CN113205280A - Electric vehicle charging station site selection method for Liqun guided attention inference network - Google Patents

Electric vehicle charging station site selection method for Liqun guided attention inference network Download PDF

Info

Publication number
CN113205280A
CN113205280A CN202110590249.2A CN202110590249A CN113205280A CN 113205280 A CN113205280 A CN 113205280A CN 202110590249 A CN202110590249 A CN 202110590249A CN 113205280 A CN113205280 A CN 113205280A
Authority
CN
China
Prior art keywords
attention
charging station
electric vehicle
vehicle charging
guiding
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
Application number
CN202110590249.2A
Other languages
Chinese (zh)
Other versions
CN113205280B (en
Inventor
殷林飞
高奇
马晨骁
高放
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Guangxi University
Original Assignee
Guangxi University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Guangxi University filed Critical Guangxi University
Priority to CN202110590249.2A priority Critical patent/CN113205280B/en
Publication of CN113205280A publication Critical patent/CN113205280A/en
Application granted granted Critical
Publication of CN113205280B publication Critical patent/CN113205280B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0637—Strategic management or analysis, e.g. setting a goal or target of an organisation; Planning actions based on goals; Analysis or evaluation of effectiveness of goals
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N5/00—Computing arrangements using knowledge-based models
    • G06N5/04—Inference or reasoning models
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce
    • G06Q30/02—Marketing; Price estimation or determination; Fundraising
    • G06Q30/0201—Market modelling; Market analysis; Collecting market data
    • G06Q30/0202—Market predictions or forecasting for commercial activities
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
    • Y02T10/00—Road transport of goods or passengers
    • Y02T10/60—Other road transportation technologies with climate change mitigation effect
    • Y02T10/70—Energy storage systems for electromobility, e.g. batteries

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Theoretical Computer Science (AREA)
  • Strategic Management (AREA)
  • Physics & Mathematics (AREA)
  • Human Resources & Organizations (AREA)
  • General Physics & Mathematics (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Development Economics (AREA)
  • Accounting & Taxation (AREA)
  • Software Systems (AREA)
  • Economics (AREA)
  • Finance (AREA)
  • Data Mining & Analysis (AREA)
  • Mathematical Physics (AREA)
  • General Business, Economics & Management (AREA)
  • Evolutionary Computation (AREA)
  • Educational Administration (AREA)
  • General Engineering & Computer Science (AREA)
  • Artificial Intelligence (AREA)
  • Game Theory and Decision Science (AREA)
  • Computing Systems (AREA)
  • Marketing (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Medical Informatics (AREA)
  • Computational Linguistics (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Tourism & Hospitality (AREA)
  • Traffic Control Systems (AREA)
  • Charge And Discharge Circuits For Batteries Or The Like (AREA)

Abstract

The invention provides an electric vehicle charging station site selection method of a lie group guiding attention reasoning network. The method combines a lie group machine learning method and a guiding attention reasoning network method and is used for electric vehicle charging station site selection. Firstly, the method of the guiding attention reasoning network is mainly used for feature extraction of traffic data and discrimination of electric vehicle flow information. Secondly, the lie group machine learning method in the method is used as a parameter of a threshold function of the guiding attention reasoning network method, and the training efficiency of the guiding attention reasoning network method is improved. And finally, gridding the flow information of the electric automobile to obtain the charging demand of the electric automobile, associating the traffic flow with grids through longitude and latitude grid division, and determining the optimal charging station address at a dense grid.

