A low-data-quality scenario-based electric vehicle charging load prediction method

Abstract

The application discloses a low-data-quality scenario electric vehicle charging load prediction method, and relates to the technical field of electric vehicles.The method comprises the following steps: acquiring first features; setting dynamic space-time weight factors according to the first features; complementing missing values in a space-time correlated charging power matrix according to the dynamic space-time weight factors to obtain complete target charging power matrix data; constructing a three-dimensional input tensor according to the complete target charging power matrix data; inputting the three-dimensional input tensor into a first model to obtain a charging load prediction value; performing trend checking on the charging load prediction value and historical charging load data to obtain a prediction trend; and if the difference between the prediction trend and a historical trend exceeds a preset threshold, correcting the charging load prediction value based on the change rate of the historical trend to obtain a target electric vehicle charging load prediction result.The method can realize high-precision load prediction and provide strong support for power grid demand response.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of electric vehicles, in particular to an electric vehicle charging load prediction method in a low data quality scenario. BACKGROUND

[0002] As a green and environmentally friendly means of transportation, the number of electric vehicles is growing explosively, but their large-scale access to the power grid poses challenges to stable operation and demand response. Accurate prediction of short-term charging load at the hourly level within the next day is crucial for power grid optimization and dispatch, effective demand response strategy development, and ensuring the safe and stable operation of the power grid. However, in practice, charging load data often has gaps due to charging pile failures, communication interruptions, storage problems, and human operations, and poor data quality seriously affects prediction accuracy, thereby hindering the implementation of demand response strategies. How to achieve short-term charging load prediction in a low data quality scenario has become an important research topic in the field of electric power systems and electric vehicles.

[0003] Current research on electric vehicle charging load prediction has made some progress. Traditional prediction methods mainly include time series analysis methods and machine learning methods (such as support vector machines). In recent years, deep learning methods have been widely used in this field. Long short-term memory networks (LSTM) and convolutional neural networks (CNN) are commonly used to process complex time series data. To further extract long-term dependencies, a new deep learning architecture called Informer is introduced. Its sparse improvement based on the self-attention mechanism can effectively capture long-term dependencies and reduce computational complexity. At the same time, through the probabilistic sparse self-attention mechanism, it reduces the processing of redundant information, achieving a good balance between prediction accuracy and time complexity, and is suitable for large-scale and complex charging load data scenarios.

[0004] Traditional time series analysis methods require high data integrity and stationarity, making it difficult to effectively handle data gaps. Traditional machine learning methods can handle data gaps to some extent, but when faced with large-scale and complex charging load data, they lack prediction accuracy and generalization ability. SUMMARY

[0005] The application provides an electric vehicle charging load prediction method in a low data quality scenario, which can achieve high-precision load prediction and provide strong support for power grid demand response.

[0006] To achieve the above purpose, the application adopts the following technical solutions:

[0007] In a first aspect, the application provides an electric vehicle charging load prediction method in a low data quality scenario, which comprises:

[0008] Obtaining a first feature that affects charging load;

[0009] setting a dynamic spatiotemporal weight factor according to the first characteristic of the influence on the charging load;

[0010] constructing a spatiotemporally correlated charging power matrix;

[0011] completing missing values in the spatiotemporally correlated charging power matrix according to the dynamic spatiotemporal weight factor to obtain complete target charging power matrix data;

[0012] constructing a three-dimensional input tensor according to the complete target charging power matrix data;

[0013] inputting the three-dimensional input tensor into a first model to obtain a charging load prediction value;

[0014] performing trend verification on the charging load prediction value and historical charging load data to obtain a prediction trend;

[0015] if a difference between the prediction trend and a historical trend exceeds a preset threshold, performing linear correction on the charging load prediction value based on a change rate of the historical trend to obtain a target electric vehicle charging load prediction result.

[0016] In some possible implementation manners, the first characteristic of the influence on the charging load includes:

[0017] obtaining first charging data from a plurality of electric vehicle charging stations and charging piles;

[0018] obtaining multi-source correlated influence data;

[0019] identifying abnormal values in the first charging data and the multi-source correlated influence data by using an abnormal value detection algorithm, and removing the abnormal values;

[0020] performing encoding conversion on text type correlated influence data in the first charging data and the multi-source correlated influence data after removing the abnormal values, and unifying a numerical value range to construct a first numerical feature set;

[0021] constructing a second numerical feature set according to non-text type correlated influence data in the first charging data and the multi-source correlated influence data after removing the abnormal values;

[0022] performing screening on the first numerical feature set and the second numerical feature set by using feature correlation analysis to obtain the first characteristic of the influence on the charging load.

[0023] In some possible implementation manners, the completing of the missing values in the spatiotemporally correlated charging power matrix according to the dynamic spatiotemporal weight factor to obtain the complete target charging power matrix data includes:

[0024] constructing a low-rank matrix completion optimization model based on the dynamic spatiotemporal weight factor;

[0025] Solve the low-rank matrix completion optimization model by using a singular value threshold algorithm to obtain a target value, replace the missing values in the space-time related charging power matrix with the target value, and obtain a completed charging power matrix;

[0026] Verify the completed charging power matrix to obtain a result that passes the verification, and use the completed charging power matrix as target charging power matrix data.

[0027] In some possible implementations, the method further includes:

[0028] building an initial neural network model;

[0029] optimizing the initial neural network model by using a flower pollination algorithm to obtain a first model.

[0030] In some possible implementations, the method further includes:

[0031] building a full-scenario multi-dimensional evaluation system including an error index, a scenario adaptability index, a robustness index, and an economic index;

[0032] determining the weights of the evaluation indexes by using an analytic hierarchy process;

[0033] obtaining actual scores under each evaluation layer index according to the target electric vehicle charging load prediction result;

[0034] obtaining a comprehensive evaluation result according to the weights of the evaluation indexes and the actual scores under each evaluation layer index;

[0035] if the comprehensive evaluation result reaches a preset standard, formulating a demand response strategy according to the target electric vehicle charging load prediction result.

[0036] In some possible implementations, the demand response strategy is formulated according to the target electric vehicle charging load prediction result, including:

[0037] constructing an early warning level according to the target electric vehicle charging load prediction result and a maximum bearable load threshold of a power grid, and dynamically adjusting the division standard of the early warning level according to a proportion of a predicted load in the target electric vehicle charging load prediction result that exceeds the maximum bearable load threshold of the power grid and a load difference change rate;

[0038] performing user stratification in combination with a user portrait and a user response behavior prediction model to obtain a user level; the user portrait includes a user charging frequency, a charging time period preference, a charging power demand, a price sensitivity degree, and a historical response record, and the user response behavior prediction model predicts a response probability of each level user under different incentive strategies based on historical data;

[0039] The incentive intensity is dynamically adjusted according to the response probability of the user, and a demand response strategy is formulated according to the incentive intensity, the early warning level and the user level.

[0040] In some possible implementation manners, the dynamic space-time weight factor comprises a region weight and a time period weight; the region weight is dynamically adjusted according to a historical load contribution degree and an operation state of a corresponding region charging station in the first feature of the influence on the charging load, and the time period weight is dynamically configured according to a charging load fluctuation feature in different time periods and a real-time power demand change in the first feature of the influence on the charging load.

[0041] In a second aspect, the present application provides a device for predicting charging load of an electric vehicle in a low data quality scenario, the device comprising:

[0042] a first feature of the influence on the charging load is obtained;

[0043] a completion module configured to set a dynamic space-time weight factor according to the first feature of the influence on the charging load, construct a space-time correlated charging power matrix, and complete missing values in the space-time correlated charging power matrix according to the dynamic space-time weight factor to obtain complete target charging power matrix data;

[0044] a prediction module configured to construct a three-dimensional input tensor according to the complete target charging power matrix data, input the three-dimensional input tensor into a first model to obtain a charging load prediction value, perform trend verification on the charging load prediction value and historical charging load data to obtain a prediction trend, and perform linear correction on the charging load prediction value based on a change rate of a historical trend if a difference between the prediction trend and the historical trend exceeds a preset threshold to obtain a target electric vehicle charging load prediction result.

[0045] In a third aspect, the present application provides a computing device comprising a memory and a processor;

[0046] One or more computer programs are stored in the memory, and the one or more computer programs comprise instructions; when the instructions are executed by the processor, the computing device performs the method according to any one of the first aspect.

