CN112613637A - Method and device for processing charging load - Google Patents
Method and device for processing charging load Download PDFInfo
- Publication number
- CN112613637A CN112613637A CN202011377051.8A CN202011377051A CN112613637A CN 112613637 A CN112613637 A CN 112613637A CN 202011377051 A CN202011377051 A CN 202011377051A CN 112613637 A CN112613637 A CN 112613637A
- Authority
- CN
- China
- Prior art keywords
- charging
- load
- data
- historical
- training
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 50
- 238000012545 processing Methods 0.000 title claims abstract description 25
- 238000012549 training Methods 0.000 claims description 76
- 238000003062 neural network model Methods 0.000 claims description 36
- 230000006870 function Effects 0.000 claims description 21
- 239000011159 matrix material Substances 0.000 claims description 11
- 238000003672 processing method Methods 0.000 claims description 6
- 238000000540 analysis of variance Methods 0.000 claims description 5
- 238000005516 engineering process Methods 0.000 abstract description 4
- 238000013528 artificial neural network Methods 0.000 description 17
- 238000009826 distribution Methods 0.000 description 13
- 238000010586 diagram Methods 0.000 description 7
- 230000008569 process Effects 0.000 description 7
- 238000004364 calculation method Methods 0.000 description 6
- 230000008859 change Effects 0.000 description 5
- 238000001914 filtration Methods 0.000 description 5
- 238000004458 analytical method Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 230000008878 coupling Effects 0.000 description 3
- 238000010168 coupling process Methods 0.000 description 3
- 238000005859 coupling reaction Methods 0.000 description 3
- 238000007405 data analysis Methods 0.000 description 3
- 238000012804 iterative process Methods 0.000 description 3
- 238000007781 pre-processing Methods 0.000 description 3
- 230000006399 behavior Effects 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 238000005034 decoration Methods 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 229920002472 Starch Polymers 0.000 description 1
- 239000002131 composite material Substances 0.000 description 1
- 238000010219 correlation analysis Methods 0.000 description 1
- 125000004122 cyclic group Chemical group 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 230000005284 excitation Effects 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 238000001543 one-way ANOVA Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 235000019698 starch Nutrition 0.000 description 1
- 239000008107 starch Substances 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 230000007704 transition Effects 0.000 description 1
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
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- Artificial Intelligence (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Human Resources & Organizations (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Economics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Strategic Management (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Data Mining & Analysis (AREA)
- Development Economics (AREA)
- Game Theory and Decision Science (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Quality & Reliability (AREA)
- Operations Research (AREA)
- Marketing (AREA)
- Entrepreneurship & Innovation (AREA)
- Charge And Discharge Circuits For Batteries Or The Like (AREA)
Abstract
The invention discloses a method and a device for processing a charging load. Wherein, the method comprises the following steps: acquiring historical charging data; dividing historical charging data to obtain historical data corresponding to a plurality of grids; and predicting the historical data corresponding to each grid by using a load prediction model to obtain the charging load of each grid in a preset time period. The invention solves the technical problem of lower accuracy of determination of the charging load in the related technology.
Description
Technical Field
The invention relates to the field of new energy automobiles, in particular to a method and a device for processing a charging load.
Background
The charging behavior of the electric vehicle user is an important factor influencing the charging load of the electric vehicle, and the difference of the charging behavior means the difference of the vehicle driving characteristics of the vehicle, and the different driving characteristics necessarily lead to different charging load requirements. However, due to the limited number of the existing charging facilities and the unreasonable layout, the actual charging requirement of the electric vehicle cannot be correctly represented by the historical charging data, and the determination accuracy of the charging load is low.
In view of the above problems, no effective solution has been proposed.
Disclosure of Invention
The embodiment of the invention provides a method and a device for processing a charging load, which are used for at least solving the technical problem of low accuracy in determination of the charging load in the related art.
According to an aspect of the embodiments of the present invention, there is provided a method for processing a charging load, including: acquiring historical charging data; dividing historical charging data to obtain historical data corresponding to a plurality of grids; and predicting the historical data corresponding to each grid by using a load prediction model to obtain the charging load of each grid in a preset time period.
Optionally, the dividing the historical charging data to obtain historical data corresponding to a plurality of grids includes: acquiring a target area corresponding to historical charging data; dividing the target area according to the longitude and latitude information to obtain a plurality of grids; and distributing the historical charging data to a plurality of grids to obtain the historical data corresponding to each grid.
Optionally, the load prediction model is obtained by training a back propagation neural network model by using a bayesian regularization algorithm.
Optionally, the method further comprises: acquiring a prediction index corresponding to the charging load, wherein the prediction index is used for representing factors influencing the charging load; constructing a back propagation neural network model; generating a training sample by using the prediction index; and training the back propagation neural network model by using a Bayesian regularization algorithm and a training sample to obtain a load prediction model.
Optionally, the obtaining of the prediction index corresponding to the charging load includes: acquiring a plurality of target data corresponding to the charging load, wherein the plurality of target data comprise: operation mode, number of charging facilities, number of users, regional function, traffic flow and number of parking spaces; processing a plurality of target data by using an analysis of variance method, and determining the importance degree of each target data; based on the degree of importance of each target data, a prediction index is determined.