Description

Electric vehicle charging station site selection method for Liqun guided attention inference network
Technical Field
The invention belongs to the field of power systems, and relates to an electric vehicle charging station site selection method based on an artificial intelligence method, which is suitable for site selection of power systems and smart city charging stations.
Background
Under the international trend of fossil energy stock crisis and carbon neutralization, automobile manufacturers are beginning to vigorously develop the electric automobile industry, get rid of the dependence on petroleum, embrace clean energy and complete the mission of social carbon neutralization. A new charging station is built in the existing urban network, so that the inclusion degree of urban electric vehicles is improved, and the short plates applied to the electric vehicles are reduced.
Charging station site selection based on data driving is a technology based on artificial intelligence, and can overcome the defects that traditional mathematical modeling cannot be completed. The information acquisition based on the machine learning method in the artificial intelligence can comprehensively and quickly lock data information, effectively extract vehicle information, determine an optimal charging station address through grid information evaluation, and enhance the economy and the practicability of the charging station address selection.
The lie group machine learning combines the advantages of manifold learning and the idea of the lie group, thereby becoming a learning paradigm with innovative characteristics in the field of machine learning. The guiding attention reasoning network is an end-to-end picture identification mode, is a machine learning method, and belongs to a specification mode between supervision and unsupervised.
Therefore, the Liqun machine learning method is used as a threshold function parameter for guiding attention to reasoning network method, the end-to-end training efficiency can be improved, a large amount of actual monitoring images are used for identifying effective traffic flow information, traffic demand positions and grid codes are divided, and finally the charging station address of the electric automobile is selected.
Disclosure of Invention
The invention provides a method for selecting the location of an electric vehicle charging station of a lie group guided attention reasoning network, which can refine a training graph label, and adds an effective mechanism to control the network attention in the learning process of machine learning, wherein the learning process is end-to-end training, the link is clear, and self-guidance can be provided in the training process, so that the generalization capability of special target tasks is improved. The method combines a Liqun machine learning method and a guiding attention reasoning network method, is used for site selection of the electric vehicle charging station, and obtains candidate positions of the electric vehicle charging station under the condition of traffic data; the electric vehicle charging station site selection method of the Liqun guide attention inference network comprises the following steps in the use process:
step (1): the method is characterized in that a guiding attention reasoning network method is used for processing traffic data to achieve the purpose of an optimal training sample set;
for a given image I, in a constrained stream Sc,lIn, let fl,kActivating a unit k in the l-th layer, wherein for each class C from the real label, the gradient of the class C with respect to the activation map is scWeight of neuronal importance
Figure BDA0003089080800000011
The gradient for the reflux will pool by global averaging:
Figure BDA0003089080800000021
in the formula, GAP (. cndot.) represents the global average pooling;
to integrate all activation maps, a weight matrix w is usedcAs kernel and applying a two-dimensional convolution on the activation map matrix flPerforming a linear rectification function ReLU operation to obtain AcAttention is drawn:
Ac＝ReLU(conv(fl,wc))
where l is from the last convolutional layer, the features are balanced between detailed spatial information and high-level semantics;
use attention AcGenerating a soft mask for the original input image to obtain I*c：
I*c＝I-(T(Ac)⊙I)
In the formula,. indicates element-level multiplication, T (A)c) Is a masking function based on threshold operations; to make T (A)c) Can be derived, using sigmoid function as T (A)c) Function: i is a given image;
Figure BDA0003089080800000022
where ω is a scale parameter when T (A)c) If σ is greater than or equal to 0, T (A) is guaranteedc) Approximately equal to 1; note that the objective of the loss function for mining losses is to minimize I for class C*cThe prediction score of (a) is determined,
Figure BDA0003089080800000023
in the formula, sc(I*c) I representing class C*cThe predicted score of (a); n is the number of real labels of the image;
final self-guiding loss LselfIs a classification loss LclAnd LamSum of (a):
Lself＝Lcl+αLam
in the formula, LclIs multi-label soft profit loss; α is a weight parameter set to 1;
step (2): the lie group machine learning method is used for guiding an attention reasoning network method to shield function parameters;
