A Spatiotemporal Prediction Method for Generalized Energy Storage of Electric Vehicles Based on BERT

Abstract

This invention discloses a BERT-based spatiotemporal prediction method for generalized energy storage in electric vehicles, comprising: collecting temporal and spatial information of electric vehicles; constructing a spatial-dimensional equivalent model of generalized energy storage for electric vehicles, and training the online-collected spatial information of the electric vehicles using a BERT-based equivalent spatial parameter identification model of generalized energy storage for electric vehicles to identify the parameters of the equivalent model; and constructing a temporal-dimensional equivalent model of generalized energy storage for electric vehicles, and training the online-collected temporal information of the electric vehicles using a BERT-based equivalent temporal parameter identification model of generalized energy storage for electric vehicles to identify the parameters of the equivalent model. This invention can accurately capture the geographical distribution of electric vehicle energy storage, optimize the regional scheduling of charging and discharging behavior, deeply mine the temporal patterns of electric vehicle energy storage behavior, and provide dynamic and accurate time series predictions.

Description

Technical Field

[0001] This invention relates to the field of power system control technology, specifically a spatiotemporal prediction method for generalized energy storage in electric vehicles based on BERT. Background Technology

[0002] Constructing a generalized spatiotemporal equivalent model of energy storage for electric vehicles (EVs) is a key technology for achieving deep integration of EVs and the power grid, yielding significant economic, social, and environmental benefits. Economically, this model optimizes the charging and discharging behavior of EVs, guiding them to charge during off-peak hours and discharge during peak hours, thus achieving grid load balancing. It also helps vehicle owners reduce charging costs and improves revenue from vehicle-to-grid (V2G) services. Furthermore, it optimizes the layout of the power grid and charging infrastructure, avoiding unnecessary investment and improving the economics of grid planning and operation. Socially, the model enhances the flexibility and reliability of the power grid by coordinating the distributed energy storage characteristics of EVs. It helps the grid more efficiently absorb intermittent renewable energy sources, reducing wind and solar power curtailment and providing technical support for the construction of the energy internet and energy structure optimization. Simultaneously, the model promotes coordinated development in the transportation and energy sectors, achieving cross-sectoral resource optimization. Environmentally, the model effectively improves the utilization rate of clean energy, reduces dependence on fossil fuels, and thus reduces carbon emissions from the power system. This technology will also contribute to low-carbon travel and green electricity trading, promoting renewable energy consumption and sustainable development.

[0003] Existing methods have the following shortcomings: 1) Lack of spatiotemporal coupling analysis of electric vehicles. The mobility of electric vehicles makes their energy storage characteristics distribution unpredictable. Existing models mainly focus on regional static analysis, ignoring the dynamic nature of the spatial dimension; they fail to fully consider the impact of the geographical distribution of electric vehicles and the location distribution of charging piles on energy storage characteristics. 2) Insufficient spatial optimization granularity. Existing methods lack spatial accuracy in resource distribution, making it difficult to support refined regional power grid scheduling; they lack in-depth modeling of the relationship between vehicle distribution, movement paths, and energy allocation, resulting in low resource utilization. 3) Insufficient dynamic prediction capability. The energy storage status of electric vehicles changes significantly over time. Existing models mostly use simple linear or time series models, making it difficult to capture complex temporal dynamic relationships; they ignore the impact of historical behavior patterns on future behavior, making it difficult to accurately predict the charging and discharging behavior of electric vehicles. 4) Inaccurate time-series optimization. There is a lack of detailed analysis of the matching of charging and discharging demand with the time of new energy output; the response capability to peak and valley demand is not reflected in time-series control, resulting in insufficient system economy and reliability. Summary of the Invention

[0004] To address the aforementioned issues, this invention proposes a BERT-based spatiotemporal prediction method for generalized energy storage of electric vehicles, capable of deeply mining the temporal patterns of electric vehicle energy storage behavior and providing dynamic and accurate time series predictions.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solution:

[0006] This invention is a spatiotemporal prediction method for generalized energy storage in electric vehicles based on BERT, comprising the following operations:

[0007] Collect time and space information of electric vehicles;

[0008] A BERT-based generalized energy storage spatiotemporal equivalent model for electric vehicles is constructed. This model is trained using the temporal and spatial information of electric vehicles, including: constructing a spatial dimension generalized energy storage equivalent model for electric vehicles; training the model with online-collected spatial information of electric vehicles using a BERT-based generalized energy storage equivalent spatial parameter identification model to identify the parameters of the generalized energy storage equivalent model; and constructing a temporal dimension generalized energy storage equivalent model for electric vehicles; training the model with online-collected temporal information of electric vehicles using a BERT-based generalized energy storage equivalent temporal parameter identification model to identify the parameters of the generalized energy storage equivalent model.

[0009] A further improvement of the present invention is that the time information of the electric vehicle includes the time when the electric vehicle starts charging, the time when the electric vehicle is fully charged for the first time, the time when the electric vehicle starts charging in the final stage before leaving, and the time when the electric vehicle leaves, expressed as:

[0010] X S (t)=[T dao (t),T dao1 (t),T li1 (t),T li (t)]

[0011] Among them, X S (t) represents the electric vehicle time information collected at time t, where T dao (t) represents the time when the electric vehicle started charging, collected at time t. dao1 (t) represents the time when the electric vehicle is first fully charged, collected at time t. li1 (t) represents the charging time from the final stage before the electric vehicle leaves, collected at time t. li (t) represents the departure time of the electric vehicle collected at time t.

[0012] A further improvement of this invention is that the spatial information of the electric vehicle includes: the electric vehicle's driving power, the power of its auxiliary equipment, the electric vehicle's equivalent state of charge, the electric vehicle's rated battery capacity, the electric vehicle's actual equivalent charging power, the electric vehicle's nominal charging power, the electric vehicle's actual equivalent discharging power, the electric vehicle's nominal discharging power, the electric vehicle's actual equivalent energy storage capacity, and the electric vehicle's maximum discharging power, expressed as:

[0013]

[0014] Among them, X K (t) represents the electric spatial information collected at time t, P drive (t) represents the driving power of the electric vehicle collected at time t, P aux (t) represents the power of the electric vehicle auxiliary equipment collected at time t, SOC. eq (t) represents the equivalent state of charge of the electric vehicle collected at time t, C bat For the rated capacity of electric vehicle batteries, Let t be the actual equivalent charging power of the electric vehicle collected at time t. The nominal charging power of electric vehicles. Let t be the actual equivalent discharge power of the electric vehicle collected at time t. E represents the nominal discharge power of an electric vehicle. eff (t) represents the actual equivalent energy storage capacity of the electric vehicle collected at time t. This represents the maximum discharge power of an electric vehicle.

[0015] A further improvement of this invention lies in the following: the expression for the spatial dimension of the generalized energy storage equivalent model of electric vehicles is:

[0016] P run (t)=P drive (t)+P aux (t)

[0017] E eq (t)=SOC eq (t)+C bat

[0018]

[0019] E eff (t)=η e ·E eq (t)

[0020]

[0021] η sys =η ch ·η dis

[0022] Among them, P run (t) represents the operating power of the generalized energy storage equivalent model of electric vehicles, E eq (t) represents the equivalent energy storage capacity of the generalized energy storage equivalent model of electric vehicles, η ch η is the equivalent charging power conversion factor in the generalized energy storage equivalent model of electric vehicles. dis η is the equivalent discharge power conversion factor for the generalized energy storage equivalent model of electric vehicles. e κ is the equivalent energy storage capacity conversion factor for the generalized energy storage equivalent model of electric vehicles. dis (t) represents the sustainable discharge parameters of the generalized energy storage equivalent model of electric vehicles, η sys The equivalent comprehensive efficiency parameters are used for the generalized energy storage equivalent model of electric vehicles.

