correlation-day combined charging scenario set in the practice of the present invention;
FIG. 8 is a flow chart illustrating the multi-node electric vehicle charging load prediction in accordance with an embodiment of the present invention;
FIG. 9 is a statistical result chart of evaluation indexes of each prediction method in the implementation of the multi-node electric vehicle charging load joint countermeasure generation interval prediction method of the present invention;
FIG. 10 is a diagram of the prediction results of the EV charging load interval in each season in the implementation of the multi-node electric vehicle charging load joint countermeasure generation interval prediction method;
fig. 11 is an evaluation index chart of the prediction result of the charging load interval in each season in the implementation of the present invention.
Detailed Description
The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.
The invention discloses a multi-node electric vehicle charging load joint countermeasure generation interval prediction method, which is implemented according to the following steps:
and a high-permeability strong-randomness electric automobile is connected into the distribution network space. After the charging load is reduced to different distribution network nodes, the charging load distribution conditions are different, and the voltage level of each node of the distribution network is affected differently. Therefore, when the electric automobile dispatching plan is reasonably arranged by the power grid, the influence of the space distribution condition of the charging load of the electric automobile on the operation of the distribution network needs to be considered. In order to meet the requirement of an electric vehicle dispatching plan arranged in a power grid, the charging load of 32 charging stations (including 229 charging piles) in a certain area is mapped to each node of an IEEE33 node power distribution network system, and multi-node charging load interval prediction is carried out by taking the charging load of each node as a prediction target. The electric vehicle charging load data required by research is acquired from actual measurement data in 2019 in a certain area. The data is the data record of the charging load of the electric automobile in each day in the area with the sampling interval of 1 hour.
The charging place selection of the electric automobile user has certain inertia, and the charging behaviors of the user in different dates have relevance. But the vehicle using behaviors and the charging frequency of the electric vehicle users in the geographical area corresponding to each node in the distribution network are different. Therefore, it is necessary to select a historical day having a strong correlation with the day to be predicted, and construct a multi-node multi-correlation day combined charging scenario. The scene set is used for depicting the influence of multiple historical days on the charging load of the multi-node combined charging scene on the day to be predicted. In order to construct a multi-node multi-correlation-day combined charging scene, 2-D correlation coefficients are adopted, and the inter-day similarity of a day to be predicted and a historical day multi-node combined charging scene and the inter-multi-node charging load similarity in the scene are comprehensively analyzed from two dimensions of time and space, so that the limitation of a traditional correlation analysis method on single-dimension analysis of one-dimensional data is overcome.
The charging scene space nodes in the IEEE33 node power distribution network system are numbered 1, … and 32, and a matrix formed by multi-node combined charging scenes on days to be predicted is defined to be represented as DntThe matrix formed by the multi-node combined charging scene in the historical day is defined as (D-i)ntD is calculated according to all historical data of the charging load of the electric automobilentAnd (D-i)ntTime-space correlation between charging loads in two matrices
The calculation formula is as follows:
in equation (1), n denotes a spatial node number in the joint charging scenario, t denotes a spatial charging load sampling time point in the joint charging scenario, and ranges of n ═ 1,2, …,32, and t ═ 1,2, …,24, respectively; and is
The method comprises the steps of obtaining a history day multi-node combined charging scene with extremely strong correlation with a day to be predicted according to correlation analysis, taking the history day multi-node combined charging scene with extremely strong correlation with the day multi-node combined charging scene as the extremely strong correlation day multi-node combined charging scene, and constructing an original multi-node multi-correlation day combined charging scene set by arranging the extremely strong correlation day multi-node combined charging scene and the day multi-node combined charging scene according to a time sequence.
According to all historical data of the charging loads of the electric automobile, the time-space correlation between the multi-node combined charging scene of the day to be predicted and the multi-node charging loads of the ten-day previous combined charging scene is calculated by using the formula (1). The method comprises the steps of constructing a multi-node multi-correlation day combined charging scene according to time sequence arrangement on the basis of extremely strong correlation day and multi-node combined charging scenes of days to be predicted, constructing a first combined charging scene from 1, 9 and 2019 on the basis of actually measured charging load data of the electric vehicle, wherein each multi-node multi-correlation day combined charging scene comprises 32 nodes and 6 days of electric vehicle charging load data, and dividing an original multi-node multi-correlation day combined charging scene set into a training set and a testing set according to a ratio of 4:1 when multi-node charging load interval prediction is carried out.
