Communication networking method and system applicable to widely dispersed charging piles

Abstract

Disclosed in the present invention are a communication networking method and system applicable to widely dispersed charging piles. The method comprises: initializing the total number of widely dispersed charging piles and coordinate parameters thereof; initializing related parameters of a hybrid learning network and using the hybrid learning network to predict and determine optimal networking nodes of the charging piles; and according to the optimal networking nodes, connecting same to form communication networking of the widely dispersed charging piles. The present invention targets communication networking applicable to widely dispersed charging piles, whereby the channel environment between the widely dispersed charging piles is evaluated, and optimal main nodes are selected on the basis of the quality of the communication environment between the charging piles, thereby forming a communication networking scheme for the widely dispersed charging piles, thus overcoming the defect whereby communication reliability cannot be guaranteed when charging piles are far apart and widely dispersed, and laying the foundation for comprehensively realizing communication networking of electric vehicles in a later period.

Description

A communication networking method and system suitable for wide-area distributed charging piles Technical Field

[0001] The present invention relates to the field of new energy vehicles, and in particular to a communication networking method and system applicable to wide-area dispersed charging piles. Background Art

[0002] The growing number of electric vehicles (EVs) poses challenges to the development of new energy vehicle charging infrastructure. Implementing EV communication networking is a fundamental issue in this infrastructure. Current research is exploring the possibility of executing charging control commands at target charging piles. Simply placing a gateway in a location with a local area network (LAN) enables communication between the charging pile and the control platform. Alternatively, wireless communication can be used to network the remaining auxiliary nodes within a cluster system channel, excluding the primary and auxiliary nodes, to complete the network and perform maintenance. These approaches are particularly effective when EV charging piles are densely distributed.

[0003] However, when charging piles are far apart and widely dispersed, current technologies cannot solve the communication networking problem of charging piles. Therefore, in order to solve the above problem, it is of great significance to provide a communication networking method suitable for widely dispersed charging piles.

[0004] Summary of the Invention

[0005] In view of the above existing problems, the present invention is proposed.

[0006] Therefore, the present invention provides a communication networking method and system suitable for widely dispersed charging piles to solve the problem that the current charging piles cannot accurately and efficiently communicate and network when the charging piles are far away and widely dispersed.

[0007] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0008] In a first aspect, the present invention provides a communication networking method applicable to wide-area distributed charging piles, comprising:

[0009] Initialize the total number of wide-area discrete charging piles and their coordinate parameters;

[0010] Initialize the relevant parameters of the hybrid learning network and use the hybrid learning network to predict and determine the optimal networking nodes of the charging piles;

[0011] A communication network of wide-area dispersed charging piles is formed based on the optimal networking nodes.

[0012] As a preferred solution of the communication networking method applicable to wide-area dispersed charging piles described in the present invention, the total number of wide-area dispersed charging piles and their coordinate parameters are initialized, including:

[0013] The coordinates of the wide-area distributed charging pile are expressed as: (x i ,y i ), i = 1, 2, 3, ..., N, N is the total number of wide-area distributed charging piles.

[0014] As a preferred solution of the communication networking method for wide-area distributed charging piles described in the present invention, the initialization of relevant parameters of the hybrid learning network and the use of the hybrid learning network to determine the optimal networking node of the charging pile include:

[0015] Define the first master node (x1, y1) as the first optimal master node (x 1,op ,y 1,op ), that is, x1=x 1,op , y1=y 1,op ;

[0016] Perform a first judgment, the first judgment including:

[0017] Determine the distance r between other charging piles and the first optimal master node i1 , expressed as:

[0018] Where i = 2, 3, ..., N;

[0019] Sort the distances of N-1 charging piles from smallest to largest, and select the n charging piles with the smallest distances to establish the second optimal master node judgment data set Φ; where the data set Φ is the number of the n charging piles with the smallest distance from the first optimal master node;

[0020] The hybrid learning network is used to predict the channel state information of the first optimal master node and the n charging piles in the data set Φ, and the prediction error e is obtained. j , j=1,2,3,...,n;

[0021] The prediction errors are sorted and the charging pile with the smallest error is defined as the second best master node. The other charging piles in the data set Φ are defined as secondary nodes.