Optionally, the back propagation neural network model comprises: the device comprises an input layer, a hidden layer and an output layer, wherein the number of nodes of the input layer is determined based on the number of the prediction indexes, the number of nodes of the output layer is 1, and the number of nodes of the hidden layer is determined based on the number of nodes of the input layer and the number of nodes of the output layer by using a trial and error method.
Optionally, generating the training sample using the prediction index includes: normalizing the prediction index to obtain normalized data; training samples are generated based on the normalized data.
Optionally, training the back propagation neural network model by using a bayesian regularization algorithm and a prediction index, and obtaining the load prediction model includes: inputting the training sample into a back propagation neural network model to obtain a prediction result of the training sample; obtaining a mean square error based on an original result of the training sample and a prediction result of the training sample; acquiring the square sum mean value of network parameters of a back propagation neural network model; obtaining the total error of the back propagation neural network model based on the mean square error and the mean square sum; and updating the regularization parameters of the back propagation neural network model based on the total error to obtain a load prediction model.
Optionally, updating the regularization parameter of the back propagation neural network model based on the total error, and obtaining the load prediction model includes: judging whether the total error is greater than a preset error or not and whether the training times reach preset times or not; if the total error is smaller than the preset error or the training times reach the preset times, determining that the training of the load prediction model is finished; and if the total error is greater than the preset error and the training times do not reach the preset times, updating the regularization parameter based on the preset learning rate.
Optionally, updating the regularization parameter based on the preset learning rate comprises: acquiring a Jacobian matrix of the total error; solving the Jacobian matrix by using a Gauss-Newton approximation method to obtain the number of effective parameters; and obtaining updated regularization parameters based on the mean square error, the mean square sum, the number of effective parameters and a preset learning rate.
According to another aspect of the embodiments of the present invention, there is also provided a processing apparatus of a charging load, including: the acquisition module is used for acquiring historical charging data; the dividing module is used for dividing the historical charging data to obtain historical data corresponding to a plurality of grids; and the prediction module is used for predicting the historical data corresponding to each grid by using the load prediction model to obtain the charging load of each grid in a preset time period.
According to another aspect of the embodiments of the present invention, there is also provided a computer-readable storage medium, where the computer-readable storage medium includes a stored program, and when the program runs, the apparatus where the computer-readable storage medium is located is controlled to execute the processing method of the charging load.
According to another aspect of the embodiments of the present invention, there is also provided a processor, configured to execute a program, where the program executes the processing method of the charging load.
In the embodiment of the invention, after the historical charging data is acquired, the historical charging data can be divided to obtain the historical data corresponding to a plurality of grids, and the load prediction model is used for predicting the historical data corresponding to each grid to obtain the charging load of each grid in a preset time period, so that the purpose of predicting the charging load of the electric automobile is realized. It is easy to notice that, since the historical charging data can be divided, and the charging load in the preset time period can be obtained through prediction, the space-time distribution of the charging load of the electric vehicle is realized, the technical effect of improving the prediction accuracy is achieved, and the technical problem of low accuracy in determining the charging load in the related technology is solved.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this application, illustrate embodiment(s) of the invention and together with the description serve to explain the invention without limiting the invention. In the drawings:
fig. 1 is a flowchart of a method of processing a charging load according to an embodiment of the present invention;
fig. 2 is a schematic diagram of a predicted value and a historical true value of a charging load according to an embodiment of the present invention;
FIG. 3 is a schematic illustration of a 24 hour charging load distribution without a charging facility grid according to an embodiment of the present invention;
FIG. 4 is a thermodynamic diagram of a peak charge load over a day for each grid in accordance with an embodiment of the present invention;
FIG. 5 is a thermodynamic diagram of the total electric vehicle charge per day for each grid in accordance with an embodiment of the present invention;
FIG. 6 is a thermodynamic diagram of the total electric vehicle charge per day for each grid according to the prior art;
fig. 7 is a schematic diagram of a processing device for charging a load according to an embodiment of the present invention.
Detailed Description
In order to make the technical solutions of the present invention better understood, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
It should be noted that the terms "first," "second," and the like in the description and claims of the present invention and in the drawings described above are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used is interchangeable under appropriate circumstances such that the embodiments of the invention described herein are capable of operation in sequences other than those illustrated or described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
Example 1
According to an embodiment of the present invention, there is provided a method for processing a charging load, it should be noted that the steps shown in the flowchart of the figure may be executed in a computer system such as a set of computer executable instructions, and that although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in an order different from that here.
Fig. 1 is a flowchart of a method for processing a charging load according to an embodiment of the present invention, as shown in fig. 1, the method including the steps of:
in step S102, historical charging data is acquired.
The historical charging data in the steps can be data such as charging pile asset information, charging transaction records, used parking space number of vehicles in a parking lot, area function proportion, traffic flow and the like in a predicted area. The predicted area may be a street, a region, etc., and may be, for example, the Haishen area of Beijing.
And step S104, dividing the historical charging data to obtain historical data corresponding to a plurality of grids.
In an optional embodiment, historical charging data in one area may be divided according to an area division manner to obtain historical charging data corresponding to each grid, that is, the historical data.
And step S106, predicting the historical data corresponding to each grid by using the load prediction model to obtain the charging load of each grid in a preset time period.
The preset time period may be a time period determined according to actual prediction needs, for example, may be a day, a week, and the like, and in the embodiment of the present invention, 24 hours a day is taken as an example.