In order to develop the time-space distribution prediction research of the charging load in the distribution network space, a multi-node multi-correlation day combined charging scene generation model is constructed through an original multi-node multi-correlation day combined charging scene set, and a multi-node multi-correlation day combined charging scene set is obtained and generated through the multi-node multi-correlation day combined charging scene generation model.
Step 2.1, generating model of multi-node multi-correlation day combined charging scene
The method comprises the steps that a gradient punishment Wassertein generation confrontation network is constructed based on an original multi-node multi-correlation day joint charging scene set, a generator and a discriminator need to be determined in the process of generating the confrontation network by the gradient punishment Wassertein, the generator excavates potential time-space distribution of electric vehicle charging loads in a distribution network space by learning multi-node multi-correlation day joint charging scene sample data probability distribution, and the discriminator is responsible for monitoring the multi-node multi-correlation day joint charging scene sample data quality and ensuring that the generated joint charging scene data and historical joint charging scene data have similar probability distribution.
When training on a countermeasure network generated by penalizing Wasserstein for gradient, alternately optimizing mutual games by a discriminator and a generator, completing model training when a Nash equilibrium point is reached, adopting Wasserstein distance to replace JS divergence to describe the difference between generated data and real data distribution, and applying the Wasserstein distance to the generation of the countermeasure network, wherein the expression is as follows:
wherein the content of the first and second substances,
as desired;
to generate a sample;
representing the result obtained by the discriminator; z is the noise vector input by the generator, and the probability distribution is Z-PZ(z); x is the characteristic vector of the original multi-node multi-correlation day combined charging scene concentrated sample, and X-PX(x)；
In order to solve the problems of difficult WGAN model training and mode collapse, a gradient penalty item is added in a discriminator loss function, the Lipschitz limit is optimized, the gradient penalty is ensured to be within a limited threshold, and the objective function of the multi-node multi-correlation day combined charging scene generation model is as follows:
in the formula, lambda is a gradient penalty coefficient,
and charging load historical data for the electric automobile and generating a multi-node multi-correlation day combined charging scene data probability distribution-based linear sampling value.
Step 2.2, generating a multi-node multi-correlation day combined charging scene set
Inputting the data in the original multi-node multi-correlation-day combined charging scene set into a multi-node multi-correlation-day combined charging scene generation model, generating a same-dimension multi-node multi-correlation-day combined charging scene with the mass similar to the probability distribution of the original combined charging scene data but with the time sequence distribution having difference, and forming the generated mass multi-node multi-correlation-day combined charging scene set by the generated mass multi-node multi-correlation-day combined charging scene.
In order to ensure the effectiveness of the multi-node multi-correlation-day combined charging scene generation, the quality of the generated multi-node multi-correlation-day combined charging scene is evaluated.
1) Generating joint charging scenario data probability distribution characterization
In order to evaluate the data generation capacity of the combined charging scene generation model, on the premise that the coupling relation among the charging loads of the electric automobiles of all nodes in the distribution network space is not considered, probability distribution characteristic analysis is respectively carried out on all data of all nodes in the multi-node multi-correlation-day combined charging scene and the multi-node multi-correlation-day combined charging scene set. And analyzing the data probability distribution characteristics of the original multi-node multi-correlation-day combined charging scene set and the multi-node multi-correlation-day combined charging scene set by applying a probability density function and an empirical cumulative distribution function. And further analyzing probability statistical characteristics of the multi-node multi-correlation day combined charging scene set sample and the original multi-node multi-correlation day combined charging scene set sample by adopting three statistics of the average value, the variance and the maximum value.
2) Inter-node charging load spatial correlation analysis in multi-node multi-correlation-day combined charging scene
The spatial correlation of the charging load among the multiple nodes in the multi-node multi-correlation-day combined charging scene needs to accord with the correlation rule of the historical combined charging scene. In order to evaluate the spatial correlation of the charging load among the multiple nodes in the multi-node multi-correlation-day combined charging scene, all the charging load data of each node in the original multi-node multi-correlation-day combined charging scene set and the generated multi-node multi-correlation-day combined charging scene set are reshaped into one line respectively (namely, 32 lines of charging load data are obtained from the original multi-node multi-correlation-day combined charging scene set and the generated multi-node multi-correlation-day combined charging scene set respectively). And after data are reshaped, respectively calculating the correlation among the charging load data of each row. Therefore, the original multi-node multi-correlation day joint charging scene set and the generated multi-node multi-correlation day joint charging scene set are obtained respectively, and the spatial correlation (namely the data correlation among matrix rows) of the charging loads among the multiple nodes is obtained. In order to further quantify the analysis result, the evaluation indexes of the structural similarity and the feature similarity are adopted. From the structural and characteristic aspects of spatial correlation among the multi-node charging loads in the combined charging scene, the similar spatial correlation exists in the original multi-node multi-correlation-day combined charging scene set and the generated multi-node multi-correlation-day combined charging scene set multi-node charging loads. When the structural similarity and the feature similarity evaluation index value are larger, the similarity degree indicating the spatial correlation is higher.