[0022] The corresponding charging pile label in Φ is taken from the total charging pile label dataset Φ 总 Remove it and get the total charging pile label dataset Φ' 总 ;

[0023] Perform a second judgment, the second judgment including:

[0024] Determine the total charging pile label data set Φ' 总 With the second best master node (x 2,op ,y 2,op ) distance ri2 , expressed as:

[0025] in, is the total dataset of charging pile labels Φ' 总 size;

[0026] right The distances of the charging piles are sorted, and the n charging piles with the smallest distance are intercepted to update the main node judgment data set Φ;

[0027] Use the hybrid learning network to predict the channel state information of the second master node and the n charging piles in the data set Φ, and obtain the prediction error e j , j=1,2,3,...,n;

[0028] Sort the prediction errors and define the charging pile with the smallest error as the third optimal master node, and the other charging piles in the data set Φ as secondary nodes; the corresponding charging pile labels in Φ are taken from the charging pile label data set Φ' 总 Remove and update the total charging pile label dataset Φ' after deletion 总 ;

[0029] Judge Φ' 总 Is it an empty set? If not, loop to the second judgment;

[0030] Output the number of the best master node in sequence.

[0031] As a preferred solution of the communication networking method for wide-area distributed charging piles described in the present invention, wherein: the hybrid learning network is used to predict the channel state information of the second master node and the n charging piles in the data set Φ to obtain the prediction error e j , j = 1, 2, 3, ..., n, including:

[0032] Use the channel to estimate the frequency domain channel state information between the optimal node and the jth charging pile Among them, K and N s are the total number of subcarriers and the number of samples of pilot subcarriers respectively;

[0033] Obtain the delay power spectrum between the optimal node and the jth charging pile through inverse discrete Fourier transform

[0034] For the lth multipath of the delay power spectrum Establish a hybrid learning network prediction model to predict the lth multipath value l=1,2,3,...,L T , L T is the total number of transmission multipaths between the optimal node and the jth charging pile;

[0035] will be predicted Perform Fourier transform to obtain the predicted frequency domain channel state information Frequency domain channel state information with theoretical Jointly calculate the root mean square error e j , expressed as:

[0036] As a preferred solution of the communication networking method for wide-area distributed charging piles described in the present invention, wherein: the first multipath of the delay power spectrum Build a hybrid learning network prediction model, including:

[0037] Initialize the number of neurons P in the first feature layer of the hybrid learning network and the number of neurons Q in the second feature layer. Then the output matrix G of the first feature layer can be expressed as G = tanh(XW in )

[0038] Among them, tanh is the hyperbolic tangent function, X is the lth multipath of the delay power spectrum in the training phase The input data is expressed as:

[0039] N T is the lth multipath delay power spectrum in the training phase The number of data used for training, N s >N T ;W in is the connection weight between the input layer and the first feature layer, W in Randomly generated between 0-1;

[0040] Calculate the output matrix U of the second feature layer: U = tanh(GW)

[0041] Where W is the connection weight between the first feature layer and the second feature layer, and W is randomly generated between 0 and 1;

[0042] Estimate the weight matrix W of the output layer through the loss function J out , expressed as:

[0043] Where Y is the lth multipath delay power spectrum In the training phase, the target matrix corresponding to the input matrix X is [X; G; U], which represents the row connection matrix of X, G, and U. ||*||2 represents the l2 norm, λ is the penalty coefficient of the l2 norm, and θ i is the i-th random coefficient between 0 and 1;

[0044] The adaptive particle sphere algorithm is used to solve the loss function J and obtain the estimated output weight matrix

[0045] The predicted value of the lth multipath of the delay power spectrum is obtained using the trained hybrid learning network

[0046] As a preferred solution of the communication networking method for wide-area distributed charging piles described in the present invention, wherein: the adaptive particle sphere algorithm is used to solve the loss function J to obtain the estimated output weight matrix include:

[0047] Initialize the population number M of particle balls s , the space vector of each particle sphere k=1,2,3,...,M s , maximum number of iterations D;

[0048] Define the number of iterations label d, let d = 1;

[0049] Bring it into the solution, including: the k-th particle ball space vector in this iteration process Bring it into the loss function J to solve the fitness value of the k-th particle ball k=1,2,3,...,M s , obtain the optimal particle sphere space vector for the dth iteration