In an alternative embodiment, the process flow of the charging load prediction of the electric vehicle is as follows: historical charging data can be divided to serve as a basic unit for space prediction of charging load of the electric automobile; and then, processing the existing historical charging data by using a pre-trained load prediction model, taking the historical charging data as input data, wherein the output result of the model is the space-time distribution of the charging load of the electric automobile.
Optionally, the load prediction model may be obtained by training a back propagation neural network model by using a bayesian regularization algorithm.
The Back Propagation (BP) neural network is a global approximation network, and can implement nonlinear mapping from an input space to an output space through composite mapping of a plurality of simple nonlinear processing units. However, the traditional BP algorithm processing has the problems of low convergence speed and easy falling into local minimum, and the generalization capability of the traditional BP algorithm processing is general, so that the accuracy of prediction is influenced. Aiming at the problem, a Bayesian regularization algorithm can be adopted in the embodiment of the invention to improve the generalization capability of the BP neural network model.
Optionally, the back propagation neural network model includes: the device comprises an input layer, a hidden layer and an output layer, wherein the number of nodes of the input layer is determined based on the number of the prediction indexes, the number of nodes of the output layer is 1, and the number of nodes of the hidden layer is determined based on the number of nodes of the input layer and the number of nodes of the output layer by using a trial and error method.
The accuracy and the training efficiency of the network model depend on the network structure and control parameters to a great extent, and the network structure can comprise network parameters, the number of neurons in each layer and the like; the control parameters may refer to input variables, allowable errors, learning rates, initial weight values, and the like.
In the embodiment of the present invention, a prediction model can be established by using a three-layer BP neural network, that is, an input layer, a hidden layer, and an output layer, and the function of each layer is the same as that of the existing one, which is not described herein again.
The number of nodes of the input layer and the output layer can be determined by actual processing requirements and is the number of variables actually input and output by the model. The output variable of the model is the charging amount of the electric automobile, so the number of nodes of the output layer is 1.
The number of hidden layer nodes is related to the complexity of the problem to be solved and the number of input and output variables, and has a great influence on the performance of the network. If the number of hidden layer nodes is too small, the problem that a neural network is difficult to process is solved, and modeling is insufficient; the number of nodes of the hidden layer is too many, which not only can cause the problems of complex structure, long iteration time, large calculation amount and the like of the neural network, but also can provide redundant degrees of freedom to adapt to noise, so that the network is over-trained, and the error is not necessarily optimal. In the embodiment of the present invention, the number of nodes in the hidden layer may be calculated according to a formula:
wherein J represents the number of nodes of the hidden layer, I and O represent the number of nodes of the input layer and the output layer respectively, a is a constant between [1 and 10], the number of nodes of the hidden layer can be determined by a trial and error method, namely, a plurality of BP neural networks with the same conditions except the number of nodes of the hidden layer are constructed, the training is performed by the same sample, and the optimal number of nodes of the hidden layer is determined by comparing the training time and the prediction precision of each network.
In an alternative embodiment, since the charging amount without the charging facility grid is added on the basis of the original historical charging amount, the historical charging data should be proportionally distributed to all grids, and the charging amount in each grid in one day is the charging amount:
wherein m and k represent the number of grids containing charging facilities and the number of grids not containing charging facilities in the target area, respectively, CiRepresents the predicted amount of charge in grid i of the neural network in one day, CtotalRepresents the total charge of all grids in the area in one day, namely the historical total charge of the grids containing the charging facility,
representing the amount of charge for grid i in one day.
The change of the electric vehicle holding capacity is approximately represented by the change of the new energy vehicle holding capacity, and the change of the electric vehicle holding capacity of the prediction region is approximately represented by the change of the national electric vehicle holding capacity. The spatial distribution of factors influencing the charging quantity of the electric automobile in the predicted area, such as the area function, the population number, the traffic flow and the like, is assumed to be basically unchanged. On the basis of the foregoing assumption, if the current new energy vehicle reserve is c thousands of vehicles and the predicted annual reserve is f thousands of vehicles, the predicted electric vehicle charge amount of each grid of the year may be represented as:
wherein the content of the first and second substances,
the predicted value of the electric vehicle charging amount when the new energy vehicle holding amount is f ten thousands of vehicles is shown,
the electric vehicle charge amount when the new energy vehicle holds c ten thousands of vehicles, that is, the historical charge amount is represented.
Through the embodiment of the invention, after the historical charging data is acquired, the historical charging data can be divided to obtain the historical data corresponding to a plurality of grids, the historical data corresponding to each grid is predicted by using the load prediction model to obtain the charging load of each grid in the preset time period, and the purpose of predicting the charging load of the electric automobile is achieved. It is easy to notice that, since the historical charging data can be divided, and the charging load in the preset time period can be obtained through prediction, the space-time distribution of the charging load of the electric vehicle is realized, the technical effect of improving the prediction accuracy is achieved, and the technical problem of low accuracy in determining the charging load in the related technology is solved.
Optionally, in the above embodiment of the present invention, the dividing the historical charging data to obtain the historical data corresponding to the multiple grids includes: acquiring a target area corresponding to historical charging data; dividing the target area according to the longitude and latitude information to obtain a plurality of grids; and distributing the historical charging data to a plurality of grids to obtain the historical data corresponding to each grid.