3) Generating joint charging scenario timing distribution profile analysis
In order to analyze and generate the time sequence distribution characteristics of the combined charging scene, a box line graph is adopted to respectively analyze the charging load data of each node of the electric automobile in the original multi-node multi-correlation day combined charging scene set and the multi-node multi-correlation day combined charging scene set generated by the countermeasure network based on the gradient penalty Wasserstein. And verifying that the charging load data of each node in the generated multi-node multi-correlation-day combined charging scene set accords with the time sequence distribution characteristic of the charging load data of each node in the original multi-node multi-correlation-day combined charging scene set.
According to the three characteristic analyses, the effectiveness of the generated multi-node multi-correlation-day combined charging scene can be obtained.
calculating a combined charging scene of extremely strong correlation days of the days to be predicted and generating a weighted 2-D correlation coefficient R of the jth scene in a multi-node multi-correlation day combined charging scene setjThe expression is:
in the formula:
2-D correlation coefficients representing the extremely strong correlated day combined charging scene of the day to be predicted, the generation of a multi-node multi-correlated day combined charging scene set D-i and the generation of the jth scene in the multi-node multi-correlated day combined charging scene set;
and sequentially selecting the first M combined charging scenes from high to low to obtain the correlation coefficient to form a day-to-day associated combined scene set to be predicted.
And (3) calculating the charging load interval prediction result and the certainty prediction result of each node by adopting an equation (6):
wherein
Respectively representing the upper limit and the lower limit of the prediction result of the node n charging load interval at the time t;
and (4) representing the electric vehicle charging load certainty prediction result of the node n at the time t.
Examples
The electric vehicle charging load data required by research is acquired from actual measurement data in 2019 in a certain area. The data is the data record of the charging load of the electric automobile in each day in the area with the sampling interval of 1 hour.
(1) Multi-node multi-correlation-day joint charging scene set construction
As shown in fig. 1, a schematic diagram of the correlation analysis between the charging loads of the multi-node combined charging scenario for the day to be predicted and the multiple history days, as can be seen from fig. 1, there are five extremely strong correlated history day charging scenarios for the day to be predicted, which are 1 day before, 2 days before, 6 days before, 7 days before, and 8 days before the day to be predicted. The multi-node combined charging scene with 5 extremely strong correlation days and days to be predicted is arranged according to a time sequence to construct a multi-node multi-correlation day combined charging scene, and the scene structure is shown in fig. 2. Based on the actual measurement data of the charging load of the electric vehicle, a first combined charging scene is constructed from 1 month and 9 days in 2019. Each multi-node multi-correlation day combined charging scene comprises 32 nodes and 6-day electric vehicle charging load data. Then 357 joint charging scenarios are contained in the original multi-node multi-dependent day joint charging scenario set. When multi-node charging load interval prediction is carried out, original multi-node multi-correlation day combined charging scene concentrated scene data are divided into a training set and a testing set according to the ratio of 4: 1.
(2) Multi-node multi-correlation day joint charging scene generation
Fig. 3 is a multi-node multi-correlation day combined charging scene generation process, which is to obtain a same-dimension multi-node multi-correlation day combined charging scene with similar probability distribution but different time sequence distribution of massive data and original combined charging scene data. When the scale of the multi-node multi-correlation day combined charging scene set is set to 5000 groups through experiments, the effect of covering the historical combined charging scene is optimal.