[0050] Update the space vector of the k-th particle ball, expressed as:

[0051] Among them, β, α and Rand are both random coefficients between 0 and 1. is the space vector of the g-th particle sphere in the d-th iteration, g=1,2,3,...,M s ,g≠k;

[0052] Determine whether d is equal to D; if not, then d = d + 1, and return to the solution;

[0053] Output the optimal particle sphere space vector

[0054] As a preferred solution of the communication networking method for wide-area distributed charging piles described in the present invention, wherein: the predicted value of the lth multipath of the delay power spectrum is obtained by using the trained hybrid learning network include:

[0055] In the prediction stage, the output matrix G of the first feature layer is calculated T: G T =tanh(X T W in )

[0056] Among them, X T The lth multipath of the delay power spectrum in the prediction stage Input data;

[0057] Calculate the output matrix U of the second feature layer T : U T =tanh(G T W)

[0058] Calculate the output matrix of the output layer

[0059] in,

[0060] In a second aspect, the present invention provides a communication networking system applicable to wide-area distributed charging piles, comprising:

[0061] Initialization module, used to initialize the total number of wide-area discrete charging piles and their coordinate parameters;

[0062] The judgment module is used to initialize the relevant parameters of the hybrid learning network and use the hybrid learning network to predict and determine the optimal networking node of the charging pile;

[0063] The networking module is used to connect the optimal networking nodes to form a communication network of wide-area dispersed charging piles.

[0064] In a third aspect, the present invention provides a computing device, comprising:

[0065] memory and processor;

[0066] The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, the steps of the communication networking method applicable to wide-area distributed charging piles are implemented.

[0067] In a fourth aspect, the present invention provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the steps of the communication networking method applicable to wide-area distributed charging piles.

[0068] Compared with the existing technology, the beneficial effects of the present invention are as follows: the present invention is aimed at communication networking suitable for wide-area dispersed charging piles, evaluates the channel environment between wide-area dispersed charging piles, and selects the optimal master node according to the quality of the communication environment between the charging piles, thereby forming a communication networking solution for wide-area dispersed charging piles; it makes up for the defect that communication reliability cannot be guaranteed when the charging piles are far away and widely dispersed, and lays the foundation for the full realization of communication networking of electric vehicles in the future. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0070] in:

[0071] FIG1 is a schematic diagram of the overall process of a communication networking method applicable to wide-area distributed charging piles according to an embodiment of the present invention;

[0072] FIG2 is a schematic diagram of an optimal node output process in a communication networking method applicable to wide-area distributed charging piles according to an embodiment of the present invention;

[0073] FIG3 is a schematic diagram of a hybrid learning network output weight optimization process in a communication networking method applicable to wide-area distributed charging piles according to an embodiment of the present invention;

[0074] FIG4 is a schematic diagram showing the connection of system devices applicable to a communication networking method for wide-area distributed charging piles according to an embodiment of the present invention. DETAILED DESCRIPTION

[0075] To make the above-mentioned objects, features, and advantages of the present invention more clearly understood, the following detailed description of the specific embodiments of the present invention is given in conjunction with the accompanying drawings. It is obvious that the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in this field without creative work should fall within the scope of protection of the present invention.

[0076] Example 1

[0077] 1-4 , an embodiment of the present invention provides a communication networking method applicable to wide-area distributed charging piles, including, as shown in FIG1-2 :

[0078] S100: Initialize the total number of wide-area discrete charging piles and their coordinate parameters;

[0079] Furthermore, the total number of wide-area discrete charging piles and their coordinate parameters are initialized, including:

[0080] The coordinates of the wide-area distributed charging pile are expressed as: (x i ,y i ), i = 1, 2, 3, ..., N, N is the total number of wide-area distributed charging piles.