In an alternative embodiment, the target area may be divided into grids of the same size with a certain precision, so as to provide a basis for space prediction of the subsequent electric vehicle charging demand and refined charging facility planning. For example, the model algorithm is validated and analyzed by taking the haih lake as an example, and considering that the traffic flow data and the population data are counted according to a 1000m × 1000m grid, and the area of the 1000m side length grid is moderate and within the range acceptable for the user to walk, the haih lake can be divided into 1000m × 1000m grids according to longitude and latitude coordinates.
It should be noted that the area meshing for the charging network planning may refer to a meshing method for the parking lot planning, which is not described herein again.
Optionally, in the above embodiment of the present invention, the method further includes: acquiring a prediction index corresponding to the charging load, wherein the prediction index is used for representing factors influencing the charging load; constructing a back propagation neural network model; generating a training sample by using the prediction index; and training the back propagation neural network model by using a Bayesian regularization algorithm and a training sample to obtain a load prediction model.
In an optional embodiment, a mat lab may be used to generate, train, and test a BP neural network model, and a trained load prediction model is used to predict a charging load of an electric vehicle, which is specifically implemented as follows: factors which have obvious influence on the charging amount of the electric automobile can be selected as prediction indexes; setting the number of hidden layer nodes and a training function by using a feedback forward net function to generate a BP neural network model; training the network by utilizing a train function to obtain the relationship between each prediction index and the charging load; and finally, inputting each prediction index of the area to be predicted into the trained network, and outputting a corresponding 24-hour charging load prediction value.
Optionally, in the above embodiment of the present invention, acquiring the prediction index corresponding to the charging load includes: acquiring a plurality of target data corresponding to the charging load, wherein the plurality of target data comprise: operation mode, number of charging facilities, number of users, regional function, traffic flow and number of parking spaces; processing a plurality of target data by using an analysis of variance method, and determining the importance degree of each target data; based on the degree of importance of each target data, a prediction index is determined.
The analysis of variance, also called variance analysis, can be used for the significance test of the mean difference of two or more samples. One-way anova can be used to investigate whether different levels of a control variable have a significant effect on the observed variable.
In an alternative embodiment, factors that may affect the charging requirements of electric vehicles (particularly those charged by charging piles at the same company) may be collected: operation mode (parking charge, electricity price, etc.), number of charging facilities of other operators than the grid company, population number, regional function (mall, residential area, work area, etc.), traffic flow, parking lot number of cars, etc. (i.e., the above-mentioned target data). And identifying the influence characteristics of the electric vehicle charging amount and all factors by using an analysis of variance method, exploring and extracting the influence degree, and evaluating the importance degree of all the influence factors, thereby taking the factors which have obvious influence on the electric vehicle charging amount as prediction indexes.
Optionally, in the foregoing embodiment of the present invention, generating the training sample using the prediction index includes: normalizing the prediction index to obtain normalized data; training samples are generated based on the normalized data.
In order to facilitate calculation, eliminate dimension influence and improve prediction accuracy, input data needs to be preprocessed. Therefore, in practical applications, it is necessary to normalize the input variables and the output variables.
In an alternative embodiment, the following data normalization formula may be used:
wherein, PnIs normalized data, P is the original input data, PminAnd PmaxRespectively the minimum and maximum in the original input data set. After this normalization process, the data values are transformed to [ -1,1 [ -1]And the training of the neural network is facilitated.
Optionally, in the foregoing embodiment of the present invention, training the back propagation neural network model by using a bayesian regularization algorithm and a prediction index, and obtaining the load prediction model includes: inputting the training sample into a back propagation neural network model to obtain a prediction result of the training sample; obtaining a mean square error based on an original result of the training sample and a prediction result of the training sample; acquiring the square sum mean value of network parameters of a back propagation neural network model; obtaining the total error of the back propagation neural network model based on the mean square error and the mean square sum; and updating the regularization parameters of the back propagation neural network model based on the total error to obtain a load prediction model.
The Bayesian regularization algorithm improves the generalization capability of the neural network by modifying the training performance function of the neural network. In general, the training performance function of the neural network adopts a mean square error mse, that is:
wherein n is the total number of samples, tiTo a desired output, aiIs the actual output of the network.
In the Bayesian regularization algorithm, the network performance function is changed into the following form:
msereg＝α.msw+β.mse，
wherein msereg is an improved performance function, msw is the mean of the sum of squares of all network weights, alpha and beta are regularization parameters, the size of the regularization parameters has a remarkable influence on the network training effect, m is the total number of the network weights, and omega isiIs the network weight.
In an alternative embodiment, the training step of the bayesian regularization BP neural network is as follows: determining a network structure, initializing hyper-parameters alpha and beta, generally setting alpha to 0 and beta to 1, and assigning initial values to network parameters according to prior distribution; training the network by using a BP algorithm to minimize the total error msereg; the regularization parameters alpha and beta are adaptively adjusted and optimized during the network training process.
The Bayesian regularization neural network is an iterative process, the total error function of each iterative process changes along with the change of the hyper-parameters alpha and beta, the minimum value point also changes, the parameters of the network are also continuously corrected, and finally the total error function is not greatly changed in the iterative process, which is also called as convergence.