[0023] A further improvement of this invention is that the expression for the generalized energy storage equivalent spatial parameter identification model for electric vehicles based on the BERT algorithm is as follows:

[0024]

[0025] H0 = Embed(X' K )

[0026] H g =Transformer(H g-1 )

[0027]

[0028] Where X' K X is the normalized spatial eigenvector. K Let X' be a spatial feature vector, μ and σ be the feature mean and standard deviation, respectively, and Embed() be the BERT embedding layer processing function, which processes the normalized spatial feature vector X'. K The processing is performed, H0 is the initial embedding layer, Transformer() is the encoding function of the Transformer layer, and H... g This is the output of the g-th layer. The spatial parameters identified for the i-th sample, y K,i For the actual spatial parameters, H L Let N be the hidden state, and L be the total number of samples. K Let θ be the loss value. K For model parameters, α K For learning efficiency, Let be the gradient descent function. This is a BERT-based model for identifying generalized energy storage space parameters for electric vehicles. To finally identify the equivalent spatial parameters of generalized energy storage for electric vehicles, To identify the operating power, To identify the equivalent energy storage capacity, The equivalent charging power conversion factor is used to identify the power. The equivalent discharge power conversion factor is used to identify the power. The equivalent storage capacity conversion factor is used to identify the capacity. To identify the sustainable discharge parameters, The equivalent comprehensive efficiency parameter is identified.

[0029] A further improvement of this invention is that the expression for the generalized energy storage equivalent model of electric vehicles is:

[0030]

[0031] Where, δ S This represents the generalized equivalent state of an electric vehicle, where 0 indicates the moving state, 1 indicates the charging state, and -1 indicates the discharging state.

[0032] A further improvement of this invention is that the expression for the identification model of the generalized energy storage equivalent time parameter of electric vehicles based on the BERT algorithm is as follows:

[0033]

[0034] in, X is a BERT-based model for identifying the time parameters of generalized energy storage for electric vehicles. S Input features for time, To finally identify the equivalent time parameters of generalized energy storage for electric vehicles, The charging start time of the electric vehicle in the identified generalized energy storage equivalent model of the electric vehicle. The time when the identified electric vehicle was first fully charged. The charging time begins in the final stage before the identified electric vehicle leaves. To identify the departure time of the electric vehicle, The generalized equivalent energy storage charging state of the identified electric vehicle.

[0035] The beneficial effects of this invention are: it can deeply explore the temporal patterns of electric vehicle energy storage behavior and provide dynamic and accurate time series forecasts. This invention supports refined grid regulation, provides the equivalent distribution of generalized energy storage in electric vehicles under different spatiotemporal conditions, and supports precise peak shaving and valley filling of loads and ancillary service scheduling within the region.

[0036] This invention can promote the consumption of new energy sources. The spatial dimension of the generalized energy storage equivalent model of electric vehicles optimizes the energy storage regulation within the region, which helps to promote the local consumption of new energy sources within the region. The temporal dimension of the generalized energy storage equivalent model of electric vehicles predicts the dynamic behavior of energy storage of electric vehicles, thereby improving the time matching degree between energy storage and new energy power generation.

[0037] This invention enhances the capabilities of virtual power plants and the energy internet, supports efficient management of electric vehicle energy storage resources in virtual power plants, optimizes vehicle-to-grid interaction, and helps achieve cross-regional (electric vehicles can move, move in different energy internets, and choose to charge or discharge in different energy internets) and cross-time energy internet resource collaborative optimization. Attached Figure Description

[0038] Figure 1 This is a schematic diagram of the method flow according to an embodiment of the present invention. Detailed Implementation

[0039] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.

[0040] like Figure 1 As shown, this embodiment is a spatiotemporal prediction method for generalized energy storage in electric vehicles based on BERT, including the following steps:

[0041] Step 1: Collect time and space information of electric vehicles.