[0081] S200: Initializing relevant parameters of the hybrid learning network, and using the hybrid learning network to predict and determine the optimal networking node of the charging pile;

[0082] Furthermore, the relevant parameters of the hybrid learning network are initialized, and the hybrid learning network is used to determine the optimal networking node of the charging pile, including:

[0083] S201: Define the first master node (x1, y1) as the first optimal master node (x 1,op ,y 1,op ), that is, x1=x 1,op , y1=y 1,op ;

[0084] Perform a first judgment, which includes:

[0085] Determine the distance r between other charging piles and the first optimal master node i1 , expressed as:

[0086] Where i = 2, 3, ..., N;

[0087] S202: Sort the distances of N-1 charging piles in ascending order, and select the n charging piles with the smallest distances to establish a second optimal master node determination data set Φ; wherein the data set Φ is the number of the n charging piles with the smallest distances to the first optimal master node;

[0088] S203: Use the hybrid learning network to predict the channel state information of the first optimal master node and the n charging piles in the data set Φ, and obtain the prediction error e j , j=1,2,3,...,n;

[0089] Furthermore, the hybrid learning network is used to predict the channel state information of the second master node and the n charging piles in the data set Φ to obtain the prediction error e j , j = 1, 2, 3, ..., n, including:

[0090] S2031: Using the channel to estimate the frequency domain channel state information between the optimal node and the j-th charging pile Among them, K and N s are the total number of subcarriers and the number of samples of pilot subcarriers respectively;

[0091] S2032: Obtain the delay power spectrum between the optimal node and the j-th charging pile through inverse discrete Fourier transform

[0092] S2033: Delay power spectrum for the lth multipath Establish a hybrid learning network prediction model to predict the lth multipath value l=1,2,3,...,L T , L T is the total number of transmission multipaths between the optimal node and the jth charging pile;

[0093] Furthermore, for the lth multipath of the delay power spectrum Building a hybrid learning network prediction model includes:

[0094] Initialize the number of neurons P in the first feature layer of the hybrid learning network and the number of neurons Q in the second feature layer. Then the output matrix G of the first feature layer can be expressed as G = tanh(XW in )

[0095] Among them, tanh is the hyperbolic tangent function, X is the lth multipath of the delay power spectrum in the training phase The input data is expressed as:

[0096] N T is the lth multipath delay power spectrum in the training phase The number of data used for training, N s >N T ;W in is the connection weight between the input layer and the first feature layer, W in Randomly generated between 0-1;

[0097] Calculate the output matrix U of the second feature layer: U = tanh(GW)

[0098] Where W is the connection weight between the first feature layer and the second feature layer, and W is randomly generated between 0 and 1;

[0099] Estimate the weight matrix W of the output layer through the loss function J out , expressed as:

[0100] Where Y is the lth multipath delay power spectrum In the training phase, the target matrix corresponding to the input matrix X is [X; G; U], which represents the row connection matrix of X, G, and U. ||*||2 represents the l2 norm, λ is the penalty coefficient of the l2 norm, and θ iis the i-th random coefficient between 0 and 1; the adaptive particle sphere algorithm is used to solve the loss function J and obtain the estimated output weight matrix

[0101] Furthermore, referring to FIG3, the adaptive particle sphere algorithm is used to solve the loss function J to obtain the estimated output weight matrix include:

[0102] A1: Initialize the population size M of particle balls s , the space vector of each particle sphere k=1,2,3,...,M s , maximum number of iterations D;

[0103] A2: Define the iteration number label d, set d = 1;

[0104] A3: Substitute the k-th particle sphere space vector into the solution, including: Bring it into the loss function J to solve the fitness value of the k-th particle ball k=1,2,3,...,M s , obtain the optimal particle sphere space vector for the dth iteration

[0105] A4: Update the space vector of the k-th particle sphere, expressed as:

[0106] Among them, β, α and Rand are both random coefficients between 0 and 1. is the space vector of the g-th particle sphere in the d-th iteration, g=1,2,3,...,M s ,g≠k;

[0107] A5: Determine whether d is equal to D; if not, then d = d + 1, and return to the original solution;

[0108] A6: Output the optimal particle sphere space vector

[0109] The predicted value of the lth multipath of the delay power spectrum is obtained using the trained hybrid learning network

[0110] Furthermore, the trained hybrid learning network is used to obtain the predicted value of the lth multipath of the delay power spectrum include:

[0111] In the prediction stage, the output matrix G of the first feature layer is calculated T : G T =tanh(XT W in )

[0112] Among them, X T The lth multipath of the delay power spectrum in the prediction stage Input data;