Optionally, in the foregoing embodiment of the present invention, updating the regularization parameter of the back propagation neural network model based on the total error, and obtaining the load prediction model includes: judging whether the total error is greater than a preset error or not and whether the training times reach preset times or not; if the total error is smaller than the preset error or the training times reach the preset times, determining that the training of the load prediction model is finished; and if the total error is greater than the preset error and the training times do not reach the preset times, updating the regularization parameter based on the preset learning rate.
The preset error may be the minimum allowable error of the output, and may obtain better learning accuracy, but there is transition fitting to the sample, and the popularization capability of the network is poor, and in the above embodiment of the present invention, the preset error may be 0.001.
The preset number may be the maximum training number of the BP neural network model, which may prevent the network from entering a local minimum value in a certain training and running indefinitely without reaching a minimum error, and in the above embodiment of the present invention, the preset number may be 1000.
In an alternative embodiment, when the total error msereg is smaller than the preset error, it may be determined that training is completed to obtain the load prediction model, and when the total error msereg is larger than the preset error, it may be determined that training is not completed and training needs to be continued. In order to avoid the unlimited operation of the network, whether the training times reach the preset times can be judged while the total error msereg is judged, and if so, the training is finished no matter the size of the total error msereg.
Optionally, in the foregoing embodiment of the present invention, updating the regularization parameter based on the preset learning rate includes: acquiring a Jacobian matrix of the total error; solving the Jacobian matrix by using a Gauss-Newton approximation method to obtain the number of effective parameters; and obtaining updated regularization parameters based on the mean square error, the mean square sum, the number of effective parameters and a preset learning rate.
For the regularization parameter, if alpha is far greater than beta, training emphasizes the generalization performance of the network, but the under-fitting phenomenon is easily caused; if α is much smaller than β, the training algorithm tends to reduce the corresponding error of the learning set network, and the network is prone to overfitting. The conventional regularization method is difficult to determine the sizes of the regularization parameters alpha and beta, and the Bayesian theory can be adopted to adaptively adjust the sizes of the regularization parameters in the network training process and enable the regularization parameters to be optimal. The regularization parameter prediction formula of the algorithm is as follows:
where γ is called the number of significant parameters:
γ＝m-2αtr(H)-1，
where H is the Hessian matrix for msereg:
in the optimization solution, a Hessian matrix of the msereg at the minimum point of the msereg needs to be calculated, and the calculation amount is large. In order to improve the calculation speed, the Hessian matrix can be further simplified by using a gauss-newton approximation method to obtain:
H≈2αIm+2βJTJ，
where J is the Jacobian matrix for mse.
The learning rate reflects the variable quantity of the weight in one-time cyclic training of the network, and is an important factor in optimization calculation, and each specific network has a proper learning rate. The reasonable speed is selected, so that the convergence speed and the training time are directly influenced, and the control precision can be greatly improved. The learning rate is typically selected to be in the range of 0.01-0.1. In the embodiment of the invention, the preset learning rate is 0.01 as an example, the convergence speed is high, and the training time is short.
In an optional embodiment, a gaussian-newton approximation method can be used to solve a Hessian matrix, the number gamma of effective parameters is calculated, and the new estimated values of the hyper-parameters alpha and beta are calculated by using the formula and combining with a preset learning rate; and updating through multiple times of training until the training is finished.
The hai lake area in Beijing is selected as a prediction area to carry out sample analysis, and firstly, the hai lake area is divided into 1000m by 1000m grids according to longitude and latitude coordinates, wherein 52 grids containing charging facilities exist, and 403 grids without the charging facilities exist. On the basis of grid division, data such as charging pile asset information, charging transaction records, the number of used cars in a parking lot, area function proportion, traffic flow and the like are imported according to each grid.
(1) Charging transaction record
Building a charging transaction record data warehouse, and importing and storing the asset information of the charging pile in Beijing and 250 ten thousand charging records from 4 months to 9 months in 2018; connecting the charging pile asset information and the charging record according to the charging pile operation number to determine the space position where the charging transaction record occurs, filtering dirty data in data preprocessing, obtaining that the effective data size of a lake region is 42 thousands of data after filtering, and then formatting the correct data; and then in a data analysis and statistics link, counting the data from a time dimension according to 24 hours a day, counting from a space dimension according to a longitude and latitude grid, and averaging 183-day data after counting so as to obtain a typical daily charging information statistical curve under different space-time distributions.
(2) Regional function
Building a regional functional data warehouse, and importing and storing 14.5 ten thousand map POI data in a Hai lake area of Beijing; filtering dirty data in data preprocessing, wherein the size of the filtered data is 13 ten thousand, and then formatting correct data; and then, in a data analysis and statistics link, statistics is carried out according to longitude and latitude grids in the spatial dimension, and then the proportion of POI points representing different industries in each grid is calculated, so that the proportion of different functional areas in each grid is obtained.
(3) Flow of traffic
Building a traffic flow data warehouse, and importing and storing 168 ten thousand pieces of traffic flow data of 9 days including 9 months 11 to 17 days, 5 months 1 day and 10 months 1 day in 2017 of Beijing; filtering dirty data in data preprocessing, obtaining the data scale of the effective data of the starch area after filtering is 29 ten thousand, and then formatting correct data; and in a data analysis and statistics link, counting the data from a time dimension according to typical working days, weekends and holidays, counting from a space dimension according to longitude and latitude grids, and calculating a weighted average value according to the proportion of the working days, weekends and holidays in one year so as to obtain typical day traffic flow data in each grid.