[0042] The collected spatial information includes the electric vehicle's drive power, auxiliary equipment power, equivalent state of charge, battery rated capacity, actual equivalent charging power, nominal charging power, actual equivalent discharging power, nominal discharging power, actual equivalent energy storage capacity, and maximum discharge power. The expression for this spatial information is:

[0043]

[0044] Among them, X K (t) represents the electric spatial information collected at time t, P drive (t) represents the driving power of the electric vehicle collected at time t, P aux (t) represents the power of the electric vehicle auxiliary equipment collected at time t, SOC. eq (t) represents the equivalent state of charge of the electric vehicle collected at time t, C batFor the rated capacity of electric vehicle batteries, Let t be the actual equivalent charging power of the electric vehicle collected at time t. The nominal charging power of electric vehicles. Let t be the actual equivalent discharge power of the electric vehicle collected at time t. E represents the nominal discharge power of an electric vehicle. eff (t) represents the actual equivalent energy storage capacity of the electric vehicle collected at time t. This represents the maximum discharge power of an electric vehicle.

[0045] The collected time information includes the time when the electric vehicle starts charging, the time when the electric vehicle is fully charged for the first time, the time when the electric vehicle starts charging in the final stage before leaving, and the time when the electric vehicle leaves. The expression for the time information is shown in equation (2), where, in T dao (t) to T dao1 Between (t), the electric vehicle is equivalent to energy storage in a charging state. dao1 (t) to T li1 Between (t), the electric vehicle is equivalent to energy storage, capable of charging or discharging. At T... li1 (t) to T li Between (t), the electric vehicle is equivalent to energy storage in a charging state. dao (t) to T li Outside of time (t), electric vehicles are equivalent to mobile energy storage.

[0046] X S (t)=[T dao (t),T dao1 (t),T li1 (t),T li (t)] (2)

[0047] Among them, X S (t) represents the electric vehicle time information collected at time t, where T dao (t) represents the time when the electric vehicle started charging, collected at time t. dao1 (t) represents the time when the electric vehicle is first fully charged, collected at time t. li1 (t) represents the charging time from the final stage before the electric vehicle leaves, collected at time t. li (t) represents the departure time of the electric vehicle collected at time t.

[0048] Step 2: Construct a BERT-based spatiotemporal equivalent model of generalized energy storage for electric vehicles, and train the BERT-based spatiotemporal equivalent model of generalized energy storage for electric vehicles using the temporal and spatial information of electric vehicles.

[0049] Step 2.1: Construct a spatial dimension generalized energy storage equivalent model for electric vehicles. The constructed generalized energy storage equivalent model for electric vehicles includes operating power, equivalent energy storage capacity, equivalent charging power conversion factor, equivalent discharging power conversion factor, equivalent energy storage capacity conversion factor, sustainable discharge parameters, and equivalent comprehensive efficiency parameters. The expressions of the generalized energy storage equivalent model for electric vehicles in this embodiment are shown in (3) to (9). Among them, expression (3) is the operating power, which represents the power demand of the electric vehicle in the operating state. Expression (4) is the equivalent energy storage capacity, which represents the equivalent energy reserve under the standardized state. Expression (5) is the equivalent charging power conversion factor, which represents the ratio of the actual input power to the nominal charging power during the charging process. Expression (6) is the equivalent discharging power conversion factor, which represents the equivalent coefficient of the output power during the discharging process. Expression (7) is the equivalent energy storage capacity conversion factor, which represents the conversion ratio of the energy storage capacity from the system output. Expression (8) is the sustainable discharge parameter, which represents the performance parameter during continuous discharge. Expression (9) is the equivalent comprehensive efficiency parameter, which represents the overall efficiency of the system during the charge and discharge cycle.