[0113] Calculate the output matrix U of the second feature layer T : U T =tanh(G T W)

[0114] Calculate the output matrix of the output layer

[0115] in,

[0116] S2034: The predicted Perform Fourier transform to obtain the predicted frequency domain channel state information Frequency domain channel state information with theoretical Jointly calculate the root mean square error e j , expressed as:

[0117] S204: Sort the prediction errors, define the charging pile with the smallest error as the second best master node, and define the other charging piles in the data set Φ as secondary nodes;

[0118] The corresponding charging pile label in Φ is taken from the total charging pile label dataset Φ 总 Remove it and get the deleted charging pile label total dataset Φ' 总 ;

[0119] S205: Perform a second judgment, which includes:

[0120] Determine the total charging pile label data set Φ' 总 With the second best master node (x 2,op ,y 2,op ) distance r i2 , expressed as:

[0121] in, is the total dataset of charging pile labels Φ' 总 size;

[0122] S206: Yes The distances of the charging piles are sorted, and the n charging piles with the smallest distance are intercepted to update the main node judgment data set Φ;

[0123] S207: Use the hybrid learning network to predict the channel state information of the second master node and the n charging piles in the data set Φ to obtain the prediction error e j , j=1,2,3,...,n;

[0124] S208: Sort the prediction errors, define the charging pile with the smallest error as the third optimal master node, and the other charging piles in the data set Φ as secondary nodes; and remove the corresponding charging pile labels in Φ from the charging pile label data set Φ' 总 Remove and update the total charging pile label dataset Φ' after deletion 总 ;

[0125] S209: Determine Φ' 总 Is it an empty set? If not, loop to the second judgment;

[0126] S210: Output the numbers of the optimal master nodes in sequence.

[0127] It should be noted that S200 evaluates the quality of the communication environment between charging piles through the delay power spectrum. Through continuous iteration, the charging pile corresponding to the optimal communication environment, that is, the optimal master node, can be selected through the optimal delay power spectrum.

[0128] S300: Forming a communication network of wide-area dispersed charging piles based on the optimal networking nodes.

[0129] The above is a schematic diagram of a communication networking method applicable to wide-area distributed charging piles according to this embodiment. It should be noted that the technical solution of the communication networking system applicable to wide-area distributed charging piles and the technical solution of the communication networking method applicable to wide-area distributed charging piles described above are based on the same concept. For details not described in detail in the technical solution of the communication networking system applicable to wide-area distributed charging piles in this embodiment, please refer to the description of the technical solution of the communication networking method applicable to wide-area distributed charging piles described above.

[0130] The communication networking system applicable to wide-area distributed charging piles in this embodiment includes:

[0131] Initialization module, used to initialize the total number of wide-area discrete charging piles and their coordinate parameters;

[0132] The judgment module is used to initialize the relevant parameters of the hybrid learning network and use the hybrid learning network to predict and determine the optimal networking node of the charging pile;

[0133] The networking module is used to connect the optimal networking nodes to form a communication network of wide-area distributed charging piles.

[0134] It should be noted that, referring to FIG4 , the hardware connection of the system may include:

[0135] Signal processor, used to obtain and process power delay spectrum information between charging piles;

[0136] Communication network group controller, used to output the optimal networking node;

[0137] The charging pile networking output module is used to form a communication network of wide-area dispersed charging piles based on the optimal networking nodes.

[0138] This embodiment further provides a computing device suitable for communication networking of widely dispersed charging piles, including:

[0139] Memory and processor; the memory is used to store computer-executable instructions, and the processor is used to execute computer-executable instructions to implement the communication networking method applicable to wide-area distributed charging piles as proposed in the above embodiment.

[0140] This embodiment also provides a storage medium having a computer program stored thereon. When the program is executed by a processor, the communication networking method applicable to wide-area distributed charging piles proposed in the above embodiment is implemented.

[0141] The storage medium proposed in this embodiment and the communication networking method for wide-area distributed charging piles proposed in the above embodiment belong to the same inventive concept. Technical details not fully described in this embodiment can be found in the above embodiment, and this embodiment has the same beneficial effects as the above embodiment.