[0047] In a fourth aspect, the present application provides a computer readable storage medium for storing a computer program, the computer program being used to perform the method according to any one of the first aspect.

[0048] In a fifth aspect, the present application provides a computer program product comprising one or more computer instructions, when the computer instructions are executed by a computer, the computer performs the method according to any one of the first aspect.

[0049] From the above technical solutions, the present application has at least the following beneficial effects:

[0050] In the present application, the first feature affecting the charging load is obtained; the dynamic space-time weight factor is set according to the first feature affecting the charging load; the space-time associated charging power matrix is constructed; the missing values in the space-time associated charging power matrix are completed according to the dynamic space-time weight factor, and the complete target charging power matrix data is obtained; the three-dimensional input tensor is constructed according to the complete target charging power matrix data; the three-dimensional input tensor is input into the first model to obtain the charging load prediction value; the charging load prediction value is trend checked with the historical charging load data to obtain the prediction trend; if the difference between the prediction trend and the historical trend exceeds the preset threshold, the charging load prediction value is linearly corrected based on the change rate of the historical trend to obtain the target electric vehicle charging load prediction result.

[0051] In the prior art, a simple interpolation method or a single model is often used to directly predict the charging load, without fully considering the space-time dynamic change characteristics of the charging load related features, lacking pertinence in completing the missing values of the charging power matrix, and lacking the verification and correction link of the prediction result and the historical trend, resulting in limited prediction result accuracy, which is difficult to reflect the actual charging load change rule, and brings adverse effects to the subsequent power grid demand response work based on the prediction result. It can be seen that, by obtaining the first feature affecting the charging load and setting the dynamic space-time weight factor, the present application can quickly capture the space-time change rule of the charging load, thereby completing the missing values of the charging power matrix and constructing a more actual three-dimensional input tensor; in combination with the first model prediction and the verification and correction of the prediction trend and the historical trend, the accuracy and reliability of the charging load prediction result are guaranteed in multiple dimensions, which provides strong data support for the efficient development of demand response, optimization of resource allocation and the like of the power grid.

[0052] It should be understood that the description of technical features, technical solutions, beneficial effects or similar language in the present application does not imply that all features and advantages can be achieved in any single embodiment. On the contrary, it can be understood that the description of a feature or a beneficial effect means that the specific technical feature, technical solution or beneficial effect is included in at least one embodiment. Therefore, the description of technical features, technical solutions or beneficial effects in the present specification does not necessarily refer to the same embodiment. Further, the technical features, technical solutions and beneficial effects described in the present embodiment can be combined in any appropriate manner. Those skilled in the art will understand that the embodiments can be implemented without one or more specific technical features, technical solutions or beneficial effects of a particular embodiment. In other embodiments, additional technical features and beneficial effects can be identified in specific embodiments that do not embody all embodiments. BRIEF DESCRIPTION OF DRAWINGS

[0053] Figure 1 A flowchart of a low-data-quality scenario electric vehicle charging load prediction method provided for an embodiment of the present application is shown in FIG. 1.

[0054] Figure 2 A schematic diagram of a prediction result of an electric vehicle charging load provided for an embodiment of the present application is shown in FIG. 2.

[0055] Figure 3 A schematic diagram of a prediction error result provided for an embodiment of the present application is shown in FIG. 3.

[0056] Figure 4 A schematic diagram of a low-data-quality scenario electric vehicle charging load prediction device provided for an embodiment of the present application is shown in FIG. 4.

[0057] Figure 5 A schematic diagram of a computing device provided for an embodiment of the present application is shown in FIG. 5. DETAILED DESCRIPTION

[0058] The terms "first", "second", and "third" and the like in the specification and the drawings of the present application are used to distinguish different objects, and are not used to limit a particular order.

[0059] In the embodiments of the present application, the words "exemplary" or "for example" are used to mean serving as an example, instance, or illustration, in no way limiting. Any embodiment or design scheme described as "exemplary" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or advantageous than other embodiments or design schemes. Rather, the use of the words "exemplary" or "for example" is intended to present relevant concepts in a specific manner.

[0060] For the sake of clear and concise description of the following embodiments, first, a brief introduction of related technologies is given:

[0061] As a green and environmentally friendly means of transportation, the number of electric vehicles is growing explosively. However, the large-scale access of electric vehicles also brings new challenges to the power grid, especially the uncertainty of charging load, which puts higher requirements on the stable operation and demand response of the power grid. Accurate prediction of short-term charging load of electric vehicles, especially the hourly charging load distribution in the future one day, is the key to power grid optimization scheduling and demand response strategy making, and is one of the factors to ensure the safe and stable operation of the power grid.

[0062] In practical application scenarios, electric vehicle charging load data often faces quality problems, and data vacancy is a typical case. The reasons for data vacancy are various: it may be charging pile failure, it may be communication interruption, such as communication link failure between charging pile and data server, resulting in part of the data unable to upload, it may also be data storage problem, and it may also be affected by human factors. Such data quality problems will seriously reduce the accuracy of electric vehicle charging load prediction, and thus hinder the effective formulation and landing of power grid demand response strategies. Therefore, how to accurately predict the short-term charging load of electric vehicles under the condition of poor data quality has become an important research direction in the field of electric power system and electric vehicles.

[0063] Currently, electric vehicle charging load prediction research has made certain progress. Traditional prediction methods are mainly divided into time series analysis methods and machine learning methods (such as support vector machine), but these methods have obvious limitations in charging load prediction under low data quality scenarios: time series analysis methods have high requirements for data integrity and stationarity, and are difficult to effectively handle data vacancy problems; traditional machine learning methods can deal with data vacancy to some extent, but when faced with large-scale and complex electric vehicle charging load data, their prediction accuracy and generalization ability still have room for improvement.

[0064] In recent years, deep learning methods have been widely applied in this field, and long short-term memory (LSTM) and convolutional neural network (CNN) are often used to process complex time series data. However, this kind of method cannot effectively extract the long-term dependence relationship of load data. To solve this problem, a new deep learning architecture, Informer, is introduced. Based on the sparsity improvement of self-attention mechanism, it can effectively capture the dependence relationship of long time span and greatly reduce the computational complexity. In addition, Informer reduces the demand for processing redundant information through probabilistic sparse self-attention mechanism, further improving the efficiency and performance of the model. Compared with traditional deep learning methods, Informer achieves a better balance between prediction accuracy and time complexity, and is very suitable for large-scale and complex electric vehicle charging load data scenarios. However, the hyperparameter setting of Informer depends on human experience, which can easily lead to unstable prediction performance.

[0065] In view of this, the embodiment of the present application provides a low data quality scene electric vehicle charging load prediction method. In the method, first, the first feature affecting the charging load is obtained; the dynamic space-time weight factor is set according to the first feature affecting the charging load; the space-time associated charging power matrix is constructed; the missing values in the space-time associated charging power matrix are completed according to the dynamic space-time weight factor, and the complete target charging power matrix data is obtained; the three-dimensional input tensor is constructed according to the complete target charging power matrix data; the three-dimensional input tensor is input into the first model to obtain the charging load prediction value; the charging load prediction value is trend checked with the historical charging load data to obtain the prediction trend; if the difference between the prediction trend and the historical trend exceeds the preset threshold, the charging load prediction value is linearly corrected based on the change rate of the historical trend to obtain the target electric vehicle charging load prediction result. The first feature affecting the charging load is obtained and the dynamic space-time weight factor is set, so that the space-time variation law of the charging load can be more accurately captured, the missing values of the charging power matrix are completed, and a three-dimensional input tensor more consistent with the actual situation is constructed. In combination with the first model prediction and the checking and correction of the prediction trend and the historical trend, the accuracy and reliability of the charging load prediction result are guaranteed in multiple dimensions, which provides strong data support for efficient demand response, optimal resource allocation and the like of the power grid.

[0066] In order to make the technical solutions of the present application clearer and easier to understand, the low data quality scene electric vehicle charging load prediction method provided by the embodiment of the present application will be introduced below in combination with the drawings. As shown in the figure, the figure is a flow chart of the low data quality scene electric vehicle charging load prediction method provided by the embodiment of the present application. Figure 1

[0067] In the embodiment of the present application, the method is applied to a processing device. The low data quality scene electric vehicle charging load prediction method comprises:

[0068] S101, obtaining the first feature affecting the charging load.