the lie group machine learning method is used for guiding the attention inference network method to shield the function parameters; σ is the internal mean of the lie group machine learning sample set:
Figure BDA0003089080800000031
wherein x isijRepresents the jth sample, n, in the ith class in the training sample set of the lie groupiRepresenting the number of training samples in the ith classification, wherein the training samples have C classes;
and (3): applying a lie group guiding attention reasoning network method to electric vehicle charging station site selection;
converting the longitude and latitude of the departure point and the destination of the traffic demand into grid numbers, and indirectly counting the traffic demand in the grid by counting the occurrence times of the grid numbers; the mesh division method is that the analysis area is divided into meshes with equal size according to a certain interval, and D is used for representing the interval of mesh division; when D is 0.01 degrees, the analysis area is divided into a plurality of grids at equal intervals according to the longitude and latitude intervals of 0.01 degrees, and each grid is a rectangular grid of 700m multiplied by 1000 m;
definition (X)min,Ymin) To analyze the lower left corner longitude and latitude coordinates of a region, (X)min,Ymin) Longitude and latitude coordinates of the upper right corner; (X)i,Yi) Is the longitude and latitude coordinate of any traffic demand point position, then (X)i,Yi) Grid number (gridC)x,gridCy) Is composed of
gridCx＝(int)((Xi-Xmin)/W)
gridCy＝(int)((Yi-Ymin)/W)；
(int) is an integer part; and selecting the grids with large demand as candidate positions for electric vehicle charging station site selection.
Compared with the prior art, the invention has the following advantages and effects:
(1) the attention-directing reasoning network method is a machine learning method between supervised and unsupervised learning, and has good recognition effect in end-to-end training in training set; meanwhile, a shielding function is added, so that over-training is prevented, and the specific object of image recognition is more stable.
(2) In the shielding function provided by the invention, parameters are derived from a lie group machine learning method, an inner mean value of the lie group learning method is used as a threshold matrix, supervision and unsupervised direct influence is realized during balance training, and the generalization capability of the system is greatly improved.
(3) The lie group guiding attention reasoning network method provided by the invention acts on a ground traffic system, analyzes road surface information, obtains the charging demand of the electric automobile, associates traffic flow with grids through longitude and latitude grid division, and finally determines the optimal charging station address at a dense grid. The factors of inaccurate and incomplete information acquisition are avoided, and the information is directly acquired from the traffic image with higher accuracy.
Drawings
FIG. 1 is a lie group guided attention inference network of the method of the present invention.
Fig. 2 is a data-driven charging station-based addressing method of the present invention.
Detailed Description
The invention provides an electric vehicle charging station site selection method for a lie group guiding attention inference network, which is described in detail in combination with the accompanying drawings as follows:
FIG. 1 is a lie group guided attention inference network of the method of the present invention. Firstly, integrating traffic data pictures into a training set and a testing set, and training a baggage group to guide attention to an inference network. After input of the training picture, stream S is usedclAnd searching the picture, and extracting a region which is helpful for identifying the category. Obtaining A Using ReLU operationcAttention is drawn to simultaneously obtain a region I outside the category C of current interest to the network*c. Lie group guided attention inference network attempts to minimize class C I*cThe most critical masking function T (A) based on threshold operationc) σ in (d) is the inner mean of the lie group machine learning sample set.
Fig. 2 is a data-driven charging station-based addressing method of the present invention. The data-driven electric vehicle charging station site selection method is characterized in that potential traffic demand distribution positions are mined by analyzing the distribution rule of mass mobile position data, and data support is provided for site selection of electric vehicle charging stations. In order to improve the efficiency of data processing, the invention provides a grid traffic demand statistical method, wherein grid is also called unitization and is a form of system division and organization, and the aim is to reduce the complexity of a system by grid the system so as to improve the management level and the control effect. And extracting potential traffic demand positions including a starting point and a destination by using the lie group guided attention inference network by using massive mobile position data. And dividing the analysis area into grids with equal intervals, associating the traffic demand position with the grids, and counting the traffic demand in each grid. And selecting the grids with large demand as candidate positions for electric vehicle charging station site selection.