The correlation analysis results of the influence factors of the electric automobile charging demand are shown in table 1, and according to the analysis results, three factors, namely an area function which has a significant influence on the electric automobile charging demand, the number of used parking lots and the traffic flow, are used as prediction indexes of an electric automobile charging load prediction model.
TABLE 1
From the above analysis, the model finally extracts 2 prediction indexes of the area function and the traffic flow, but the area function is divided into 19 types such as a residential area, a business area, an office area, and the like, and the ratio of each of them in the mesh is used as input data of the model, so the number of the input layer nodes is 20. The output quantity of the model is 24 hours of charging load of the electric automobile, so the number of nodes of the output layer is 24. The number of hidden layer nodes is determined to be 12 by trial and error.
And (3) constructing a BP neural network by using MATLAB, setting an excitation function between an input layer and a hidden layer as tansig, setting a training function as a trainbr function of a Bayes regularization algorithm, wherein the learning rate is 0.01, the maximum training frequency is 1000, the training error is 0.01, and the rest parameters are default values. And taking each prediction index of a charging facility grid in the area and the historical charging load which is averaged for 24 hours a day as input and output data of neural network training, and training the network by utilizing a train function to obtain the relation between each prediction index and the charging load. And finally, inputting all the prediction indexes without the charging facility grids into the trained neural network, and outputting the 24-hour charging load prediction value without the charging facility grids.
The comparison between the predicted value and the historical true value of the charging load obtained by the BP neural network model is shown in FIG. 2, and it can be seen from the graph that the error between the predicted value and the true value of the charging load is small, so that the BP neural network model can accurately predict the 24-hour distribution of the charging load of the electric automobile.
The 24-hour distribution of the electric vehicle charging load in each grid can be obtained by a neural network prediction model, as shown in table 2. On the basis of the 24-hour distribution of the charging load, the peak value of the charging load and the total charging capacity in one day of each grid can be obtained, as shown in tables 3 and 4. Fig. 3 shows the 24-hour charging load distribution of 3 grids without charging facilities from the predictive model. The peak value of the charging load in one day of each grid is shown in fig. 4, and the total charging capacity in one day is shown in fig. 5.
TABLE 2
TABLE 3
TABLE 4
The total charge amount of the electric vehicle in each grid for one day obtained from the historical charging data of the existing charging facility is shown in table 5, and the thermodynamic diagram of the total charge amount is shown in fig. 6.
TABLE 5
As can be seen from fig. 6, the number of the existing charging facilities is limited, and only a few grids exist, and the historical charging data cannot correctly reflect the spatial distribution of the charging requirements of the electric vehicle. By adopting the method for predicting the charging load of the electric vehicle provided by the embodiment of the invention, the charging load of the existing charging facility can be distributed to all grids, and the 24-hour distribution of the charging load in one day and the total charging amount in one day of each grid can be obtained. Comparing fig. 5 and fig. 6, it can be seen that the magnitude of the predicted value of the charging capacity in fig. 5 is significantly smaller than the historical true value of the charging capacity in fig. 6, because the historical charging capacity in fig. 6 is distributed to all grids to ensure that the total charging capacity is not changed.
According to the scheme, on the basis of planning regional grid division, historical charging data including charging facility grids are reasonably distributed to each grid through a Bayesian regularization BP neural network algorithm, an electric vehicle charging load prediction model is established by taking the grids as basic units, and the electric vehicle charging load prediction model comprises electric vehicle charging frequency prediction and average charging load calculation of one-time charging, so that the problem that the actual charging requirement of an electric vehicle cannot be correctly reflected by the historical charging data due to the fact that the existing charging facilities are limited in number and unreasonable in layout is solved.
Example 2
According to an embodiment of the present invention, a device for processing a charging load is provided, where the device may perform the method for processing a charging load in the foregoing embodiment, and a specific implementation scheme and a preferred application scenario are the same as those in the foregoing embodiment, and are not described herein again.
Fig. 7 is a schematic diagram of a processing apparatus for charging a load according to an embodiment of the present invention, as shown in fig. 7, the apparatus including:
an obtaining module 72, configured to obtain historical charging data;
the dividing module 74 is configured to divide the historical charging data to obtain historical data corresponding to a plurality of grids;
and a predicting module 76, configured to predict, by using the load prediction model, historical data corresponding to each grid, so as to obtain a charging load of each grid within a preset time period.
Example 3
According to an embodiment of the present invention, there is provided a computer-readable storage medium including a stored program, wherein when the program runs, a device in which the computer-readable storage medium is located is controlled to execute the processing method of the charging load.
Example 4
According to an embodiment of the present invention, a processor for running a program is provided, where the program runs to execute the processing method of the charging load.
The above-mentioned serial numbers of the embodiments of the present invention are merely for description and do not represent the merits of the embodiments.
In the above embodiments of the present invention, the descriptions of the respective embodiments have respective emphasis, and for parts that are not described in detail in a certain embodiment, reference may be made to related descriptions of other embodiments.
In the embodiments provided in the present application, it should be understood that the disclosed technology can be implemented in other ways. The above-described embodiments of the apparatus are merely illustrative, and for example, the division of the units may be a logical division, and in actual implementation, there may be another division, for example, multiple units or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, units or modules, and may be in an electrical or other form.
The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.
In addition, functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit. The integrated unit can be realized in a form of hardware, and can also be realized in a form of a software functional unit.