[0069] ​Firstly, multi-source data collection is carried out. From multiple electric vehicle charging stations and charging piles, with the help of intelligent sensing devices and communication modules built-in the charging piles, real-time first charging data is obtained. The first charging data here covers key information such as charging power, charging start time, charging end time, and charging power, which can reflect the electric energy parameters in the electric vehicle charging process. Then, through the data platform of the meteorological department, the holiday management system, and the local power department price policy release channel, multi-source associated influence data is obtained, including weather conditions (such as temperature, humidity, and light intensity, different weather will affect user travel and charging demand), holiday information (holiday users will change their travel mode, and charging behavior will also change accordingly), and local electricity price policy (time-of-use electricity price policy will guide users to adjust the charging time) and other data.

[0070] Then, data preprocessing operation is carried out. Anomaly value detection algorithm is used to identify the anomaly values in the first charging data and multi-source associated influence data, and the anomaly values are excluded, including: using anomaly value detection algorithm, such as 3σ principle detection algorithm based on statistics or isolation forest algorithm, to identify the anomaly values in the first charging data and multi-source associated influence data. The cause of the anomaly value may be equipment failure (such as charging pile sensor failure leading to abnormal charging power data), communication interference (error in data transmission process) and the like. The identified anomaly values are excluded to ensure the reliability of subsequent data.

[0071] The text type associated influence data in the first charging data and multi-source associated influence data after excluding the anomaly values is encoded and converted, and the numerical value range is unified to construct the first numerical feature set; for example, holiday information and the like is encoded and converted by using label encoding or one-hot encoding, and the text information is converted into a numerical form recognizable by a computer. At the same time, all data is normalized or standardized, and the numerical value range is unified, for example, the data is mapped to the interval [0, 1], thereby constructing the first numerical feature set.

[0072] And for the non-text type associated influence data in the first charging data and multi-source associated influence data after excluding the anomaly values, such as charging power, temperature and the like, which are numerical data, are directly arranged to construct the second numerical feature set.

[0073] Finally, the feature screening work is carried out. Through feature correlation analysis methods such as Pearson correlation coefficient analysis or mutual information method, the first numerical feature set and the second numerical feature set are screened. The correlation degree between each feature and the charging load is calculated, the features with high correlation with the charging load are retained, and the redundant features with low correlation and small contribution to the charging load prediction are removed, so as to obtain the first features that affect the charging load. Among them, high correlation means that the absolute value of the Pearson correlation coefficient is greater than or equal to the correlation threshold, and low correlation means that the absolute value of the Pearson correlation coefficient is less than the correlation threshold. In this application, the correlation threshold can be set to 0.5; if the absolute value of the Pearson correlation coefficient between a feature and the charging load is less than 0.5 (i.e. low correlation), it means that the linear correlation between the feature and the charging load is weak, and the feature cannot provide sufficient effective information reflecting the change rule of the charging load for the prediction model, and the positive help for the prediction of the charging load is small, which belongs to the case of small contribution; even if the absolute value of the Pearson correlation coefficient between a feature and the charging load is not less than 0.5 (not extremely low correlation), but if it cannot provide additional effective information for the prediction model, such as the time period ID (a simple sequence number for the time period), the association between the feature and the charging load only comes from the mapping of the time period, and the independent contribution of the feature to the prediction is extremely low, which also belongs to the case of small contribution. These screened features will be faster used for subsequent construction of the charging load prediction model.

[0074] S102, setting a dynamic space-time weight factor according to the first features affecting the charging load.

[0075] The dynamic space-time weight factor includes a region weight and a time period weight.

[0076] For the region weight, the historical load data of each regional charging station is extracted from the first features affecting the charging load, and the historical load contribution degree is calculated, that is, the proportion of the charging load of the regional charging station in the overall load in the past period of time. Combined with the operation state of each regional charging station, such as the number of available charging piles, the health degree of the equipment (judged by equipment failure records, maintenance period, etc.), and other factors.

[0077] If the historical load contribution degree of a regional charging station is high and the operation state is good, it can stably provide charging services, so the region weight is increased; on the contrary, if the historical contribution degree is low or the operation state is poor (such as frequent charging pile failures), the region weight is reduced.

[0078] In this way, the region weight is dynamically adjusted, and the influence degree of different regions of the charging station on the overall charging load can be reflected. For the time period weight, according to the charging load fluctuation characteristics of different time periods in the first feature of the influence on the charging load, the change law of the charging load of each time period (such as early peak, midday, late peak, night, etc.) in history is analyzed, such as the charging load is usually high and fluctuates greatly in the late peak period. Combined with the real-time power demand change, the time period weight is dynamically configured by monitoring the power load of the power grid and the real-time data of the user charging request. In the time period of high charging load and tight real-time power demand, the time period weight is increased to highlight the influence of the charging load on the whole; in the time period of low charging load and loose real-time power demand, the time period weight is reduced.

[0079] By setting the dynamic space-time weight factor, the importance of different regions and different time periods in subsequent analysis and prediction can be dynamically adjusted according to the actual influencing factors of the charging load, so that the subsequent processing based on these weights is more in line with the actual charging load change.

[0080] Compared with the fixed weight method, dynamically adjusting the region weight and the time period weight can more flexibly and accurately reflect the real influence of different regions and time periods on the charging load, improve the description accuracy of the influencing factors of the charging load, and provide more suitable weight basis for the subsequent steps, thereby improving the accuracy of the subsequent charging load related processing.

[0081] S103, constructing a charging power matrix associated with space and time.

[0082] Firstly, the original charging power data of each region charging station in different time periods is collected. Then, the dynamic space-time weight factor set in S102 is used to perform weighted processing on the original data. For the charging power data of each region and each time period, the corresponding region weight and time period weight are multiplied to strengthen the effect of the charging power data of the region and time period that has a large influence on the charging load.

[0083] After that, the charging power data of each region and each time period after the weighted processing is integrated to construct a two-dimensional charging power matrix according to the dimensions of the region and the time period. Among them, the rows of the matrix can represent different regions, the columns can represent different time periods, and each element in the matrix can represent the charging power value of the corresponding region in the corresponding time period after the space-time weight adjustment. The space-time associated charging power matrix constructed in this way can simultaneously reflect the space (region) and time (time period) correlation characteristics of the charging load.

[0084] The constructed space-time related charging power matrix integrates the space-time related characteristics of the charging load in one matrix, provides a basic data structure containing space-time related information for subsequent steps of charging power matrix missing value completion, charging load prediction and the like, and facilitates subsequent models to learn and mine the space-time rules of the charging load.

[0085] The constructed space-time related charging power matrix integrates the information in the space-time dimension, so that the correlation in space and time can be considered simultaneously when subsequent charging power data is processed, which helps to more comprehensively grasp the change rule of the charging load, lays a good data foundation for subsequent charging load prediction and the like, and thus improves the accuracy and reliability of the charging load prediction.

[0086] S104, according to the dynamic space-time weight factor, the missing values in the space-time related charging power matrix are completed to obtain complete target charging power matrix data.

[0087] A low-rank matrix completion optimization model based on the dynamic space-time weight factor is constructed, a singular value threshold algorithm is used to solve the low-rank matrix completion optimization model to obtain a target value, the target value is replaced in the missing values in the space-time related charging power matrix to obtain a completed charging power matrix. The completed charging power matrix is verified to obtain a result of passing the verification, and the completed charging power matrix is taken as the target charging power matrix data. The specific steps and the formula involved are as follows:

[0088] First, a low-rank matrix completion optimization model based on a dynamic space-time weight factor is constructed. Let the space-time related charging power matrix be (M is the space-time related charging power matrix, m is the number of regions, n is the number of time periods, R is a mathematical standard symbol, and is used to indicate that the elements in the matrix or vector discussed are real numbers), wherein there are missing values in the charging power matrix, and the index set of the known elements in the space-time related charging power matrix is denoted as , the index set of the missing elements in the space-time related charging power matrix is denoted as .