[0050] P run (t)=P drive (t)+P aux (t) (3)

[0051] E eq (t)=SOC eq (t)+C bat (4)

[0052]

[0053] E eff (t)=η e ·E eq (t) (7)

[0054]

[0055] η sys =η ch ·η dis (9)

[0056] Among them, P run (t) represents the operating power of the generalized energy storage equivalent model of electric vehicles, E eq (t) represents the equivalent energy storage capacity of the generalized energy storage equivalent model of electric vehicles, η ch η is the equivalent charging power conversion factor in the generalized energy storage equivalent model of electric vehicles. dis η is the equivalent discharge power conversion factor for the generalized energy storage equivalent model of electric vehicles. e κ is the equivalent energy storage capacity conversion factor for the generalized energy storage equivalent model of electric vehicles. dis(t) represents the sustainable discharge parameters of the generalized energy storage equivalent model of electric vehicles, η sys The equivalent comprehensive efficiency parameters are used for the generalized energy storage equivalent model of electric vehicles.

[0057] Different types of electric vehicles have different parameters, and these parameters change during actual operation, so online parameter identification is required. In this embodiment, the electric vehicle generalized energy storage equivalent spatial parameter identification model based on the BERT algorithm is used to train the spatial information of the electric vehicle collected online, and to identify the operating power, equivalent energy storage capacity, equivalent charging power conversion factor, equivalent discharging power conversion factor, equivalent energy storage capacity conversion factor, sustainable discharge parameters, and equivalent comprehensive efficiency parameters of the electric vehicle generalized energy storage equivalent model. The operation of constructing the electric vehicle generalized energy storage equivalent spatial parameter identification model based on the BERT algorithm in this embodiment is shown in expressions (10) to (17), where expression (10) is the input feature normalization operation model, and the input feature is the spatial feature vector. Expression (11) is the BERT embedding layer processing model, which processes the normalized spatial feature vector X'. K Processing is performed. Expression (12) is the Transformer layer encoding (multi-layer self-attention mechanism). BERT uses multiple Transformers to model the input, and defines the output of the l-th layer as H. l Expression (13) is the output layer model, and the hidden state H of the last layer is... L The parameters are fed into a fully connected layer to predict the generalized equivalent energy storage space parameters of electric vehicles. Expression (14) is the loss function, using mean squared error (MSE) as the loss function. Expression (15) is the parameter optimization model, using gradient descent (such as the Adam optimizer) to update the model parameters. Expression (16) is, and Expression (17) represents the BERT-based model for identifying the generalized energy storage space parameters of electric vehicles. Based on the BERT algorithm, the parameters of X are... K The data was used for training, and the spatial parameters that were finally identified are shown in formula (17).

[0058]

[0059] H0 = Embed(X' K (11)

[0060] H g =Transformer(H g-1 (12)

[0061]

[0062] Among them, X' K X is the normalized spatial eigenvector. KLet X' be a spatial feature vector, μ and σ be the feature mean and standard deviation, respectively, and Embed() be the BERT embedding layer processing function, which processes the normalized spatial feature vector X'. K The processing is performed, H0 is the initial embedding layer, Transformer() is the encoding function of the Transformer layer, and H... g This is the output of the g-th layer. The spatial parameters identified for the i-th sample, y K,i For the actual spatial parameters, H L Let N be the hidden state, and L be the total number of samples. K Let θ be the loss value. K For model parameters, α K For learning efficiency, Let be the gradient descent function. This is a BERT-based model for identifying generalized energy storage space parameters for electric vehicles. To finally identify the equivalent spatial parameters of generalized energy storage for electric vehicles, To identify the operating power, To identify the equivalent energy storage capacity, The equivalent charging power conversion factor is used to identify the power. The equivalent discharge power conversion factor is used to identify the power. The equivalent storage capacity conversion factor is used to identify the capacity. To identify the sustainable discharge parameters, The equivalent comprehensive efficiency parameter is identified.

[0063] Step 2.2, construct the time-dimensional generalized energy storage equivalent model of electric vehicles, as shown in expression (18), where, in T dao (t) to T dao1 Between (t), the electric vehicle is equivalent to energy storage in a charging state. dao1 (t) to T li1 Between (t), the electric vehicle is equivalent to energy storage, capable of charging or discharging. At T... li1 (t) to T li Between (t), the electric vehicle is equivalent to energy storage in a charging state. dao (t) to T li Outside of time (t), electric vehicles are equivalent to mobile energy storage.