The integrated unit, if implemented in the form of a software functional unit and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a removable hard disk, a magnetic or optical disk, and other various media capable of storing program codes.
The foregoing is only a preferred embodiment of the present invention, and it should be noted that, for those skilled in the art, various modifications and decorations can be made without departing from the principle of the present invention, and these modifications and decorations should also be regarded as the protection scope of the present invention.
Claims (13)
1. A method for processing a charging load, comprising:
acquiring historical charging data;
dividing the historical charging data to obtain historical data corresponding to a plurality of grids;
and predicting historical data corresponding to each grid by using a load prediction model to obtain the charging load of each grid in a preset time period.
2. The method of claim 1, wherein the step of dividing the historical charging data to obtain historical data corresponding to a plurality of grids comprises:
acquiring a target area corresponding to the historical charging data;
dividing the target area according to longitude and latitude information to obtain a plurality of grids;
distributing the historical charging data to the grids to obtain historical data corresponding to each grid.
3. The method of claim 1, wherein the load prediction model is trained using a bayesian regularization algorithm on a back propagation neural network model.
4. The method of claim 3, further comprising:
acquiring a prediction index corresponding to the charging load, wherein the prediction index is used for representing factors influencing the charging load;
constructing a back propagation neural network model;
generating a training sample by using the prediction index;
and training the back propagation neural network model by using a Bayesian regularization algorithm and the training samples to obtain the load prediction model.
5. The method of claim 4, wherein obtaining the prediction index corresponding to the charging load comprises:
acquiring a plurality of target data corresponding to the charging load, wherein the plurality of target data comprise: operation mode, number of charging facilities, number of users, regional function, traffic flow and number of parking spaces;
processing the target data by using an analysis of variance method, and determining the importance degree of each target data;
and determining the prediction index based on the importance degree of each target data.
6. The method of claim 4, wherein the back propagation neural network model comprises: the device comprises an input layer, a hidden layer and an output layer, wherein the number of nodes of the input layer is determined based on the number of the prediction indexes, the number of nodes of the output layer is 1, and the number of nodes of the hidden layer is determined based on the number of nodes of the input layer and the number of nodes of the output layer by using a trial and error method.
7. The method of claim 4, wherein generating training samples using the predictor comprises:
normalizing the prediction index to obtain normalized data;
generating the training sample based on the normalized data.
8. The method of claim 4, wherein training the back propagation neural network model using a Bayesian regularization algorithm and the prediction index to obtain the load prediction model comprises:
inputting the training sample into the back propagation neural network model to obtain a prediction result of the training sample;
obtaining a mean square error based on the original result of the training sample and the prediction result of the training sample;
acquiring the square sum mean value of the network parameters of the back propagation neural network model;
obtaining a total error of the back propagation neural network model based on the mean square error and the mean square sum;
and updating the regularization parameters of the back propagation neural network model based on the total error to obtain the load prediction model.
9. The method of claim 8, wherein updating the regularization parameters of the back propagation neural network model based on the total error to obtain the load prediction model comprises:
judging whether the total error is larger than a preset error or not and whether the training times reach preset times or not;
if the total error is smaller than the preset error or the training times reach the preset times, determining that the training of the load prediction model is finished;
and if the total error is greater than the preset error and the training times do not reach the preset times, updating the regularization parameter based on a preset learning rate.
10. The method of claim 9, wherein updating the regularization parameter based on a preset learning rate comprises:
obtaining a Jacobian matrix of the total error;
solving the Jacobian matrix by using a Gauss-Newton approximation method to obtain the number of effective parameters;
and obtaining updated regularization parameters based on the mean square error, the square sum mean, the number of effective parameters and the preset learning rate.
11. A processing apparatus for a charging load, comprising:
the acquisition module is used for acquiring historical charging data;
the dividing module is used for dividing the historical charging data to obtain historical data corresponding to a plurality of grids;
and the prediction module is used for predicting the historical data corresponding to each grid by using the load prediction model to obtain the charging load of each grid in a preset time period.
12. A computer-readable storage medium, comprising a stored program, wherein when the program runs, the program controls a device where the computer-readable storage medium is located to execute the processing method of the charging load according to any one of claims 1 to 10.