[0089] In the dynamic space-time weight factor, the full-region weight is , and the full-time period weight is , so the low-rank matrix completion optimization model integrating the dynamic space-time weight is:

[0090]

[0091]

[0092] , wherein represents finding a matrix X such that the rank of the matrix X is as small as possible, is the low-rank matrix to be completed, is the element in the i-th row and j-th column of X, i.e., the element in the position of the matrix to be completed, is the spatiotemporal correlated charging power matrix is the known element in the i-th row and j-th column of X, represents the rank of the matrix, embodying the low-rank constraint; is the regularization parameter balancing the low-rank constraint and the data fitting term; is the region weight corresponding to region i, i is an integer, and , is the time period weight corresponding to time period j, j is an integer, and , and fuses the dynamic spatiotemporal weight, so that the fitting at the known element is more focused on the regions and time periods with high weight, and T is the transpose.

[0093] Then, the singular value threshold algorithm is used to solve the low-rank matrix completion optimization model. Singular value decomposition is performed on the matrix , and the specific formula is as follows:

[0094]

[0095]

[0096] wherein, is the first orthogonal matrix, is the second orthogonal matrix, is the transpose of the second orthogonal matrix, is a diagonal matrix, the elements on the diagonal of which are the singular values of the low-rank matrix X to be completed, and these singular values are arranged in descending order, is a specific description of the diagonal matrix, , ,···, is the singular value of X, and k is the effective rank related dimension of the matrix, i.e., the number of non-zero singular values of X, which is closely related to the rank of the matrix, embodying the effective information dimension of the matrix, .

[0097] The singular value threshold algorithm realizes the low-rank approximation by threshold processing of the singular values, i.e.,

[0098]

[0099] wherein, is a diagonal matrix obtained after threshold processing of the diagonal matrix, is a threshold parameter, which is determined by iterative optimization.

[0100] Solving for the completed matrix yields the solution. The completed matrix Chinese correspondence The element value at the index is the target value, and the calculation formula is:

[0101]

[0102] Will Chinese correspondence Element value at index Replace the missing values ​​in the spatiotemporally correlated charging power matrix M to obtain the initially completed charging power matrix M'. The elements in the charging power matrix M' are... The calculation formula is:

[0103]

[0104] Finally, the completed charging power matrix is ​​validated. A portion of the known data (data without missing values) from the completed matrix is ​​selected, and the mean square error is calculated. The formula is as follows:

[0105]

[0106] If MSE is less than the preset threshold If the completion is satisfactory, the model parameters are considered to meet the requirements; otherwise, the model parameters are adjusted (e.g., ...). , (etc.) Resolve. After verification, the complete target charging power matrix data is obtained.

[0107] By constructing a low-rank matrix completion optimization model that incorporates dynamic spatiotemporal weighting factors, and using the singular value thresholding algorithm to solve the problem and combining it with mean square error verification, the missing values ​​of the spatiotemporal correlated charging power matrix are completed, providing complete and high-quality charging power data support for subsequent charging load prediction and other processes.

[0108] S105. Construct a three-dimensional input tensor based on the complete target charging power matrix data.

[0109] First, let's clarify that the complete target charging power matrix is ​​denoted as... This is a matrix describing the charging power in different regions and time periods, encompassing charging power information for m regions (m being the number of regions) and n time periods (n being the number of time periods). To allow this matrix to be processed by the subsequent first model, it needs to be constructed as a three-dimensional input tensor. The first step is to... Expand the dimensions.

[0110] Because in the charging load prediction, not only the current area, the current period of charging power will have an impact, the historical time of charging power will also play a role in the prediction result due to the existence of time sequence correlation. Therefore, the application selects the charging power data of T1 continuous time windows, such as selecting the charging power data of the past T1 hours as the time window.

[0111] Next, for each area i (i takes values ) and each period j (j takes values ), collect their charging power data in these continuous T1 time windows, denoted as (where t takes values ).

[0112] Finally, based on these collected data, a three-dimensional input tensor is constructed. The three-dimensional input tensor is a tensor with m area dimensions, n period dimensions and T1 time window dimensions. Each element in the three-dimensional input tensor corresponds to the charging power data of area i, period j and the first T1 time window. The three-dimensional input tensor constructed in this way can simultaneously reflect the information of charging load in the area, period and time sequence dimensions, providing the necessary data basis for the subsequent model to capture the space-time and time sequence correlation of the charging load.

[0113] S106, input the three-dimensional input tensor into the first model to obtain the charging load prediction value.

[0114] An initial neural network model is constructed, and the initial neural network model is optimized by adopting the flower pollination algorithm to obtain the first model, specifically including:

[0115] The initial neural network model adopts a network architecture based on an attention mechanism, which is an Informer model. The core is to capture the long-term dependence relationship of the charging load data through a self-attention mechanism.

[0116] The overall architecture of the Informer model is composed of an encoder and a decoder. The model input covers charging data such as charging power, charging start time, charging end time, and charging capacity, and also collects external influencing factor data related to electric vehicle load, such as weather conditions (including temperature, humidity, wind speed, and other specific meteorological indicators), holiday arrangement information, and local electricity price policy, etc.

[0117] Both the encoder and the decoder are composed of multiple neural network modules, which include multi-head probability sparse self-attention layers and convolution distillation layers. The model first encodes the input time series data, and then extracts deep features by multiple neural network modules composed of multi-head probability sparse self-attention layers and convolution layers. In the decoder, the variable information related to the electric vehicle charging load is input into the model, which is operated with the deep feature maps extracted by the encoder to capture the correlation between the features. Finally, the prediction result of the electric vehicle charging load value is obtained through the full connection layer.

[0118] In the multi-head probability sparse self-attention layer, the query matrix (Q) is used to reduce redundant calculations, so that the time complexity is reduced from O(L 2 ) of the traditional self-attention to O(LlnL). The formula is:

[0119]

[0120] wherein, is the filtered query matrix, K is the key matrix, is the transpose of the key matrix, V is the value matrix, and L represents the sequence length. is the dimension of the key matrix K, the function is used for normalization of The initial neural network model structure includes the encoder and the decoder part. The encoder extracts features from the input three-dimensional tensor, and the decoder generates the charging load prediction result based on the extracted features.

[0121] In the overall architecture of the Informer, the encoder uses sparse self-attention instead of the traditional self-attention mechanism. This method can reduce the size of the network and increase the robustness of the model by layer-by-layer stacking. The decoder receives long sequence input, fills the target element with zero, and combines the feature maps through weighted attention to quickly output the element in a generated manner.

[0122] The main function of the encoder is to extract the long-distance correlation and deep features of the input sequence. In the probability sparse self-attention calculation process, the query vector with low attention value is directly averaged. In order to make the model structure pay more attention to the dominant attention in the deep feature extraction process and highlight its role, we use convolution and pooling operations to distill the self-attention feature map. In this way, the dominant features with dominant role can be extracted with privilege. The feature transmission process from the rth layer to the r+1th layer is as follows:

[0123]

[0124] In the formula, is the output feature of the r+1th layer at the tth time. is the input feature of the r-th layer at the t-th time, Conv is a one-dimensional convolution operation used to extract local features; is an exponential linear unit, and MaxPool is a max-pooling operation. is an attention block.

[0125] In the decoder part, a standard decoder structure is adopted. It consists of two identical multi-head attention layers. This structure is used to alleviate the problem of speed decline in long sequence generation inference process. First, the hidden multi-head attention layer (Masked Multi-Head Attention) is used to avoid the leakage of future information in autoregressive prediction. Then, the deep features output by the encoder are spliced with the input features of the decoder. The splicing formula is:

[0126]

[0127] In the formula: is the spliced feature input to the decoder at the t-th time, is the feature related to the token (token can be understood as a unit used to identify or convey specific information) at the t-th time, is the input feature at the t-th time, except the spliced feature and the feature related to the token, is the length parameter related to the token, which is used to represent the sequence length of the token-related feature, is the length parameter of the feature related to the token (such as the target-related feature), is a real number space with a dimension of , i.e. the space dimension of the spliced feature vector, and Concat is the splicing operation, which splices the features of the encoder and the decoder together. is the dimension of the model, which represents the dimension of the feature vector.

[0128] In the probability sparse self-attention, a hidden multi-head attention mechanism (masked multi-head attention) is adopted. This mechanism avoids the autoregressive problem of paying attention to the next position at each position. Finally, the preliminary charging load prediction result is output through a fully connected layer.