Claims (1)

1. A lie group guiding attention reasoning network electric vehicle charging station site selection method is characterized in that the method combines a lie group machine learning method and an attention reasoning network guiding method, is used for electric vehicle charging station site selection, and obtains electric vehicle charging station candidate positions under the condition of traffic data; the electric vehicle charging station site selection method of the Liqun guide attention inference network comprises the following steps in the use process:
step (1): the method is characterized in that a guiding attention reasoning network method is used for processing traffic data to achieve the purpose of an optimal training sample set;
for a given image I, in a constrained stream Sc,lIn, let fl,kActivating a unit k in the l-th layer, wherein for each class C from the real label, the gradient of the class C with respect to the activation map is scWeight of neuronal importance
Figure FDA0003089080790000011
The gradient for the reflux will pool by global averaging:
Figure FDA0003089080790000012
in the formula, GAP (. cndot.) represents the global average pooling;
to integrate all activation maps, a weight matrix w is usedcAs kernel and applying a two-dimensional convolution on the activation map matrix flPerforming a linear rectification function ReLU operation to obtain AcAttention is drawn:
Ac＝ReLU(conv(fl,wc))
where l is from the last convolutional layer, the features are balanced between detailed spatial information and high-level semantics;
use attention AcGenerating a soft mask for the original input image to obtain I*c：
Figure FDA0003089080790000013
In the formula (I), the compound is shown in the specification,
Figure FDA0003089080790000014
representing multiplication at the element level, T (A)c) Is a masking function based on threshold operations; to make T (A)c) Can be derived, using sigmoid function as T (A)c) Function: i is a given image;
Figure FDA0003089080790000015
where ω is a scale parameter when T (A)c) If σ is greater than or equal to 0, T (A) is guaranteedc) Approximately equal to 1; note that the objective of the loss function for mining losses is to minimize I for class C*cThe prediction score of (a) is determined,
Figure FDA0003089080790000016
in the formula, sc(I*c) I representing class C*cThe predicted score of (a); n is the number of real labels of the image;
final self-guiding loss LselfIs a classification loss LclAnd LamSum of (a):
Lself＝Lcl+αLam
in the formula, LclIs multi-label soft profit loss; α is a weight parameter set to 1;
step (2): the lie group machine learning method is used for guiding an attention reasoning network method to shield function parameters;
the lie group machine learning method is used for guiding the attention inference network method to shield the function parameters; σ is the internal mean of the lie group machine learning sample set:
Figure FDA0003089080790000021
i＝1,2,…,c；j＝1,2,…,ni
wherein x isijRepresents the jth sample, n, in the ith class in the training sample set of the lie groupiRepresenting the number of training samples in the ith classification, wherein the training samples have C classes;
and (3): applying a lie group guiding attention reasoning network method to electric vehicle charging station site selection;
converting the longitude and latitude of the departure point and the destination of the traffic demand into grid numbers, and indirectly counting the traffic demand in the grid by counting the occurrence times of the grid numbers; the mesh division method is that the analysis area is divided into meshes with equal size according to a certain interval, and D is used for representing the interval of mesh division; when D is 0.01 degrees, the analysis area is divided into a plurality of grids at equal intervals according to the longitude and latitude intervals of 0.01 degrees, and each grid is a rectangular grid of 700m multiplied by 1000 m;
definition (X)min,Ymin) To analyze the lower left corner longitude and latitude coordinates of a region, (X)min,Ymin) Longitude and latitude coordinates of the upper right corner; (X)i,Yi) Is the longitude and latitude coordinate of any traffic demand point position, then (X)i,Yi) Grid number (gridC)x,gridCy) Is composed of
gridCx＝(int)((Xi-Xmin)/W)
gridCy＝(int)((Yi-Ymin)/W)；
(int) is an integer part; and selecting the grids with large demand as candidate positions for electric vehicle charging station site selection.
CN202110590249.2A 2021-05-28 2021-05-28 Electric vehicle charging station address selection method of plum cluster guiding attention reasoning network Active CN113205280B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110590249.2A CN113205280B (en) 2021-05-28 2021-05-28 Electric vehicle charging station address selection method of plum cluster guiding attention reasoning network

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110590249.2A CN113205280B (en) 2021-05-28 2021-05-28 Electric vehicle charging station address selection method of plum cluster guiding attention reasoning network

Publications (2)

Publication Number Publication Date
CN113205280A true CN113205280A (en) 2021-08-03
CN113205280B CN113205280B (en) 2023-06-23

Family

ID=77023470

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110590249.2A Active CN113205280B (en) 2021-05-28 2021-05-28 Electric vehicle charging station address selection method of plum cluster guiding attention reasoning network

Country Status (1)

Country Link
CN (1) CN113205280B (en)