13. A processor, configured to execute a program, wherein the program executes to perform the processing method of the charging load according to any one of claims 1 to 10.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011377051.8A CN112613637A (en) | 2020-11-30 | 2020-11-30 | Method and device for processing charging load |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011377051.8A CN112613637A (en) | 2020-11-30 | 2020-11-30 | Method and device for processing charging load |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112613637A true CN112613637A (en) | 2021-04-06 |
Family
ID=75228224
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011377051.8A Pending CN112613637A (en) | 2020-11-30 | 2020-11-30 | Method and device for processing charging load |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112613637A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| 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 |
| CN114140176A (en) * | 2022-01-30 | 2022-03-04 | 国网浙江电动汽车服务有限公司 | Adjustable capacity prediction method and device for load aggregation platform |
| CN116148700A (en) * | 2023-01-10 | 2023-05-23 | 中国第一汽车股份有限公司 | Method for predicting state of battery and storage medium |
| CN116613867A (en) * | 2023-07-20 | 2023-08-18 | 西安峰盛智能电子有限公司 | Wireless power transmission system for AGV and control method thereof |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109409614A (en) * | 2018-11-16 | 2019-03-01 | 国网浙江瑞安市供电有限责任公司 | A kind of Methods of electric load forecasting based on BR neural network |
| CN109978241A (en) * | 2019-03-11 | 2019-07-05 | 清华四川能源互联网研究院 | A kind of determination method and device of electric car charging load |
| CN110889527A (en) * | 2018-09-07 | 2020-03-17 | 国网甘肃省电力公司电力科学研究院 | Electric vehicle charging load prediction method based on LSTM neural network |
| WO2020082973A1 (en) * | 2018-10-22 | 2020-04-30 | 新智数字科技有限公司 | Load prediction method and apparatus based on neural network |
| CN111476441A (en) * | 2020-05-29 | 2020-07-31 | 南方电网科学研究院有限责任公司 | Load prediction method for electric vehicle charging equipment and related device |
-
2020
- 2020-11-30 CN CN202011377051.8A patent/CN112613637A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110889527A (en) * | 2018-09-07 | 2020-03-17 | 国网甘肃省电力公司电力科学研究院 | Electric vehicle charging load prediction method based on LSTM neural network |
| WO2020082973A1 (en) * | 2018-10-22 | 2020-04-30 | 新智数字科技有限公司 | Load prediction method and apparatus based on neural network |
| CN109409614A (en) * | 2018-11-16 | 2019-03-01 | 国网浙江瑞安市供电有限责任公司 | A kind of Methods of electric load forecasting based on BR neural network |
| CN109978241A (en) * | 2019-03-11 | 2019-07-05 | 清华四川能源互联网研究院 | A kind of determination method and device of electric car charging load |
| CN111476441A (en) * | 2020-05-29 | 2020-07-31 | 南方电网科学研究院有限责任公司 | Load prediction method for electric vehicle charging equipment and related device |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| 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 |
| CN114140176A (en) * | 2022-01-30 | 2022-03-04 | 国网浙江电动汽车服务有限公司 | Adjustable capacity prediction method and device for load aggregation platform |
| CN114140176B (en) * | 2022-01-30 | 2022-04-26 | 国网浙江电动汽车服务有限公司 | Adjustable capacity prediction method and device for load aggregation platform |
| CN116148700A (en) * | 2023-01-10 | 2023-05-23 | 中国第一汽车股份有限公司 | Method for predicting state of battery and storage medium |
| CN116148700B (en) * | 2023-01-10 | 2024-04-12 | 中国第一汽车股份有限公司 | Method for predicting state of battery and storage medium |
| CN116613867A (en) * | 2023-07-20 | 2023-08-18 | 西安峰盛智能电子有限公司 | Wireless power transmission system for AGV and control method thereof |
| CN116613867B (en) * | 2023-07-20 | 2023-12-26 | 上海木链工业互联网科技有限公司 | Wireless power transmission system for AGV and control method thereof |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112613637A (en) | Method and device for processing charging load | |
| Babel et al. | A multivariate econometric approach for domestic water demand modeling: an application to Kathmandu, Nepal | |
| Bhat et al. | A comparison of two alternative behavioral choice mechanisms for household auto ownership decisions | |
| Pfaffenbichler et al. | The integrated dynamic land use and transport model MARS | |
| CN109858706A (en) | A kind of project cost data management system and method | |
| CN108985570A (en) | A kind of load forecasting method and its system | |
| CN106063067A (en) | Electricity-demand prediction device, electricity supply system, electricity-demand prediction method, and program | |
| Ajitha et al. | Design and development of Residential Sector Load Prediction model during COVID-19 Pandemic using LSTM based RNN | |
| Jeyaraj et al. | Computer‐assisted demand‐side energy management in residential smart grid employing novel pooling deep learning algorithm | |
| Cullinan | A spatial microsimulation approach to estimating the total number and economic value of site visits in travel cost modelling | |
| CN116205329A (en) | Holiday passenger flow prediction method | |
| CN116934366A (en) | Charging pricing method and device for charging station, storage medium and computer equipment | |
| CN112508734B (en) | Method and device for predicting power generation capacity of power enterprise based on convolutional neural network | |
| Khan et al. | Comparison of electric vehicle load forecasting across different spatial levels with incorporated uncertainty estimation | |
| JPH05314094A (en) | Sales prediction supporting device | |
| CN111967686B (en) | Random load prediction method based on power utilization probability distribution function | |
| CN113435663A (en) | CNN-LSTM combined load prediction method considering electric vehicle charging load influence | |
| CN117495056A (en) | Power consumption data monitoring and optimizing method and system | |
| Alhmoud et al. | Short-term load forecasting for Jordan’s power system using neural network based different | |
| Helmus et al. | SEVA: A data driven model of electric vehicle charging behavior | |
| Carvalho et al. | A data-driven model to evaluate the medium-term effect of contingent pricing policies on residential water demand | |
| Kim et al. | Spatial-Temporal Graph Convolutional-Based Recurrent Network for Electric Vehicle Charging Stations Demand Forecasting in Energy Market | |
| Tabassum et al. | Short Term Load Forecasting of Residential and Commercial Consumers of Karnataka Electricity Board using CFNN | |
| CN111325413B (en) | Short-term load interval prediction method based on neural network and proportionality coefficient method | |
| CN114912760B (en) | Method, system and medium for assigning take-out packaging garbage population weight |
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 |