[0129] Secondly, in order to improve the prediction accuracy and generalization ability of the Informer deep neural network in short-term electric vehicle charging load prediction, the Flower Pollination Algorithm (FPA) is introduced to optimize the key hyperparameters of the model. First, the hyperparameters of the Informer model (including the number of attention heads, the number of encoder layers, the number of decoder layers, and the learning rate) are encoded as individual position vectors of the Flower Pollination Algorithm, and each vector corresponds to a set of parameter combinations to be evaluated. Then, the population is randomly initialized to be uniformly distributed within the parameter search space. During the optimization process, the algorithm performs global pollination or local pollination with a certain probability p (p=0.8 in this application):

[0130]

[0131] where: is the position of the s-th individual in the d+1-th generation, is the position of the s-th individual in the d-th generation, is the step control factor; L( ) represents a random step length following a Levy distribution; is the optimal solution of the current population; in the local pollination stage, two individuals and are randomly selected for short-distance information exchange and update, and the update formula is:

[0132]

[0133] where: is the position of the randomly selected f-th individual in the d-th generation, is the position of the randomly selected z-th individual in the d-th generation, and ε is a random number following a uniform distribution.

[0134] After each parameter update, the corresponding Informer model is trained on the training set, and the prediction error (MAPE) is calculated on the validation set, which is used as the fitness value; if the fitness is better than the historical optimal value of the current individual, it is replaced. The algorithm terminates when the maximum number of iterations or the fitness convergence condition is met, and outputs the globally optimal parameter combination, which is used for training and testing the final Informer model. Through this method, the balance between global optimality and convergence speed of parameter search can be achieved, thereby improving the prediction accuracy and stability of the model for electric vehicle charging load.

[0135] The initial neural network model is optimized by the flower pollination algorithm to obtain an optimized first model. The three-dimensional input tensor constructed in S105 is input into the optimized first model. The encoder first extracts the spatio-temporal features of the three-dimensional input tensor, captures the spatial correlation of regional charging load and the time sequence dependence of different time periods through multi-head probability sparse self-attention, and then compresses the feature dimension through the convolution distillation layer. The decoder receives the vector obtained by splicing the deep features output by the encoder and the multi-source associated features, matches the feature correlation through the hidden multi-head attention layer, and finally outputs the hourly charging load prediction value for the next 24 hours through the full connection layer to complete the prediction calculation.

[0136] S107, trend verification is performed on the charging load prediction value and the historical charging load data to obtain a prediction trend.

[0137] First, sufficient historical charging load data is collected. These data need to cover charging load conditions in different time spans, different seasons and different special periods, ensuring that the data are comprehensive and representative.

[0138] Next, the charging load prediction value and the historical charging load data are respectively subjected to trend extraction. For the historical charging load data, a time series analysis method, such as the moving average method, is used to calculate the average charging load within a certain time window, and then the long-term trend and short-term fluctuation trend of the historical charging load are obtained. For the charging load prediction value, the same time granularity and analysis method as the historical data are used to extract the predicted load trend.

[0139] Then, trend verification is performed. The prediction trend is compared with the historical trend in multiple dimensions, including the direction of the trend, the change amplitude of the trend and the periodic characteristics of the trend, etc. The difference between the two can be observed intuitively by drawing a trend comparison chart, or the matching degree of the prediction trend and the historical trend can be quantified by calculating a trend similarity index, such as the cosine similarity. Through these verification operations, the fit of the prediction trend and the historical trend is determined, and a detailed prediction trend description is obtained to provide a basis for the subsequent steps.

[0140] S108, if the difference between the prediction trend and the historical trend exceeds a preset threshold, the charging load prediction value is linearly corrected based on the change rate of the historical trend to obtain a target electric vehicle charging load prediction result.

[0141] First, a preset threshold is determined. The setting of this threshold needs to consider factors such as the fluctuation range of historical charging load data, the accuracy requirement of the prediction task, etc. The difference distribution between the historical trend and the ideal trend (or the reliable predicted trend for multiple times) can be calculated by statistical analysis of historical data, and an appropriate quantile (such as the 90% quantile) can be selected as the preset threshold to ensure that when the difference between the predicted trend and the historical trend exceeds the threshold, it indicates that the predicted trend has a large deviation and needs to be corrected.

[0142] When the difference between the predicted trend and the historical trend exceeds the preset threshold, the change rate of the historical trend is calculated.

[0143] The change rate of the historical trend can be obtained by linear fitting of the historical charging load data. The fitting straight line of the historical charging load data in the time interval is y = k1x + b, where the slope k1 is the change rate of the historical trend, which reflects the speed and direction of the historical charging load change over time. Then, the predicted charging load value is linearly corrected based on the change rate of the historical trend. Let the predicted charging load value be , the corresponding time is t, and the corrected charging load prediction value is calculated according to the historical trend change rate k1 , and the correction formula can be expressed as (where is the time difference from the historical reference time to the prediction time, which can be adjusted according to the actual trend change). Such linear correction can use the change rule of the historical trend to reasonably correct the predicted value with large deviation, making it more consistent with the actual load change trend. Finally, the target electric vehicle charging load prediction result after linear correction is output. At the same time, record the correction reason (i.e. the difference between the predicted trend and the historical trend exceeds the threshold) and the historical trend change rate used in the correction process, etc. to evaluate and optimize the correction effect in the future.

[0144] In some embodiments, in order to comprehensively evaluate the electric vehicle charging load prediction result, a full-scenario multi-dimensional evaluation system including error indicators, scenario adaptability indicators, robustness indicators and economic indicators is constructed; the weights of each evaluation indicator are determined by the analytic hierarchy process; the actual scores under each evaluation layer indicator are obtained according to the target electric vehicle charging load prediction result; the comprehensive evaluation result is obtained according to the weights of each evaluation indicator and the actual scores under each evaluation layer indicator; if the comprehensive evaluation result reaches the preset standard, the demand response strategy is formulated according to the target electric vehicle charging load prediction result.

[0145] The evaluation result is usually a comprehensive score, calculated by weighting the actual scores of each evaluation indicator with weights determined by the analytic hierarchy process (AHP). For example, the comprehensive score might be a value from 0 to 100; a higher score indicates better performance of the charging load prediction across multiple scenarios and dimensions. Preset Standards Example: Preset standards are manually set thresholds to determine whether the prediction results meet requirements. For example, a comprehensive score of 80 or higher might be set as meeting the preset standard. Specific examples assume: an error indicator weight of 0.3 and an actual score of 90 (indicating a small prediction error, such as an average absolute error far lower than the historical average error); a scenario adaptability indicator weight of 0.2 and an actual score of 85 (indicating good adaptability across different scenarios, such as weekdays and holidays); a robustness indicator weight of 0.3 and an actual score of 75 (meaning that the prediction results will not deviate significantly when the input data fluctuates slightly); and an economic efficiency indicator weight of 0.2 and an actual score of 95 (indicating that charging pricing strategies based on this prediction result can bring good economic benefits). According to the comprehensive evaluation calculation method: (90×0.3+85×0.2+75×0.3+95×0.2=27+17+22.5+19=85.5) points. Since 85.5 points ≥ 80 points, meeting the preset standard, a demand response strategy can be formulated based on the electric vehicle charging load forecast results.

[0146] Error metrics are key indicators for measuring the degree of deviation between predicted and actual values. These metrics include Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE), Root Mean Squared Error (RMSE), and prediction trend accuracy. Scenario adaptability metrics include prediction performance under different data missing rates and different weather scenarios. Robustness metrics include prediction stability under data noise interference and hardware failure simulation scenarios. Economic metrics include changes in grid dispatch costs and user charging fees after formulating demand response strategies based on prediction results.

[0147] The formulas for calculating MAE, MAPE, and RMSE are as follows:

[0148] Mean Absolute Error (MAE) directly reflects the average magnitude of prediction error. The calculation formula is:

[0149]

[0150] Mean absolute percentage error Then focus on the proportion of error to the true value, can more intuitive reflect the relative error, the formula is:

[0151]

[0152] Root mean square error RMSE is more sensitive to large errors, can highlight the influence of abnormal values on the prediction effect, the formula is:

[0153]

[0154] In the formula: C is the number of prediction points, c is greater than or equal to 1, less than or equal to C, is the predicted value of the cth electric vehicle charging load, is the true value of the cth electric vehicle charging load, MAE is the mean absolute error, MAPE is the mean absolute percentage error, and RMSE is the root mean square error.