[0064]

[0065] Where, δ S This represents the generalized equivalent state of an electric vehicle, where 0 indicates the moving state, 1 indicates the charging state, and -1 indicates the discharging state.

[0066] This embodiment uses the electric vehicle generalized energy storage equivalent time parameter identification model based on the BERT algorithm to train the time information of the electric vehicle collected online, and identifies the parameters of the electric vehicle generalized energy storage equivalent model, as shown in expressions (19) to (21):

[0067]

[0068]

[0069] in, X is a BERT-based model for identifying the time parameters of generalized energy storage for electric vehicles. S Input features for time, To finally identify the equivalent time parameters of generalized energy storage for electric vehicles, The charging start time of the electric vehicle in the identified generalized energy storage equivalent model of the electric vehicle. The time when the identified electric vehicle was first fully charged. The charging time begins in the final stage before the identified electric vehicle leaves. To identify the departure time of the electric vehicle, The generalized equivalent energy storage charging state of the identified electric vehicle.

[0070] The proposed BERT-based spatiotemporal equivalent model for generalized energy storage in electric vehicles improves prediction accuracy, precisely captures the geographical distribution of electric vehicle energy storage, and optimizes regional scheduling of charging and discharging behavior. It also deeply mines the temporal patterns of electric vehicle energy storage behavior, providing dynamic and accurate time-series predictions. This supports refined grid regulation, promotes the integration of new energy sources, improves the temporal matching between energy storage and new energy power generation, enhances the capabilities of virtual power plants and the energy internet, and facilitates cross-regional and cross-temporal collaborative optimization of energy internet resources.

[0071] It will be understood by those skilled in the art that, unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. It should also be understood that terms such as those defined in general dictionaries should be understood to have the same meaning as in the context of the prior art, and should not be interpreted in an idealized or overly formal sense unless defined as herein.

[0072] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A spatiotemporal prediction method for generalized energy storage in electric vehicles based on BERT, characterized in that: This includes the following operations: Collect time and space information of electric vehicles; A BERT-based generalized energy storage spatiotemporal equivalent model for electric vehicles is constructed. This model is trained using the temporal and spatial information of electric vehicles, including: constructing a spatial dimension generalized energy storage equivalent model for electric vehicles; training the model with online-collected spatial information of electric vehicles using a BERT-based generalized energy storage equivalent spatial parameter identification model to identify the parameters of the generalized energy storage equivalent model; and constructing a temporal dimension generalized energy storage equivalent model for electric vehicles; training the model with online-collected temporal information of electric vehicles using a BERT-based generalized energy storage equivalent temporal parameter identification model to identify the parameters of the generalized energy storage equivalent model. The time information of the electric vehicle includes the time when the electric vehicle starts charging, the time when the electric vehicle is fully charged for the first time, the time when the electric vehicle starts charging in the final stage before leaving, and the time when the electric vehicle leaves, expressed as follows: X S (t)=[T dao (t),T dao1 (t),T li1 (t),T li (t)] Among them, X S (t) represents the electric vehicle time information collected at time t, where T dao (t) represents the time when the electric vehicle started charging, collected at time t. dao1 (t) represents the time when the electric vehicle is first fully charged, collected at time t. li1 (t) represents the charging time from the final stage before the electric vehicle leaves, collected at time t. li (t) represents the departure time of the electric vehicle collected at time t; The spatial information of the electric vehicle includes: electric vehicle drive power, electric vehicle auxiliary equipment power, electric vehicle equivalent state of charge, electric vehicle battery rated capacity, electric vehicle actual equivalent charging power, electric vehicle nominal charging power, electric vehicle actual equivalent discharging power, electric vehicle nominal discharging power, electric vehicle actual equivalent energy storage capacity, and electric vehicle maximum discharge power, expressed as: Among them, X K (t) represents the spatial information of the electric vehicle collected at time t, P drive (t) represents the driving power of the electric vehicle collected at time t, P aux (t) represents the power of the electric vehicle auxiliary equipment collected at time t, SOC. eq (t) represents the equivalent state of charge of the electric vehicle collected at time t, C bat For the rated capacity of electric vehicle batteries, Let t be the actual equivalent charging power of the electric vehicle collected at time t. The nominal charging power of electric vehicles. Let t be the actual equivalent discharge power of the electric vehicle collected at time t. E represents the nominal discharge power of an electric vehicle. eff (t) represents the actual equivalent energy storage capacity of the electric vehicle collected at time t. This represents the maximum discharge power of an electric vehicle. The expression for the spatial dimension of the generalized energy storage equivalent model for electric vehicles is: P run (t)=P drive (t)+P aux (t) IN eq (t)=SOC eq (t)+C bat AND eff (t)=η e ·AND eq (t) or sys =the ch ·or dis Among them, P run (t) represents the operating power of the generalized energy storage equivalent model of electric vehicles, E eq (t) represents the equivalent energy storage capacity of the generalized energy storage equivalent model of electric vehicles, η ch η is the equivalent charging power conversion factor in the generalized energy storage equivalent model of electric vehicles. dis η is the equivalent discharge power conversion factor for the generalized energy storage equivalent model of electric vehicles. e κ is the equivalent energy storage capacity conversion factor for the generalized energy storage equivalent model of electric vehicles. dis (t) represents the sustainable discharge parameters of the generalized energy storage equivalent model of electric vehicles, η sys The equivalent comprehensive efficiency parameters for the generalized energy storage equivalent model of electric vehicles; The expression for the time-dimensional generalized energy storage equivalent model of electric vehicles is as follows: Where, δ S This represents the generalized equivalent state of an electric vehicle, where 0 indicates the moving state, 1 indicates the charging state, and -1 indicates the discharging state.