As shown in fig. 4, which is an analysis result of probability distribution characteristics of all data of all nodes, it can be seen from the analysis result that, compared with the Wasserstein generation countermeasure network, the gradient penalty Wasserstein generation countermeasure network generates a highly fitted curve of PDF and ECDF curves of the joint charging scene set generated by the Wasserstein generation countermeasure network and a corresponding curve of the historical joint charging scene set. It is explained that, as a whole, the generated combined charging scenario data has similar probability distribution characteristics to the historical combined charging scenario data. After the data probability distribution characteristic is evaluated, the probability statistical characteristics of the original multi-node multi-correlation day combined charging scene set sample and the generated multi-node multi-correlation day combined charging scene set sample are further analyzed by adopting three statistics of the average value, the variance and the maximum value shown in table 1. x represents a multi-node multi-correlation day combined charging scene sample, and the average value represents the charging load distribution characteristic of the combined charging scene; the variance represents the discrete degree of the charging load of the combined charging scene sample; the maximum value represents the maximum charging load of the electric automobile in the combined charging scene sample.
TABLE 1
As shown in fig. 5, three statistics analysis results are combined for the charging scenario sample. The analysis results shown in the figure show that compared with the Wasserstein generated countermeasure network, the multi-node multi-correlation day combined charging scene set samples generated by the Wasserstein generated countermeasure network are more closely distributed with the historical combined charging scene samples, historical combined charging scene sample scatter points can be effectively covered, potential charging loads conforming to the charging behavior rules of electric vehicle users are contained, and the multi-node multi-correlation day combined charging scene set generated by the Wasserstein generated countermeasure network can more effectively reflect the charging load fluctuation rules of electric vehicles in a distribution network space and reflect the effectiveness of generating the multi-node multi-correlation day combined charging scene set samples.
As shown in fig. 6, an original multi-node multi-correlation-day combined charging scenario set and a multi-node multi-correlation-day combined charging scenario set multi-node charging load inter-space correlation analysis result are generated. In order to further quantify the visualization analysis result of fig. 6, the structural similarity and the feature similarity evaluation indexes shown in table 2 are adopted to verify that similar spatial correlation exists among the multi-node charging loads of the original multi-node multi-correlation-day combined charging scene set and the generated multi-node multi-correlation-day combined charging scene set. As can be seen from the analysis results shown in fig. 6 and table 2, compared with the Wasserstein generated countermeasure network model, the structural similarity and the characteristic similarity index values of the overall data of the multi-node multi-correlation day joint charging scene set generated by the Wasserstein generated countermeasure network are respectively 0.94 and 0.97 based on the gradient penalty, and the spatial correlation among the multi-node charging loads of the historical joint charging scene is effectively described. And proving the effectiveness of the joint union scene generated by the countermeasure network generated by the gradient penalty Wasserstein.
TABLE 2
Fig. 7 shows the time sequence distribution analysis results of the charging loads of the nodes in the original multi-node multi-correlation-day combined charging scenario set and the generated multi-node multi-correlation-day combined charging scenario set. As can be seen from fig. 7, the charging load variation rule of each node in the generated combined charging scene conforms to the charging load time sequence distribution characteristic of the historical combined charging scene. Meanwhile, compared with a historical combined charging scene, the upper edge of the charging load data distribution of each node in the generated combined charging scene is higher, the lower edge of the charging load data distribution is lower, and the data outliers are more. The multi-node multi-correlation-day combined charging scene set can cover all historical charging scenes of all nodes and contains potential charging loads according with the charging behavior rules of electric vehicle users.
(3) Multi-node electric vehicle charging load interval prediction
FIG. 8 is a multi-node electric vehicle charging load interval prediction process based on a multi-node multi-correlation day combined charging scenario set.
Fig. 9 shows the evaluation index statistical results of the prediction methods. The input feature set of the comparison experiment GPR model comprises 97-dimensional features of all the charging loads of the nodes to be predicted at 8 th day, 7 th day, 6 th day and 2 nd day before the day and the charging loads of the nodes at the t moment of 1 day before the day, the proportion of the training set and the test set is consistent with the method, the method runs in the MatlabR2018b environment, the confidence coefficient is set to be 95%, and the kernel function is an ardexponential function. The minimum values of PICP indexes of prediction results of charging load intervals of all nodes obtained by the GPR method and the method are 77.9% and 90.4% respectively, and the minimum values of the average values are 80.9% and 92.7% respectively. The maximum values of PINAW indexes were 36.7% and 32.1%, respectively, and the maximum values of the average values were 34.2% and 29.2%, respectively. Through analysis, the PICP index value of the prediction result of the method is larger. Therefore, the method can ensure that the charging load prediction interval of the electric automobile at each node is more reliable and has higher precision. The maximum values of the MAPE indexes of the deterministic prediction results of the two methods are 22.7 percent of GPR and 17.7 percent of the method, and the maximum values of the mean values are 19.7 percent of GPR and 15.8 percent of the method. Through comparison, the method disclosed by the invention has the advantages that the deterministic prediction result MAPE index value is smaller, and the accuracy of the prediction result is higher.