[0155] In addition, there is a prediction trend accuracy rate, which is used to judge the degree of agreement between the predicted load trend and the actual trend, such as comparing the direction and slope of the predicted trend segment and the actual trend segment to calculate the accuracy rate.

[0156] Scenario adaptability index is used to evaluate the performance of the prediction model under different actual scenarios. Under different data missing rate scenarios, it is necessary to test whether the model can still maintain good prediction performance through data completion and feature extraction when the charging data is missing at different proportions (such as 10%, 20%, 30% missing rate); Under different weather scenarios, such as high temperature, low temperature, rain, and strong wind, since they will affect the travel and charging demand of electric vehicles, it is necessary to evaluate the prediction accuracy of the model under these scenarios, for example, compare the changes of MAE, MAPE and other error indicators of the model in sunny and rainy days.

[0157] Robustness index focuses on the stability of the model when it is disturbed. In terms of data noise interference, different intensities of random noise (such as Gaussian noise, noise intensity measured by signal-to-noise ratio) are added to the charging data to observe the fluctuation of the model prediction results; In the hardware failure simulation scenario, such as simulating the failure of the charging pile data acquisition module, causing some period data to be abnormal or missing, testing whether the model can effectively identify and exclude interference and maintain the stability of the prediction.

[0158] The economy index measures the economic impact on the power grid and users after the demand response strategy is formulated based on the prediction results from the perspective of practical application value. For the change of power grid dispatching cost, the change of power grid peak shaving cost and power transmission loss cost after adjusting the power generation plan and power flow distribution according to the prediction results needs to be analyzed. For the change of user charging cost, the saving or increase of user charging cost needs to be calculated after the user adjusts the charging time according to the predicted charging load peak and valley period under the time-of-use electricity price policy.

[0159] After the indexes are constructed, the weights of the evaluation indexes are determined by the analytic hierarchy process (AHP). First, the evaluation target is decomposed into error, scene adaptability, robustness, and economy as the criterion layer, and further decomposed into each specific index as the scheme layer to construct a hierarchical structure model. Then, the judgment matrix is constructed by expert scoring method, the maximum eigenvalue of the judgment matrix and the corresponding eigenvector are calculated, consistency check is performed to ensure that the judgment matrix meets the logic, the weights of the evaluation indexes are obtained, and a complete weight system is formed according to the weights of the evaluation indexes; then, the actual scores of each evaluation index are calculated for the target electric vehicle charging load prediction result; finally, the actual scores of each evaluation layer index are multiplied by the corresponding total weight, and the sum of all the products is calculated to obtain the comprehensive evaluation score, so as to complete the weight integration and form a comprehensive evaluation model that can directly input the prediction result and output the comprehensive evaluation result.

[0160] Then, the target electric vehicle charging load prediction result is input into the comprehensive evaluation model. The comprehensive evaluation model will weight and sum the scores of the error index, the scene adaptability index, the robustness index, and the economy index according to the weights of the indexes to obtain the comprehensive evaluation result. If the comprehensive evaluation result reaches the preset standard, it means that the charging load prediction result has high reliability and application value, and at this time, the demand response strategy is formulated according to the target electric vehicle charging load prediction result, such as guiding the user to stagger the charging during the peak period of the predicted charging load, or adjusting the power supply plan of the power grid to ensure stable and economic power supply.

[0161] In some embodiments, a warning level is constructed according to the target electric vehicle charging load prediction result and the maximum loadable load threshold of the power grid, and the division standard of the warning level is dynamically adjusted according to the proportion of the predicted load exceeding the maximum loadable load threshold of the power grid and the load difference change rate in the target electric vehicle charging load prediction result;

[0162] Firstly, the maximum load threshold of the power grid is determined, which is determined by the power transmission capacity, power transformation equipment capacity, line loss and other factors, and is a key indicator to ensure the safe and stable operation of the power grid. Then, the warning levels are constructed according to the target electric vehicle charging load prediction results and the threshold.

[0163] Specifically, different warning intervals are divided according to the proportion of the predicted load exceeding the maximum load threshold of the power grid, such as yellow warning when the exceeding proportion is 0-10%, orange warning when the exceeding proportion is 10-20%, and red warning when the exceeding proportion is more than 20%. At the same time, the division standard of the warning levels is dynamically adjusted in combination with the load difference change rate. The load difference change rate reflects the change speed of the difference between the predicted load and the maximum load threshold of the power grid. If the change rate is fast, it means that the load growth or decline trend is obvious, and the exceeding proportion range corresponding to each warning level needs to be adjusted appropriately to respond to the dynamic changes of the power grid load more timely.

[0164] The user hierarchy is obtained by combining the user portrait and the user response behavior prediction model. The user portrait includes the user charging frequency, charging time preference, charging power demand, price sensitivity and historical response record. The user response behavior prediction model predicts the response probability of users at different levels under different incentive strategies based on historical data. Specifically as follows:

[0165] When performing user stratification, a comprehensive user portrait is first constructed. The user portrait covers the user charging frequency, i.e. the number of charging times per unit time, which can reflect the user's use intensity of electric vehicles; the charging time preference, which understands the user's habit of charging at which time of the day, which is crucial for peak-shaving charging guidance; the charging power demand, which is different for different vehicle models and different use scenarios, and high-power demand has a greater impact on power grid load; the price sensitivity, i.e. the user's response to price changes, price-sensitive users are more likely to be affected by incentive strategies to adjust charging behavior; and the historical response record, which records the user's participation and response level in similar demand response scenarios in the past.

[0166] Next, a user response behavior prediction model is built. The model is trained based on a large amount of historical data, including the response of each user under different incentive strategies (such as price discount, points reward, etc.), through machine learning algorithms (such as logistic regression, decision tree, etc.). The model can predict the response probability of users at different levels (divided according to the user portrait) under different incentive strategies, providing data support for subsequent development of incentive strategies.

[0167] According to the user response probability obtained by the user response behavior prediction model, the incentive intensity is dynamically adjusted. For users with high response probability, the incentive intensity can be appropriately reduced, because such users are more likely to respond even with small incentives; for users with low response probability, the incentive intensity needs to be increased to improve the response willingness of such users. Combined with the early warning level and user level constructed in the foregoing, a differentiated incentive strategy is formulated. For example, in the red early warning and user charging power demand, the low level of price sensitivity, a high discount and integral reward are provided to encourage the adjustment of the charging period; in the yellow early warning and user charging frequency, the flexible level of charging period preference, a moderate integral reward is given to guide the peak shifting charging.

[0168] According to the differentiated incentive strategy, a personalized demand response scheme is obtained; based on the formulated differentiated incentive strategy, a personalized demand response scheme is generated for the specific situation of each user. The scheme informs the user in detail about the time period for adjusting the charging behavior, the amount of price discount or integral reward that can be obtained, and the benefits to the power grid and itself after adjusting the charging behavior, etc., to ensure that the user can clearly understand and make a response decision.

[0169] An edge computing-based vehicle-grid interaction platform is built. Edge computing has the characteristics of low delay and high bandwidth, and can perform data processing and communication at the network edge side close to the electric vehicle on-board terminal, charging pile control system and user APP. A low-delay communication protocol, such as the Message Queuing Telemetry Transport (MQTT) protocol, is used to realize real-time communication with the electric vehicle on-board terminal, charging pile control system and user APP. Through the platform, the generated personalized demand response scheme can be timely pushed to the user, ensuring the timeliness of information transmission, so that the user can quickly obtain and adjust the charging behavior according to the scheme.

[0170] A real-time feedback and dynamic adjustment mechanism is established. On the one hand, the response of users to the individualized demand response scheme is collected, including whether to adjust the charging behavior according to the scheme, how much to adjust, etc.; on the other hand, the actual load data of the power grid is monitored in real time, and the difference between the predicted load and the actual load is compared. According to these feedback information, the range of response users is dynamically adjusted, such as expanding or reducing the incentive coverage of a certain type of users if the response effect of the users is not good; the differential incentive strategy is adjusted, such as modifying the discount rate of electricity price, the rules of integral reward, etc.; and the warning level is adjusted to ensure that the warning can accurately reflect the current load condition of the power grid. If the load reduction does not reach the pre-set target value, it means that the demand response effect is not as expected, and the demand response strategy needs to be optimized, such as increasing the incentive intensity, adjusting the incentive method, etc.; if the user dissatisfaction value exceeds the pre-set value, it means that the current strategy may bring inconvenience or dissatisfaction to the users, and the strategy also needs to be optimized to balance the demand response effect and user satisfaction while ensuring the load demand of the power grid.