[0142] Through the above description of the embodiments, those skilled in the art can clearly understand that the present invention can be implemented with the help of software and necessary general-purpose hardware. Of course, it can also be implemented with hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory (FLASH), hard disk or optical disk, etc., and includes a number of instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute the methods of various embodiments of the present invention.

[0143] Example 2

[0144] Based on the previous embodiment, this embodiment provides an application example of a communication networking method applicable to wide-area distributed charging piles:

[0145] The method was verified by collecting data from distributed charging piles in an industrial park in Jiangning District, Nanjing. The industrial park covers an area of ​​20 square kilometers and has 42 distributed charging piles.

[0146] The method of the present invention was experimentally verified using the communication networking method disclosed in this invention, using 20,000 channel state information sampling points collected from each charging pile between 11:30 and 11:31 a.m. on December 24th. The resulting communication networking scheme resulted in the first 10 charging piles being numbered 23-9-41-28-18-14-18-31-2. This demonstrates that the communication networking scheme disclosed in this invention can effectively address the communication networking issues faced by widely dispersed charging piles.

[0147] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A communication networking method applicable to wide-area distributed charging piles, characterized in that, Including: Initializing the total number of wide-area discrete charging piles and their coordinate parameters; Initializing the relevant parameters of the hybrid learning network, and using the hybrid learning network to make predictions and determine the optimal networking nodes of the charging piles; Connecting according to the optimal networking nodes to form a communication network for the wide-area distributed charging piles.

2. The communication networking method applicable to wide-area distributed charging piles according to claim 1, wherein, Initialize the total number of the wide-area discrete charging piles and their coordinate parameters, including: The coordinates of the wide-area dispersed charging piles are represented as: (x i , y i ), where i = 1, 2, 3,..., N, and N is the total number of the wide-area dispersed charging piles.

3. The communication networking method applicable to wide-area distributed charging piles according to claim 1 or 2, characterized in that The initializing the relevant parameters of the hybrid learning network and using the hybrid learning network to determine the optimal networking nodes of the charging piles includes: Define the first main node (x1, y1) as the first optimal main node (x 1,op , y 1,op ), that is, x1 = x 1,op , y1 = y 1,op ; Performing a first judgment, and the first judgment includes: Determine the distance r between other charging piles and the first optimal master node i1 , which is expressed as: where i = 2, 3,..., N; Sorting the distances of N - 1 charging piles from small to large, and intercepting the n charging piles with the smallest distances to establish a second optimal master node judgment data set Φ; where the data set Φ is the numbers of the n charging piles with the smallest distances from the first optimal master node; Using the hybrid learning network to calculate the distances between the first optimal master node and the n charging Predict the channel state information of the pile and obtain the prediction error e j , j = 1, 2, 3,..., n; Sorting the prediction errors, defining the charging pile with the smallest error as the second optimal master node, and defining the other charging piles in the data set Φ as secondary nodes; Remove the charging pile labels corresponding to Φ from the total dataset Φ of charging pile labels 总 to obtain the reduced total dataset Φ' of charging pile labels 总 ; Performing a second judgment, and the second judgment includes: Judge the total data set Φ' of charging pile labels 总 from the second optimal master node (x 2,op , y 2,op ), with the distance r i2 , expressed as: Among them, For the total data set Φ' of charging pile labels 总 in size; Pair Sorting the distances between the first optimal master node and the n charging piles, and intercepting the n charging piles with the smallest distances to update the master node judgment data set Φ; Predict the channel state information of the second master node and n charging piles in the data set Φ using a hybrid learning network to obtain the prediction error e j , j = 1, 2, 3,..., n; Sort the prediction errors, define the charging pile with the smallest error as the third optimal main node, and the other charging piles in the dataset Φ as secondary nodes; remove the corresponding charging pile labels in Φ from the charging pile label dataset Φ' 总 in it, and update the total dataset Φ' of the charging pile labels after deletion 总 ; Determine Φ' 总 whether it is an empty set; if not, loop the second determination; Sequentially outputting the numbers of the optimal master nodes.