Fig. 10 shows the prediction results, and fig. 11 shows the evaluation indices. In order to conveniently show the prediction effect of each method interval, the prediction results of the node 3, the node 26 and the node 31 with relatively poor prediction effect are selected for display, and it can be known through analyzing the prediction results of different nodes in different date types in fig. 10 and fig. 11 that when the charging load of the electric automobile in the distribution network space changes dramatically, the capacity of the GPR method for predicting the interval to track the charging load change is limited, so that the interval prediction effect is poor. Compared with a GPR method for respectively predicting the charging load of each node of the electric vehicle, the method for predicting the inter-node charging load space correlation has the advantages that the PICP value of the prediction interval is higher, and the reliability of the prediction interval is higher; the PINAW value is lower, and the prediction interval can be stronger close to the actual charging load; meanwhile, the electric vehicle charging load certainty prediction result MAPE value of each node is smaller, and the prediction precision is higher; therefore, the effectiveness of the charging load interval prediction method considering the correlation among the charging loads of the electric vehicles at all the nodes is proved.
By the mode, the invention discloses a multi-node EV charging load joint countermeasure generation interval prediction method, and the interval prediction considering the spatial correlation among the multi-node EV charging loads has better prediction indexes, can effectively predict the EV charging loads in the distribution network space in terms of time-space distribution, and is more favorable for improving the stability and the economical efficiency of the operation of the distribution network.
Claims (6)
1. The multi-node electric vehicle charging load joint countermeasure generation interval prediction method is characterized by being implemented according to the following steps:
step 1, mapping historical charging load data of an electric vehicle into an IEEE33 node power distribution network system, and constructing an original multi-node multi-correlation day combined charging scene set based on the historical charging load data of the electric vehicle;
step 2, constructing a multi-node multi-correlation day combined charging scene generation model through the original multi-node multi-correlation day combined charging scene set, and obtaining a multi-node multi-correlation day combined charging scene set through the multi-node multi-correlation day combined charging scene generation model;
step 3, analyzing and generating the correlation between the multi-node multi-correlation-day combined charging scene and the extremely strong correlation historical-day charging scene used for prediction, and selecting the high correlation degree as a day-to-be-predicted correlation combined scene set;
and 4, obtaining a multi-node charging load interval prediction result and a certainty prediction result according to the last day data of the day-related combined scene set to be predicted.
2. The multi-node electric vehicle charging load joint countermeasure generation interval prediction method according to claim 1, wherein the specific process of the step 1 is as follows: mapping the historical charging load data of the electric automobile into an IEEE33 node power distribution network system, numbering space nodes of a charging scene in the IEEE33 node power distribution network system by 1, … and 32 to obtain the historical charging load data of the electric automobile corresponding to each node, and defining a multi-node combined charging scene to be predicted on day to be expressed as a matrix DntThe multi-node combined charging scene of the historical day is expressed as a matrix (D-i)ntD is calculated according to all historical data of the charging load of the electric automobilentAnd (D-i)ntTime-space correlation between charging loads in two matrices
The calculation formula is as follows:
in equation (1), n denotes a spatial node number in the joint charging scenario, t denotes a spatial charging load sampling time point in the joint charging scenario, and ranges of n ═ 1,2, …,32, and t ═ 1,2, …,24, respectively; and is
and obtaining a multi-node combined charging scene of the extremely strong correlation day of the day to be predicted according to correlation analysis, and constructing an original multi-node multi-correlation day combined charging scene set by arranging the multi-node combined charging scene of the extremely strong correlation day and the multi-node combined charging scene of the day to be predicted according to a time sequence.