[0171] In order to make the present application clearer, the present application provides a schematic diagram of the prediction result of the electric vehicle charging load, as shown in Figure 2 The red curve in the figure is the predicted value of the first model, and the blue curve is the actual true value. From the figure, it can be seen that within the time span of 0-100 15-minute periods, the prediction curve of the first model has high overall fitting degree with the true curve, especially in the key nodes of load change, such as the stage of rapid load rise around 20-40 periods, and the subsequent peak-valley change process, the predicted value can well follow the trend of the true value, and the mean absolute percentage error (MAPE) is 4.7905%, which reflects that the model has good prediction ability for short-term electric vehicle charging load.

[0172] The present application provides a schematic diagram of the prediction error result, as shown in Figure 3 The red curve in the figure represents the prediction error (the difference between the predicted value and the true value) of the Informer model. Within the time span of 0-100 15-minute periods, the error fluctuates within a certain range, and the absolute value of the error is relatively small in most periods, but there are large error peaks in some periods (such as around period 40). However, overall, the MAPE index of 4.7905% indicates that the prediction error of the model is within an acceptable range, and also reflects that in some special load change scenarios, the model still has room for further optimization of error.

[0173] The present application also provides a comparison table of the short-term electric vehicle charging load prediction results of the first model, CNN and LSTM model, as shown in Table 1:

[0174] Table 1: Comparison table of short-term electric vehicle charging load prediction results of the first model, CNN and LSTM model

[0175]

[0176] Based on the above, the embodiment of the application provides a low-data-quality-scenario electric vehicle charging load prediction method. In the method, first, a first feature affecting the charging load is acquired; a dynamic space-time weight factor is set according to the first feature affecting the charging load; a space-time correlated charging power matrix is constructed; missing values in the space-time correlated charging power matrix are completed according to the dynamic space-time weight factor, to obtain complete target charging power matrix data; a three-dimensional input tensor is constructed according to the complete target charging power matrix data; the three-dimensional input tensor is input into a first model to obtain a charging load prediction value; the charging load prediction value is trend checked with historical charging load data to obtain a prediction trend; if a difference between the prediction trend and a historical trend exceeds a preset threshold, the charging load prediction value is linearly corrected based on a change rate of the historical trend to obtain a target electric vehicle charging load prediction result. The application can more accurately capture the space-time variation law of the charging load by acquiring the first feature affecting the charging load and setting the dynamic space-time weight factor, to complete the missing values of the charging power matrix and construct a three-dimensional input tensor that is more in line with the actual situation; in combination with the first model prediction and the checking and correction of the prediction trend and the historical trend, the accuracy and reliability of the charging load prediction result are guaranteed in multiple dimensions, to provide strong data support for efficient demand response, optimal resource allocation and the like of the power grid.

[0177] The above is combined Figures 1 to 3 The low-data-quality-scenario electric vehicle charging load prediction method provided by the embodiment of the application is described in detail, and the device and equipment provided by the embodiment of the application will be described below with reference to the drawings.

[0178] The embodiment of the application further provides a low-data-quality-scenario electric vehicle charging load prediction device, as shown in Figure 4 The device provided by the embodiment of the application is a low-data-quality-scenario electric vehicle charging load prediction device, and as shown in the drawing, the device comprises:

[0179] The acquisition module 401 is configured to acquire a first feature affecting the charging load.

[0180] The completion module 402 is configured to set a dynamic space-time weight factor according to the first feature affecting the charging load; construct a space-time correlated charging power matrix; and complete missing values in the space-time correlated charging power matrix according to the dynamic space-time weight factor, to obtain complete target charging power matrix data.

[0181] The prediction module 403 is configured to construct a three-dimensional input tensor according to complete target charging power matrix data; input the three-dimensional input tensor into a first model to obtain a charging load prediction value; perform trend checking on the charging load prediction value and historical charging load data to obtain a prediction trend; if a difference between the prediction trend and a historical trend exceeds a preset threshold, perform linear correction on the charging load prediction value based on a change rate of the historical trend to obtain a target electric vehicle charging load prediction result.

[0182] In some possible implementation manners, the acquisition module 401 is specifically configured to acquire first charging data from a plurality of electric vehicle charging stations and charging piles; acquire multi-source associated influence data; identify outliers in the first charging data and the multi-source associated influence data by using an outlier detection algorithm, and eliminate the outliers; perform encoding conversion on text type associated influence data in the first charging data and the multi-source associated influence data after the outliers are eliminated, and unify numerical ranges to construct a first numerical feature set; construct a second numerical feature set according to non-text type associated influence data in the first charging data and the multi-source associated influence data after the outliers are eliminated; and perform screening on the first numerical feature set and the second numerical feature set by using feature correlation analysis to obtain first features that affect charging load.

[0183] In some possible implementation manners, the completion module 402 is specifically configured to construct a low-rank matrix completion optimization model based on the dynamic spatiotemporal weight factor; solve the low-rank matrix completion optimization model by using a singular value threshold algorithm to obtain a target value; replace missing values in a spatiotemporal associated charging power matrix with the target value to obtain a completed charging power matrix; and perform checking on the completed charging power matrix to obtain a result of passing the checking, and take the completed charging power matrix as target charging power matrix data.

[0184] In some possible implementation manners, the apparatus further includes:

[0185] The optimization module is configured to construct an initial neural network model; and optimize the initial neural network model by using a pollination algorithm to obtain the first model.

[0186] In some possible implementation manners, the apparatus further includes:

[0187] The evaluation module is configured to construct a full-scene multi-dimensional evaluation system including error indicators, scene adaptability indicators, robustness indicators, and economic indicators; determine weights of the evaluation indicators by using an analytic hierarchy process; obtain actual scores under each evaluation layer indicator according to the target electric vehicle charging load prediction result; obtain a comprehensive evaluation result according to the weights of the evaluation indicators and the actual scores under each evaluation layer indicator; and if the comprehensive evaluation result reaches a preset standard, formulate a demand response strategy according to the target electric vehicle charging load prediction result.

[0188] In some possible implementation manners, the evaluation module is specifically configured to construct a warning level according to the target electric vehicle charging load prediction result and the maximum bearable load threshold of the power grid, dynamically adjust the division standard of the warning level according to a proportion of a predicted load exceeding the maximum bearable load threshold of the power grid and a load difference change rate in the target electric vehicle charging load prediction result, perform user stratification in combination with a user portrait and a user response behavior prediction model to obtain a user level, wherein the user portrait includes a user charging frequency, a charging time period preference, a charging power demand, a price sensitivity degree, and a historical response record, and the user response behavior prediction model predicts a response probability of each level user under different incentive strategies based on historical data.

[0189] The incentive intensity is dynamically adjusted according to the response probability of the user, and a demand response strategy is formulated according to the incentive intensity, the warning level, and the user level.

[0190] In some possible implementation manners, the dynamic space-time weight factor includes a region weight and a time period weight, wherein the region weight is dynamically adjusted according to a historical load contribution degree and an operation state of a corresponding region charging station in the first feature of the charging load influence, and the time period weight is dynamically configured according to a charging load fluctuation feature in different time periods and a real-time electricity demand change in the first feature of the charging load influence.

[0191] The electric vehicle charging load prediction device in the low data quality scenario according to the embodiments of the present application can correspond to the method described in the embodiments of the present application, and the above other operations and / or functions of each module / unit of the electric vehicle charging load prediction device in the low data quality scenario are respectively implemented to realize the corresponding flow of each method in the embodiments of the present application. For brevity, they will not be repeated here. Figure 1 The above other operations and / or functions of each module / unit of the electric vehicle charging load prediction device in the low data quality scenario are respectively implemented to realize the corresponding flow of each method in the embodiments of the present application. For brevity, they will not be repeated here.