2. The spatiotemporal prediction method for generalized energy storage of electric vehicles based on BERT according to claim 1, characterized in that: The expression for the BERT-based generalized energy storage equivalent spatial parameter identification model for electric vehicles is as follows: H0=Embed(X' K ) H g =Transformer(H g-1 ) Where X' K X is the normalized spatial eigenvector. K Let X' be a spatial feature vector, μ and σ be the feature mean and standard deviation, respectively, and Embed() be the BERT embedding layer processing function, which processes the normalized spatial feature vector X'. K The processing is performed, H0 is the initial embedding layer, Transformer() is the encoding function of the Transformer layer, and H... g This is the output of the g-th layer. The spatial parameters identified for the i-th sample, y K,i For the actual spatial parameters, H L Let N be the hidden state, and L be the total number of samples. K Let θ be the loss value. K For model parameters, α K For learning efficiency, Let be the gradient descent function. This is a BERT-based model for identifying generalized energy storage space parameters for electric vehicles. To finally identify the equivalent spatial parameters of generalized energy storage for electric vehicles, To identify the operating power, To identify the equivalent energy storage capacity, The equivalent charging power conversion factor is used to identify the power. The equivalent discharge power conversion factor is used to identify the power. The equivalent storage capacity conversion factor is used to identify the capacity. To identify the sustainable discharge parameters, Let ||·|| be the equivalent comprehensive efficiency parameter identified, and ||·|| be the vector norm.

3. The spatiotemporal prediction method for generalized energy storage of electric vehicles based on BERT according to claim 1, characterized in that: The expression for the BERT-based electric vehicle generalized energy storage equivalent time parameter identification model is as follows: in, X is a BERT-based model for identifying the time parameters of generalized energy storage for electric vehicles. S Input features for time, To finally identify the equivalent time parameters of generalized energy storage for electric vehicles, The charging start time of the electric vehicle in the identified generalized energy storage equivalent model of the electric vehicle. The time when the identified electric vehicle was first fully charged. The charging time begins in the final stage before the identified electric vehicle leaves. To identify the departure time of the electric vehicle, The generalized equivalent energy storage charging state of the identified electric vehicle.

Images