3. The multi-node electric vehicle charging load joint countermeasure generation interval prediction method according to claim 1, wherein the step 2 specifically comprises the following steps:
step 2.1, constructing a gradient penalty Wasserstein generation countermeasure network based on an original multi-node multi-correlation day combined charging scene set, optimizing a generator and a discriminator in the countermeasure network, and taking the generated network after optimization as a multi-node multi-correlation day combined charging scene generation model;
and 2.2, inputting the concentrated data of the original multi-node multi-correlation-day combined charging scene into a multi-node multi-correlation-day combined charging scene generation model, generating a same-dimension multi-node multi-correlation-day combined charging scene with the mass similar to the data of the original combined charging scene in probability distribution but different in time sequence distribution, and forming a multi-node multi-correlation-day combined charging scene set by the generated mass multi-node multi-correlation-day combined charging scenes.
4. The multi-node electric vehicle charging load joint countermeasure generation interval prediction method according to claim 3, wherein the specific process of optimizing the generator and the discriminator in the countermeasure network in step 2.1 is as follows:
using the Wasserstein distance instead of the JS divergence to describe the difference between the generated data and the true data distribution, applying the Wasserstein distance to the generation of the countermeasure network, expressed as:
wherein the content of the first and second substances,
as desired;
to generate a sample;
representing the result obtained by the discriminator; z is the noise vector input by the generator, and the probability distribution is Z-PZ(z); x is the characteristic vector of the original multi-node multi-correlation day combined charging scene concentrated sample, and X-PX(x)；
Adding a gradient penalty item in a discriminator loss function, wherein an objective function of a multi-node multi-correlation day combined charging scene generation model is as follows:
in the formula, lambda is a gradient penalty coefficient,
and charging load historical data for the electric automobile and generating a multi-node multi-correlation day combined charging scene data probability distribution-based linear sampling value.
5. The multi-node electric vehicle charging load joint countermeasure generation interval prediction method according to claim 1, wherein the specific process of step 3 is as follows:
calculating a historical day multi-node combined charging scene extremely strongly related to a multi-node combined charging scene on a day to be measured and generating a weighted 2-D correlation coefficient R of the jth scene in a multi-node multi-related day combined charging scene setjThe expression is:
in the formula:
2-D correlation coefficients of a history day multi-node combined charging scene D-i which is extremely strongly related to a multi-node combined charging scene on a day to be measured and a jth scene in a multi-node multi-related day combined charging scene set are represented;
and sequentially selecting the first M combined charging scenes from high to low to obtain the correlation coefficient to form a day-to-day associated combined scene set to be predicted.
6. The multi-node electric vehicle charging load joint countermeasure generation interval prediction method according to claim 1, characterized in that the specific process of step 4 is as follows: according to the charging load of each node in the last day scene in the day-related combined scene set to be predicted, the charging load of each node in the day to be predicted is
Where n represents the node number and n ∈ [1,32 ]]，
And (3) calculating the charging load interval prediction result and the certainty prediction result of each node by adopting an equation (6):
wherein
Respectively representing the upper limit and the lower limit of the prediction result of the node n charging load interval at the time t;
and (4) representing the electric vehicle charging load certainty prediction result of the node n at the time t.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210079212.8A CN114465256B (en) | 2022-01-24 | 2022-01-24 | Multi-node electric vehicle charging load combined countermeasure generation interval prediction method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210079212.8A CN114465256B (en) | 2022-01-24 | 2022-01-24 | Multi-node electric vehicle charging load combined countermeasure generation interval prediction method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114465256A true CN114465256A (en) | 2022-05-10 |
| CN114465256B CN114465256B (en) | 2024-06-21 |
Family
ID=81412406
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210079212.8A Active CN114465256B (en) | 2022-01-24 | 2022-01-24 | Multi-node electric vehicle charging load combined countermeasure generation interval prediction method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114465256B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109308541A (en) * | 2018-08-02 | 2019-02-05 | 华南理工大学 | A kind of electric car charging load forecasting method based on scene analysis |