[0192] The embodiments of the present application also provide a computing device. As shown in Figure 5 The figure is a schematic diagram of a computing device provided by the embodiments of the present application. The computing device 500 includes a bus 501, a processor 502, a communication interface 503, and a memory 504. The processor 502, the memory 504, and the communication interface 503 communicate through the bus 501.

[0193] The bus 501 can be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For the sake of brevity, Figure 5Only one bus is shown for simplicity, but there can be more than one bus, and many of the buses can be of different types.

[0194] The processor 502 can be any one or more of a central processing unit (CPU), a graphics processing unit (GPU), a microprocessor (MP), or a digital signal processor (DSP), etc.

[0195] The communication interface 503 is configured to communicate with the outside.

[0196] The memory 504 can include a volatile memory, such as a random access memory (RAM), and / or a non-volatile memory, such as a read-only memory (ROM), a floppy disk, a hard disk, or a solid state drive (SSD).

[0197] The memory 504 stores executable code, and the processor 502 executes the executable code to perform the foregoing method for predicting electric vehicle charging load in a low data quality scenario.

[0198] Specifically, in the case of implementing the embodiment shown in the figure, and Figure 4 In the case of implementing the embodiment shown in the figure, and Figure 4 In the case of implementing the embodiment shown in the figure, and Figure 4 In the case of implementing the embodiment shown in the figure, and

[0199] The embodiments of the present application also provide a computer readable storage medium. The computer readable storage medium can be any available medium or data storage device that can store data which can be accessed by a computing device, or a data center containing one or more available media or data storage devices. The available medium can be a magnetic medium (such as a floppy disk, a hard disk, a magnetic tape), an optical medium (such as a DVD), or a semiconductor medium (such as a solid state disk), etc. The computer readable storage medium includes instructions, which instruct the computing device to execute the foregoing method for predicting electric vehicle charging load in a low data quality scenario.

[0200] The embodiments of the present application also provide a computer program product including one or more computer instructions. When the computer instructions are loaded and executed on a computing device, the computer instructions produce, in whole or in part, the processes or functions described in the embodiments of the present application.

[0201] The computer instructions can be stored in a computer readable storage medium or transmitted from one computer readable storage medium to another computer readable storage medium, for example, the computer instructions can be transmitted from one website site, computer or data center to another website site, computer or data center through wired (for example, coaxial cable, optical fiber) or wireless (for example, infrared, wireless, microwave, etc.).

[0202] The computer program product is executed by a computer, and the computer executes any of the foregoing low-data-quality-scenario electric vehicle charging load prediction method. The computer program product can be a software installation package, and when the foregoing low-data-quality-scenario electric vehicle charging load prediction method is needed, the computer program product can be downloaded and executed on the computer.

[0203] The description of the flow or structure corresponding to each of the above-mentioned figures has its own emphasis, and the parts not described in detail in a certain flow or structure can be referred to the related description of other flows or structures.

[0204] The above is merely specific embodiments of the present application, but the protection scope of the present application is not limited to this, any change or replacement within the technical scope disclosed in the present application should be covered in the protection scope of the present application.

Claims

1. A method for predicting electric vehicle charging load in low data quality scenarios, characterized in that, The method includes: Obtain the first characteristic of the impact on charging load; A dynamic spatiotemporal weighting factor is set based on the first characteristic of the impact on charging load; Construct a spatiotemporally correlated charging power matrix; The missing values ​​in the spatiotemporally correlated charging power matrix are filled in according to the dynamic spatiotemporal weighting factor to obtain complete target charging power matrix data; Construct a three-dimensional input tensor based on the complete target charging power matrix data; The three-dimensional input tensor is input into the first model to obtain the predicted charging load value; The predicted charging load value is compared with historical charging load data to obtain the predicted trend. If the difference between the predicted trend and the historical trend exceeds a preset threshold, the predicted charging load is linearly corrected based on the rate of change of the historical trend to obtain the predicted charging load of the target electric vehicle.

2. The method according to claim 1, characterized in that, The first feature for obtaining the impact on charging load includes: First charging data is obtained from multiple electric vehicle charging stations and charging piles; Obtain multi-source correlation impact data; An outlier detection algorithm is used to identify outliers in the first charging data and multi-source correlation impact data, and outliers are removed. The first charging data after removing outliers and the text-related impact data in the multi-source impact data are encoded and converted, and the numerical range is unified to construct the first numerical feature set. A second numerical feature set is constructed based on the first charging data after removing outliers and the non-textual correlation impact data in the multi-source correlation impact data. By filtering the first and second numerical feature sets through feature correlation analysis, the first feature affecting the charging load is obtained.

3. The method according to claim 1, characterized in that, The step of filling in the missing values ​​in the spatiotemporally correlated charging power matrix according to the dynamic spatiotemporal weighting factor to obtain complete target charging power matrix data includes: Construct a low-rank matrix completion optimization model based on the aforementioned dynamic spatiotemporal weighting factor; The singular value thresholding algorithm is used to solve the low-rank matrix completion optimization model to obtain the target value. The target value is then used to replace the missing values ​​in the spatiotemporally correlated charging power matrix to obtain the completed charging power matrix. The completed charging power matrix is ​​verified, and the verification result is obtained. The completed charging power matrix is ​​then used as the target charging power matrix data.

4. The method according to claim 1, characterized in that, The method further includes: Construct the initial neural network model; The initial neural network model was optimized using a flower pollination algorithm to obtain the first model.

5. The method according to claim 1, characterized in that, The method further includes: Construct a multi-dimensional evaluation system covering all scenarios, including error indicators, scenario adaptability indicators, robustness indicators, and economic indicators; The weights of each evaluation indicator are determined using the analytic hierarchy process (AHP). The actual scores for each evaluation layer index are obtained based on the target electric vehicle charging load prediction results. The comprehensive evaluation result is obtained based on the weight of each evaluation indicator and the actual score under each evaluation level indicator; If the comprehensive evaluation results meet the preset standards, a demand response strategy will be formulated based on the target electric vehicle charging load forecast results.

6. The method according to claim 5, characterized in that, The process of formulating a demand response strategy based on the target electric vehicle charging load forecast includes: Early warning levels are constructed based on the target electric vehicle charging load forecast results and the maximum load carrying capacity threshold of the power grid. The classification criteria for early warning levels are dynamically adjusted based on the proportion of the forecasted load in the target electric vehicle charging load forecast results that exceeds the maximum load carrying capacity threshold of the power grid and the rate of change of the load difference. By combining user profiles and user response behavior prediction models, user segments are obtained. The user profiles include user charging frequency, charging time preferences, charging power requirements, electricity price sensitivity, and historical response records. The user response behavior prediction model predicts the response probability of users at each level under different incentive strategies based on historical data. The incentive level is dynamically adjusted based on the user's response probability, and a demand response strategy is formulated based on the incentive level, the warning level, and the user level.

7. The method according to claim 1, characterized in that, The dynamic spatiotemporal weighting factor includes regional weight and time period weight; wherein, the regional weight is dynamically adjusted based on the historical load contribution and operating status of the corresponding regional charging station in the first feature of the impact on charging load, and the time period weight is dynamically configured based on the charging load fluctuation characteristics and real-time electricity demand changes in different time periods in the first feature of the impact on charging load.

8. A device for predicting electric vehicle charging load in low data quality scenarios, characterized in that, The device includes: The acquisition module is used to acquire the first feature that affects the charging load; The completion module is used to set a dynamic spatiotemporal weighting factor based on the first feature affecting the charging load; construct a spatiotemporally correlated charging power matrix; and complete the missing values ​​in the spatiotemporally correlated charging power matrix according to the dynamic spatiotemporal weighting factor to obtain complete target charging power matrix data. The prediction module is used to construct a three-dimensional input tensor based on the complete target charging power matrix data; input the three-dimensional input tensor into the first model to obtain the predicted charging load value; perform trend verification between the predicted charging load value and historical charging load data to obtain the predicted trend; if the difference between the predicted trend and the historical trend exceeds a preset threshold, then perform linear correction on the predicted charging load value based on the rate of change of the historical trend to obtain the target electric vehicle charging load prediction result.

9. A computing device, characterized in that, Including memory and processor; The memory stores one or more computer programs, the one or more computer programs including instructions; when the instructions are executed by the processor, the computing device performs the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium is used to store a computer program for performing the method as described in any one of claims 1 to 7.

Images