4. The communication networking method applicable to a wide-area distributed charging pile according to claim 3, wherein Predicting the channel state information of the second master node and n charging piles in the data set Φ by using the hybrid learning network to obtain the prediction error e j , j = 1, 2, 3,..., n, including: Utilize the channel estimation to obtain the frequency-domain channel state information of the optimal node and the j-th charging pile Among them, K and N s are the total number of subcarriers and the sampling number of pilot subcarriers, respectively; Obtain the time-delay power spectrum of the optimal node and the j-th charging pile through the inverse discrete Fourier transform The l-th multipath of the delay power spectrum Build a hybrid learning network prediction model to predict the predicted value of the l-th multipath L T is the total number of transmission multipaths between the optimal node and the j-th charging pile; The predicted Perform Fourier transform to obtain the predicted frequency-domain channel state information With the theoretical frequency-domain channel state information Jointly calculate the root mean square error e j , expressed as:

5. The communication networking method applicable to wide-area distributed charging piles according to claim 4, characterized in that The l-th multipath of the delay power spectrum Establish a hybrid learning network prediction model, including: Initializing the number of neurons P in the first feature layer and the number of neurons Q in the second feature layer of the hybrid learning network, then the output matrix G of the first feature layer can be expressed as G = tanh(XW in ) where tanh is the hyperbolic tangent function, and X is the l-th multipath with respect to the delay power spectrum in the training stage The input data, expressed as: N T For the l-th multipath of the time-delay power spectrum in the training phase The number of data for training, N s >N T ; W in is the connection weight between the input layer and the first feature layer, W in is randomly generated between 0 and 1; Calculating the output matrix U of the second feature layer: U = tanh(GW) where W is the connection weight between the first feature layer and the second feature layer, and W is randomly generated between 0 and 1; Estimate the weight matrix W of the output layer through the loss function J out , expressed as: where Y is the l-th multipath of the delay power spectrum The target matrix corresponding to the input matrix X in the training phase, [X; G; U] represents the row concatenation matrix with respect to X, G, and U, ||*||2 represents the l2 norm, λ is the penalty coefficient of the l2 norm, and θ i is the i-th random coefficient between 0 and 1; Solve the loss function J using the adaptive particle sphere algorithm to obtain the estimated output weight matrix Obtain the predicted value of the \(l\)-th multipath of the delay power spectrum using the trained hybrid learning network 6. The communication networking method applicable to wide-area distributed charging piles according to claim 5, wherein Solving the loss function J by using the adaptive particle sphere algorithm to obtain an estimated output weight matrix including: Initialize the population size M of the particle spheres s , and the space vector of each particle sphere The maximum number of iterations D; Defining an iteration number label d, and setting d = 1; Perform substitution and solution, including: the spatial vector of the k-th particle sphere in the current iteration process Substitute it into the loss function J to solve the fitness value of the k-th particle sphere Particle sphere space vector that is optimal for the d-th iteration Update the spatial vector of the k-th particle sphere, expressed as: where β, Both α and Rand are random coefficients between 0 and 1. is the space vector of the g-th particle sphere in the d-th iteration process, where g = 1, 2, 3, ..., M s , where g ≠ k; Judging whether d is equal to D; if not, then d = d + 1, and returning to the above substitution and solution; Output the optimal particle sphere space vector 7. The communication networking method applicable to wide-area distributed charging piles according to claim 5 or 6, characterized in that Obtaining the predicted value of the l-th multipath of the delay power spectrum by using the trained hybrid learning network including: During the prediction stage, calculate the output matrix G of the first feature layer T : G T = tanh(X T W in ) Among them, X T is the l-th multipath of the delay power spectrum in the prediction stage The input data; Calculate the output matrix U of the second feature layer T : U T = tanh(G T W) Calculate the output matrix of the output layer Among them, 8. A communication networking system suitable for wide-area distributed charging piles, characterized in that: An initialization module for initializing the total number of wide-area discrete charging piles and their coordinate parameters; A judgment module for initializing the relevant parameters of the hybrid learning network and using the hybrid learning network to make predictions and determine the optimal networking nodes of the charging piles; A networking module for connecting according to the optimal networking nodes to form a communication network for the wide-area distributed charging piles.

9. An electronic device, including: A memory and a processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, the steps of the communication networking method for wide-area distributed charging piles according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing computer-executable instructions, and when the computer-executable instructions are executed by a processor, the steps of the communication networking method for wide-area distributed charging piles according to any one of claims 1 to 7 are implemented.

Images