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101017572A (en) * 2006-02-09 2007-08-15 三菱电机株式会社 Computerized method for tracking object in sequence of frames
CN110188597A (en) * 2019-01-04 2019-08-30 北京大学 A kind of dense population counting and accurate positioning method and system based on attention mechanism circulation scaling
CN111126485A (en) * 2019-12-24 2020-05-08 武汉大学 Lie-KFDA scene classification method and system based on Lie group machine learning kernel function
CN111476199A (en) * 2020-04-26 2020-07-31 国网湖南省电力有限公司 Power transmission and transformation project common grave ground identification method based on high-definition aerial survey image
CN111680930A (en) * 2020-06-17 2020-09-18 云南省设计院集团有限公司 Electric vehicle charging station site selection evaluation method based on characteristic reachable circle
US10873533B1 (en) * 2019-09-04 2020-12-22 Cisco Technology, Inc. Traffic class-specific congestion signatures for improving traffic shaping and other network operations

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101017572A (en) * 2006-02-09 2007-08-15 三菱电机株式会社 Computerized method for tracking object in sequence of frames
CN110188597A (en) * 2019-01-04 2019-08-30 北京大学 A kind of dense population counting and accurate positioning method and system based on attention mechanism circulation scaling
US10873533B1 (en) * 2019-09-04 2020-12-22 Cisco Technology, Inc. Traffic class-specific congestion signatures for improving traffic shaping and other network operations
CN111126485A (en) * 2019-12-24 2020-05-08 武汉大学 Lie-KFDA scene classification method and system based on Lie group machine learning kernel function
CN111476199A (en) * 2020-04-26 2020-07-31 国网湖南省电力有限公司 Power transmission and transformation project common grave ground identification method based on high-definition aerial survey image
CN111680930A (en) * 2020-06-17 2020-09-18 云南省设计院集团有限公司 Electric vehicle charging station site selection evaluation method based on characteristic reachable circle

Also Published As

Publication number Publication date
CN113205280B (en) 2023-06-23

Similar Documents

Publication Publication Date Title
Sun et al. RSOD: Real-time small object detection algorithm in UAV-based traffic monitoring
Wen et al. A novel spatiotemporal convolutional long short-term neural network for air pollution prediction
CN110245709B (en) 3D point cloud data semantic segmentation method based on deep learning and self-attention
CN111028255B (en) Farmland area pre-screening method and device based on priori information and deep learning
Chen et al. Deep learning from multiple crowds: A case study of humanitarian mapping
Li et al. Graph CNNs for urban traffic passenger flows prediction
CN112733800B (en) Remote sensing image road information extraction method and device based on convolutional neural network
Jiang et al. Hierarchical semantic segmentation of urban scene point clouds via group proposal and graph attention network
Lunga et al. Resflow: A remote sensing imagery data-flow for improved model generalization
Xiao et al. Open-pit mine road extraction from high-resolution remote sensing images using RATT-UNet
Gao et al. Road extraction using a dual attention dilated-linknet based on satellite images and floating vehicle trajectory data
CN116796810A (en) Deep neural network model compression method and device based on knowledge distillation
Moraguez et al. Convolutional neural network for detection of residential photovoltalc systems in satellite imagery
CN110287970B (en) Weak supervision object positioning method based on CAM and covering
Mabu et al. Landslide classification from synthetic aperture radar images using convolutional neural network with multichannel information
Liang et al. Human-guided flood mapping: From experts to the crowd
Chen et al. DGCNN network architecture with densely connected point pairs in multiscale local regions for ALS point cloud classification
CN113205280A (en) Electric vehicle charging station site selection method for Liqun guided attention inference network
CN115880660A (en) Track line detection method and system based on structural characterization and global attention mechanism
Lin et al. Network-wide multi-step traffic volume prediction using graph convolutional gated recurrent neural network
Sharma et al. Change detection and feature extraction using high-resolution remote sensing images
Endri et al. Road damage detection utilizing convolution neural network and principal component analysis
Wang et al. Information Extraction of the Vehicle from High-Resolution Remote Sensing Image Based on Convolution Neural Network
Jiménez-López et al. Artificial neural networks in the application of the growth of the urban sprawl
Jiang et al. A method for identifying marine targets based on mining of multi-characteristic movement patterns

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
GR01 Patent grant