| CN110956329A (en) * | 2019-12-02 | 2020-04-03 | 国网浙江省电力有限公司绍兴供电公司 | Load prediction method based on distributed photovoltaic and electric vehicle space-time distribution |
| CN111898822A (en) * | 2020-07-30 | 2020-11-06 | 东北电力大学 | Charging load interval prediction method based on multi-correlation-day scene generation |
| WO2021098352A1 (en) * | 2019-11-22 | 2021-05-27 | 国网福建省电力有限公司 | Active power distribution network planning model establishment method taking into consideration site selection and capacity determination of electric vehicle charging stations |
| CN113450023A (en) * | 2021-07-28 | 2021-09-28 | 东北电力大学 | Electric automobile ordered charging method under gridding time-of-use electricity price |
| KR20210148759A (en) * | 2020-06-01 | 2021-12-08 | 한국전력공사 | Charge/Discharge Scheduling Management System, Charge/Discharge Scheduling Server Apparatus and Charge/Discharge Scheduling Method for Electric Vehicle with SOC control |
-
2022
- 2022-01-24 CN CN202210079212.8A patent/CN114465256B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109308541A (en) * | 2018-08-02 | 2019-02-05 | 华南理工大学 | A kind of electric car charging load forecasting method based on scene analysis |
| WO2021098352A1 (en) * | 2019-11-22 | 2021-05-27 | 国网福建省电力有限公司 | Active power distribution network planning model establishment method taking into consideration site selection and capacity determination of electric vehicle charging stations |
| CN110956329A (en) * | 2019-12-02 | 2020-04-03 | 国网浙江省电力有限公司绍兴供电公司 | Load prediction method based on distributed photovoltaic and electric vehicle space-time distribution |
| KR20210148759A (en) * | 2020-06-01 | 2021-12-08 | 한국전력공사 | Charge/Discharge Scheduling Management System, Charge/Discharge Scheduling Server Apparatus and Charge/Discharge Scheduling Method for Electric Vehicle with SOC control |
| CN111898822A (en) * | 2020-07-30 | 2020-11-06 | 东北电力大学 | Charging load interval prediction method based on multi-correlation-day scene generation |
| CN113450023A (en) * | 2021-07-28 | 2021-09-28 | 东北电力大学 | Electric automobile ordered charging method under gridding time-of-use electricity price |
Non-Patent Citations (1)
| Title |
|---|
| ZHU J等: "Short-Term Load Forecasting for Electric Vehicle Charging Stations Based on Deep Learning Approaches", 《APPLIED SCIENCES》, vol. 9, no. 9, 31 December 2019 (2019-12-31), pages 1 - 7 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114465256B (en) | 2024-06-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107610021B (en) | Comprehensive analysis method for space-time distribution of environment variables | |
| CN106980877A (en) | A kind of Prediction of Blasting Vibration method based on particle cluster algorithm Support Vector Machines Optimized | |
| CN104091216A (en) | Traffic information predication method based on fruit fly optimization least-squares support vector machine | |
| CN115270965A (en) | Power distribution network line fault prediction method and device | |
| CN110909924A (en) | City expansion multi-scenario simulation cellular automata method based on cross entropy optimizer | |
| CN113222316A (en) | Change scene simulation method based on FLUS model and biodiversity model | |
| CN112419711B (en) | Closed parking lot parking demand prediction method based on improved GMDH algorithm | |
| CN102222313A (en) | Urban evolution simulation structure cell model processing method based on kernel principal component analysis (KPCA) | |
| CN115948964A (en) | Road flatness prediction method based on GA-BP neural network | |
| Chao et al. | Estimation of species richness and shared species richness | |
| CN108830405B (en) | Real-time power load prediction system and method based on multi-index dynamic matching | |
| CN117114176A (en) | Land utilization change prediction method and system based on data analysis and machine learning | |
| CN114465256B (en) | Multi-node electric vehicle charging load combined countermeasure generation interval prediction method | |
| CN116108963A (en) | Electric power carbon emission prediction method and equipment based on integrated learning module | |
| CN111898822B (en) | Charging load interval prediction method based on multi-correlation-day scene generation | |
| CN115906669A (en) | Dense residual error network landslide susceptibility evaluation method considering negative sample selection strategy | |
| CN105160065A (en) | Topological relation based remote sensing information similarity evaluation method | |
| Uribe‐Rivera et al. | The predictive performance of process‐explicit range change models remains largely untested | |
| CN114372615A (en) | Short-term photovoltaic power prediction method and system considering spatial correlation | |
| CN111882100B (en) | Hydrologic set interval forecast building method based on multi-model random linear combination | |
| CN116611588B (en) | Precipitation multi-driving factor segmentation rating optimization forecasting method and system | |
| Abdullahia et al. | Linear and non-linear ensemble concepts for pan evaporation modeling | |
| CN116306237A (en) | Method for obtaining land utilization change simulation result based on neighborhood weight matrix | |
| CN105574350A (en) | Event prediction method | |
| CN116882580A (en) | Public institution-oriented LSTM circulating neural network carbon emission prediction method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant |