Charging station load prediction method based on artificial intelligence prediction model and related product

Abstract

The invention provides a charging station load prediction method based on an artificial intelligence prediction model and a related product, and relates to the field of energy management. The charging station load prediction method based on the artificial intelligence prediction model comprises the following steps: obtaining static charging characteristics of each vehicle charged in a charging station; determining a charging strategy type of each vehicle according to the static charging characteristics; triggering corresponding artificial intelligence prediction models for different charging strategy types, and predicting a charging power curve of each vehicle based on the artificial intelligence prediction models; and superposing the charging power curves of all vehicles to obtain the future charging load of the charging station. According to the scheme provided by the invention, the charging power curve of each vehicle is predicted by using the artificial intelligence prediction model according to the strategies, so that the matching between the prediction of the charging power curve and the actual charging strategy of the vehicle is ensured, and the problem of low prediction precision caused by modeling without distinguishing the charging strategy is solved; therefore, the accuracy of load prediction of the charging station is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of energy management, and in particular to a charging station load prediction method based on an artificial intelligence prediction model and related products. BACKGROUND

[0002] Currently, the mainstream methods for charging station load prediction mainly include statistical models, machine learning, optimization, and hybrid methods. Statistical models use historical data and statistical analysis to predict load, and can use time series analysis, regression models, etc. to establish a prediction model. Machine learning methods train models using historical data, and predict by learning the patterns and relationships of the data. Common algorithms include support vector machines, random forests, and neural networks. Optimization methods convert the load prediction problem into an optimization problem, taking into account charging demand, energy costs, etc., and solving the optimal charging strategy through optimization algorithms. Hybrid methods combine the advantages of multiple methods, using statistical models to handle long-term trends and seasonal effects, and using machine learning methods to handle complex nonlinear relationships. For example, by introducing charging station distance, surrounding store / school / hotel distribution, etc. static geographic factors to build a user selection model, and using the Monte Carlo method to predict charging station load, the core process includes attraction calculation, vehicle quantity dynamic update, and load aggregation.

[0003] However, the above scheme only models user spatial selection behavior and does not involve the power demarcation of charging power characteristics under supercharging, fast charging, slow charging, etc. charging strategies, cannot reflect the dynamic change law of power curve under different charging strategies, and has the problem of insufficient accuracy. At the same time, the load prediction only relies on static attraction charging characteristics, and does not model the power changes in the constant current and constant voltage stages during charging, resulting in the problem that the predicted results cannot accurately match the coupling relationship between actual charging time and power fluctuations, and further leading to time prediction deviation. In addition, when calculating the overall electricity load of the charging station, only the load is calculated by simply superimposing the number of vehicles, without considering the power superposition effect of different charging strategies, making it difficult to support fine management needs such as grid peak shaving and valley filling. SUMMARY

[0004] In view of the above problems, a charging station load prediction method based on an artificial intelligence prediction model and related products are proposed to overcome the above problems or at least partially solve the above problems.

[0005] An object of the present application is to provide a charging station load prediction method based on an artificial intelligence prediction model that can improve the accuracy of charging station load prediction.

[0006] A further object of the present application is to improve the real-time and accuracy of charging power curve prediction, thereby further improving the accuracy of charging station load prediction.

[0007] A further object of the present application is to improve the matching of the charging power curve and the actual charging stage, thereby further improving the accuracy of the charging power curve prediction.

[0008] In particular, the present application provides a charging station load prediction method based on an artificial intelligence prediction model. The charging station load prediction method based on the artificial intelligence prediction model comprises:

[0009] Obtaining static charging characteristics of each vehicle being charged in the charging station;

[0010] Determining the charging strategy type of each vehicle according to the static charging characteristics;

[0011] Calling an artificial intelligence prediction model corresponding to the charging strategy type, and predicting the charging power curve of each vehicle based on the artificial intelligence prediction model;

[0012] Superimposing the charging power curves of all vehicles to obtain the overall power consumption load curve of the charging station.

[0013] Optionally, the step of predicting the charging power curve of each vehicle based on the artificial intelligence prediction model comprises:

[0014] Real-time acquisition of dynamic charging characteristics of the vehicle and historical charging data corresponding to the vehicle, wherein the dynamic charging characteristics include environmental temperature, real-time power sequence and current state of charge;

[0015] Inputting the static charging characteristics, dynamic charging characteristics and historical charging data into the artificial intelligence prediction model to generate the charging power curve by the artificial intelligence prediction model;

[0016] According to a preset time interval, the real-time acquired dynamic charging characteristics are periodically input into the artificial intelligence prediction model to dynamically correct the charging power curve generated by the artificial intelligence prediction model.

[0017] Optionally, the artificial intelligence prediction models corresponding to different charging strategy types are configured with different stage determination conditions; and the step of generating the charging power curve by the artificial intelligence prediction model comprises:

[0018] According to the environmental temperature, the historical charging data are weighted to obtain a weighted reference power curve;

[0019] According to the real-time power sequence, the current state of charge and the stage determination condition, the charging stage of the vehicle is determined, wherein the charging stage of the vehicle includes a constant current stage and a constant voltage stage;

[0020] Based on the charging stage, the weighted reference power curve, the static charging characteristics and the dynamic charging characteristics, the predicted expected charging time and the expected power curve of the vehicle in each charging stage during the entire charging process are predicted;

[0021] generate the charging power curve based on the expected charging duration and the expected power curve of each charging phase.

[0022] Optionally, the static charging characteristics include a battery capacity of the vehicle, a target state of charge, a maximum charging rate, and a rated charging power; and the step of predicting the expected charging duration and the expected power curve of each charging phase of the vehicle during the entire charging process based on the charging phase, the weighted reference power curve, the static charging characteristics, and the dynamic charging characteristics comprises:

[0023] predicting an expected initial power of the constant-current phase and a model predicted duration of the constant-current phase according to the charging phase and the weighted reference power curve;

[0024] calculating an expected charging duration of the constant-current phase based on the model predicted duration, the static charging characteristics, and the dynamic charging characteristics;

[0025] generating an expected power curve of the constant-current phase according to the expected initial power and the expected charging duration of the constant-current phase.

[0026] Optionally, the calculation formula of the expected charging duration of the constant-current phase is:

[0027]

[0028] In the calculation formula, T1 is the expected charging duration of the constant-current phase, T LSTM is the model predicted duration, SOC target is the target state of charge, SOC now is the current state of charge, C is the battery capacity of the vehicle, the coefficient 60 represents a time conversion coefficient, P avg is the current average power of the vehicle.

[0029] Optionally, the static charging characteristics include a battery capacity of the vehicle, a target state of charge, a maximum charging rate, and a rated charging power; and the step of predicting the expected charging duration and the expected power curve of each charging phase of the vehicle during the entire charging process based on the charging phase, the weighted reference power curve, the static charging characteristics, and the dynamic charging characteristics comprises:

[0030] predicting an expected initial power of the constant-voltage phase according to the charging phase and the weighted reference power curve;

[0031] generating an expected power curve of the constant-voltage phase according to the expected initial power of the constant-voltage phase and a preset power decay model, wherein the power decay model dynamically corrects a decay coefficient based on an ambient temperature;

[0032] calculating an expected charging duration of the constant-voltage phase based on the expected power curve of the constant-voltage phase and a charging termination power.

[0033] Optionally, the formula of the power attenuation model is:

[0034] P(t) = P0·e -k·t ,

[0035] In the formula of the power attenuation model, P0 is the expected starting power of the constant voltage stage, and k is the attenuation coefficient;

[0036] The dynamic correction formula of the attenuation coefficient is:

[0037]

[0038] In the dynamic correction formula, k base is a preset initial attenuation coefficient, T env is the ambient temperature, and C is the battery capacity of the vehicle; and

[0039] The calculation formula of the expected charging duration of the constant voltage stage is:

[0040]

[0041] In the calculation formula, T2 is the expected charging duration of the constant voltage stage, and P cutoff is the charging termination power.

[0042] Optionally, the historical charging data includes a plurality of historical charging curves and historical temperatures recorded when each historical charging curve is recorded;

[0043] The weight calculation formula for weighted calculation of the historical charging data is:

[0044]

[0045] In the weight calculation formula, W i represents the weight of the i-th historical charging curve in the historical charging data, T env is the ambient temperature, is the historical temperature recorded when the i-th historical charging curve is recorded; and

[0046] The formula of the weighted reference power curve is:

[0047]

[0048] In the formula of the weighted reference power curve, n is the number of historical charging curves, represents the power value of the i-th historical charging curve at time t.

[0049] Optionally, the stage determination condition includes:

[0050] σ P <δ s , wherein σP is a standard deviation of the real-time power sequence, δ s is a standard deviation threshold corresponding to the charging strategy type; and

[0051] SOC now < θ b , wherein the SOC now is a current state of charge, θ b is a battery turning point state of charge of the vehicle; and

[0052] The step of determining the charging phase in which the vehicle is located according to the real-time power sequence, the current state of charge and the phase determination condition comprises:

[0053] sequentially determining whether the real-time power sequence and the current state of charge satisfy the phase determination condition;

[0054] in a case where the determination results are both yes, determining that the vehicle is in the constant current phase;

[0055] in a case where the determination results are both no, determining that the vehicle is in the constant voltage phase.

[0056] Optionally, the static charging feature comprises a battery type of the vehicle and a charging request power, and the charging strategy type comprises a plurality of charging strategies with different rated charging powers; and

[0057] The step of determining the charging strategy type of each vehicle according to the static charging feature comprises:

[0058] comparing the charging request power with preset power threshold intervals, wherein different power threshold intervals are respectively set according to rated charging powers of the vehicle under different charging strategies;

[0059] determining the charging strategy type of the vehicle according to the power threshold interval to which the charging request power belongs and the battery type.

[0060] Optionally, the charging strategy type comprises a slow charging strategy, a fast charging strategy and an ultra-fast charging strategy with rated charging powers increasing in sequence; and

[0061] The step of determining the charging strategy type of the vehicle according to the power threshold interval to which the charging request power belongs and the battery type comprises:

[0062] in a case where the power threshold interval to which the charging request power belongs corresponds to the slow charging strategy or the fast charging strategy, taking the charging strategy corresponding to the power threshold interval to which the charging request power belongs as the charging strategy type of the vehicle;

[0063] in a case where the power threshold interval to which the charging request power belongs corresponds to the ultra-fast charging strategy, determining whether the vehicle supports the ultra-fast charging according to the battery type;

[0064] If yes, the charging strategy type of the vehicle is determined as the super charging strategy;

[0065] If no, the charging strategy type of the vehicle is determined as the fast charging strategy.

[0066] Optionally, the static charging feature further comprises a preset stay duration; and

[0067] After the step of predicting the charging power curve of each vehicle based on the artificial intelligence prediction model, the charging station load prediction method based on the artificial intelligence prediction model further comprises:

[0068] In the case that the charging strategy type of the vehicle is the slow charging strategy or the fast charging strategy, the charging power curve of the vehicle is optimized according to the preset stay duration.

[0069] According to another aspect of the present application, there is also provided a computer program product comprising a computer program which, when executed by a processor, implements the steps of any of the charging station load prediction methods based on the artificial intelligence prediction model described above.

[0070] According to still another aspect of the present application, there is also provided a computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the steps of any of the charging station load prediction methods based on the artificial intelligence prediction model described above.

[0071] According to still another aspect of the present application, there is also provided a computer device comprising a memory, a processor and a computer program stored on the memory, the processor executing the computer program to implement the steps of any of the charging station load prediction methods based on the artificial intelligence prediction model described above.

[0072] The charging station load prediction method based on the artificial intelligence prediction model of the present application realizes the early prediction of the power consumption load of each vehicle by obtaining the static charging characteristics of each vehicle charging in the charging station, determining the charging strategy type of each vehicle according to the static charging characteristics, calling the artificial intelligence prediction model corresponding to the charging strategy type, and predicting the charging power curve of each vehicle based on the artificial intelligence prediction model. In addition, the charging station load prediction method based on the artificial intelligence prediction model of the present application also obtains the power consumption load curve of the charging station as a whole by superimposing the charging power curves of all vehicles, thereby realizing the early prediction of the overall power consumption load of the charging station. Therefore, the charging station load prediction method based on the artificial intelligence prediction model of the present application divides the charging strategy of each vehicle and uses the artificial intelligence prediction model to predict the charging power curve of each vehicle according to the strategy, thereby ensuring the matching of the charging power curve prediction and the actual charging strategy of the vehicle, solving the problem of low prediction accuracy caused by not distinguishing the charging strategy modeling, thereby improving the accuracy of the charging station load prediction, and further providing accurate load prediction data for the operation management of the charging station and the dispatching of the power grid, which is helpful to realize the stable operation of the power grid and the reasonable allocation of resources.

[0073] Further, the charging station load prediction method based on the artificial intelligence prediction model of the present application realizes the dynamic correction of the charging power curve generated by the artificial intelligence prediction model by real-time acquisition of the dynamic charging characteristics such as the environmental temperature, real-time power sequence and current state of charge of the vehicle and the historical charging data corresponding to the vehicle, inputting the static charging characteristics, dynamic charging characteristics and historical charging data into the artificial intelligence prediction model, and periodically inputting the real-time acquired dynamic charging characteristics into the artificial intelligence prediction model according to the preset time interval. Therefore, the charging station load prediction method based on the artificial intelligence prediction model of the present application inputs the static charging characteristics, dynamic charging characteristics and historical charging data into the artificial intelligence prediction model, so that the model can comprehensively predict multiple factors to generate a charging power curve that is more in line with the actual situation, thereby improving the accuracy of the charging power curve prediction. At the same time, the prediction result is dynamically corrected based on the real-time acquired dynamic charging characteristics, thereby improving the real-time performance of the charging power curve prediction, so that the prediction can timely reflect the changes in the vehicle charging process, thereby improving the timeliness and accuracy of the prediction, and further improving the accuracy of the charging station load prediction.

[0074] Further, the charging station load prediction method based on the artificial intelligence prediction model of the present application realizes the strategy-based prediction of the charging power curve by configuring different stage determination conditions for the artificial intelligence prediction model corresponding to different charging strategies, obtains the weighted reference power curve by weighting the historical charging data according to the environmental temperature, determines the charging stage of the vehicle according to the real-time power sequence, the current state of charge and the stage determination condition, and predicts the expected charging time and the expected power curve of each charging stage of the vehicle in the entire charging process based on the charging stage, the weighted reference power curve, the static charging characteristics and the dynamic charging characteristics, and generates the photovoltaic power curve and the load power curve of the micro-grid predicted by the charging power curve, so as to realize the generation of the charging power curve in stages by the artificial intelligence prediction model. Therefore, the charging station load prediction method based on the artificial intelligence prediction model of the present application realizes the fine modeling of the charging power dynamic change law in stages through a multi-step prediction mechanism, solves the prediction deviation problem caused by the lack of stage-based dynamic modeling, and further improves the timing accuracy and reliability of the power curve, so that the prediction result can more intuitively reflect the charging process of the vehicle, improves the matching of the charging power curve and the actual charging stage, and further improves the accuracy of the charging power curve prediction, and further guarantees the accuracy of the overall power load prediction of the charging station.

[0075] The above and other objects, advantages and features of the present application will become more apparent from the following detailed description of some embodiments thereof, taken in conjunction with the accompanying drawings. BRIEF DESCRIPTION OF DRAWINGS

[0076] Some embodiments of the present application will now be described in detail in connection with the accompanying drawings, which are described below. The same reference numerals are used in different drawings to refer to the same or like parts or components. It should be understood that these drawings are not necessarily to scale. In the drawings:

[0077] Figure 1 is a control block diagram of the charging station load prediction method based on the artificial intelligence prediction model according to an embodiment of the present application;

[0078] Figure 2 is a control block diagram of the artificial intelligence prediction model prediction in the charging station load prediction method based on the artificial intelligence prediction model according to an embodiment of the present application;

[0079] Figure 3 is a schematic diagram of a charging power curve in the charging station load prediction method based on the artificial intelligence prediction model according to an embodiment of the present application;

[0080] Figure 4is a schematic diagram of an overall power consumption curve of a charging station in a charging station load prediction method based on an artificial intelligence prediction model according to an embodiment of the present application;

[0081] Figure 5 is a flowchart of a control instance of a charging station load prediction method based on an artificial intelligence prediction model according to an embodiment of the present application;

[0082] Figure 6 is a schematic diagram of a computer program product according to an embodiment of the present application;

[0083] Figure 7 is a schematic diagram of a computer readable storage medium according to an embodiment of the present application;

[0084] Figure 8 is a schematic block diagram of a computer device according to an embodiment of the present application. DETAILED DESCRIPTION

[0085] Those skilled in the art will understand that the embodiments described below are only a part of the embodiments of the present application, rather than all the embodiments of the present application, and are intended to explain the technical principles of the present application, rather than limit the protection scope of the present application. Based on the embodiments provided by the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor shall fall within the protection scope of the present application.

[0086] Figure 1 is a control block diagram of a charging station load prediction method based on an artificial intelligence prediction model according to an embodiment of the present application. As shown in Figure 1 , the statistical method of process query of the present embodiment can generally include:

[0087] In step S102, static charging characteristics of each vehicle charging in the charging station are acquired. It should be noted that the static charging characteristics of the vehicle in the present embodiment can be acquired when the vehicle is plugged into the charging pile of the charging station, and can include static parameters such as battery type, battery capacity C, charging request power, preset stay duration, target state of charge (SOC), maximum charging rate R max and rated charging power P max In addition, the preset stay duration can be set by the user.

[0088] In step S104, the charging strategy type of each vehicle is determined according to the static charging characteristics. It should be noted that the charging strategy type can include multiple charging strategies with different rated charging powers. For example, the charging strategy type can include slow charging strategy, fast charging strategy and super charging strategy with rated charging power increasing in turn.

[0089] Step S106, an artificial intelligence prediction model corresponding to the charging strategy type is called, and the charging power curve of each vehicle is predicted based on the artificial intelligence prediction model. It should be noted that a dedicated artificial intelligence prediction model is set for each charging strategy type. In addition, the artificial intelligence prediction model in this step can adopt an LSTM (Long Short-Term Memory, Long Short-Term Memory Neural Network) model. That is, for the slow charging strategy, a slow charging dedicated LSTM model is called; for the fast charging strategy, a fast charging dedicated LSTM model is called; and for the super charging strategy, a super charging dedicated LSTM model is called.

[0090] Step S108, the charging power curves of all vehicles are superimposed to obtain the overall power consumption load curve of the charging station. That is, through the strategy prediction and linear superposition algorithm, the overall power consumption load of the charging station in the future preset time period is obtained to support the dynamic scheduling of the power grid and the power market bidding.

[0091] Therefore, the charging station load prediction method based on the artificial intelligence prediction model of the embodiment of the present application divides the charging strategy of each vehicle and uses the LSTM model to predict the charging power curve of each vehicle, which ensures the matching of the charging power curve prediction and the actual charging strategy of the vehicle, solves the problem of low prediction accuracy caused by not distinguishing the charging strategy modeling, thereby improving the accuracy of the charging station load prediction, and further providing accurate load prediction data for the operation management of the charging station and the scheduling of the power grid, which helps to realize the stable operation of the power grid and the reasonable allocation of resources.

[0092] In some embodiments, in the above step S102, the step of obtaining the battery type and the battery capacity can be specifically performed by obtaining the vehicle VIN code of each vehicle charged in the charging station to obtain the battery type and the battery capacity.

[0093] In addition, the above step S104 can include the following steps: comparing the charging request power with a preset power threshold interval, wherein different power threshold intervals are set according to the rated charging power of the vehicle under different charging strategies; and determining the charging strategy type of the vehicle according to the power threshold interval to which the charging request power belongs and the battery type.

[0094] Therefore, the charging station load prediction method based on the artificial intelligence prediction model of the embodiment of the present application determines the charging strategy type by combining the charging request power and the battery type, realizes the automatic classification of the charging strategy based on the inherent characteristics of the vehicle, thereby solving the problem of undifferentiated strategy caused by the traditional method relying only on the historical mean or simple time sequence, and further laying a differentiated foundation for the selection of the LSTM model and the power prediction.

[0095] In some embodiments, in the step S104 described above, the slow charging strategy, the fast charging strategy and the super charging strategy can correspond to power threshold intervals with continuous and sequentially increasing data ranges, respectively. For example, the upper limit value of the power threshold interval corresponding to the slow charging strategy and the lower limit value of the power threshold interval corresponding to the fast charging strategy are both the first power threshold value, and the upper limit value of the power threshold interval corresponding to the fast charging strategy and the lower limit value of the power threshold interval corresponding to the super charging strategy are both the second power threshold value. The first power threshold value is less than the second power threshold value, and the first power threshold value and the second power threshold value are respectively set according to the rated charging power of the vehicle under the slow charging strategy and the fast charging strategy. For example, when the rated charging power of the vehicle under the slow charging strategy is 50KW and the rated charging power of the vehicle under the fast charging strategy is 150KW, the first power threshold value can be set to 50KW and the second power threshold value can be set to 150KW. That is, the power threshold interval corresponding to the slow charging strategy can be (0, 50KW), which is suitable for a long-time parking scene; the power threshold interval corresponding to the fast charging strategy can be [50KW, 150KW), which is suitable for a regular charging scene; and the power threshold interval corresponding to the fast charging strategy can be [150KW, 1500KW), which is suitable for a high-SOC fast power compensation scene.

[0096] On this basis, the step of determining the charging strategy type of the vehicle according to the power threshold interval to which the charging request power belongs and the battery type described above can include the following steps: in the case where the power threshold interval to which the charging request power belongs corresponds to the slow charging strategy or the fast charging strategy, taking the charging strategy corresponding to the power threshold interval to which the charging request power belongs as the charging strategy type of the vehicle; in the case where the power threshold interval to which the charging request power belongs corresponds to the super charging strategy, judging whether the vehicle supports super charging according to the battery type; if yes, determining the charging strategy type of the vehicle as the super charging strategy; and if no, determining the charging strategy type of the vehicle as the fast charging strategy.

[0097] That is, first, the charging strategies are preliminarily divided according to the power threshold interval to which the charging request power belongs, and then when the charging strategy type of the vehicle is preliminarily divided as the super charging strategy, it is further determined whether the vehicle supports the super charging strategy according to the battery type of the vehicle, so as to avoid the battery damage caused by directly charging under the super charging strategy of the vehicle which is not suitable for super charging, thereby improving the charging safety.

[0098] In some embodiments, the step S106 can include the following steps: acquiring the dynamic charging feature of the vehicle and the historical charging data corresponding to the vehicle in real time; inputting the static charging feature, the dynamic charging feature and the historical charging data into the LSTM model to generate the charging power curve by the LSTM model; and inputting the dynamic charging feature acquired in real time into the LSTM model at a preset time interval to dynamically correct the charging power curve generated by the LSTM model. That is, based on the static charging feature, the dynamic charging feature of the vehicle and the historical charging data corresponding to the vehicle, the charging power curve is generated by the LSTM model, and every preset time interval, the prediction result is dynamically corrected according to the newly acquired dynamic charging feature to obtain the latest prediction result. In addition, it should be noted that the dynamic charging feature can include the current environmental temperature T env of the charging station, the real-time power sequence P real of the vehicle in the actual charging process and the current state of charge SOC now of the vehicle.

[0099] Therefore, the charging station load prediction method based on the artificial intelligence prediction model of the embodiment of the application can make the model predict by comprehensively considering various factors by inputting the static charging feature, the dynamic charging feature and the historical charging data into the LSTM model, generate a charging power curve that is more in line with the actual situation, improve the accuracy of the charging power curve prediction, and at the same time, dynamically correct the prediction result based on the dynamic charging feature acquired in real time at a regular time interval, improve the real-time performance of the charging power curve prediction, so that the prediction can timely reflect the changes in the vehicle charging process, improve the timeliness and accuracy of the prediction, thereby further improving the accuracy of the charging station load prediction.

[0100] In some embodiments, in the step S104, the LSTM models corresponding to different charging strategy types are configured with different stage determination conditions to be applicable to the charging power curve prediction under different charging strategy types. Based on this, the step of generating the charging power curve by the LSTM model can include the following steps: performing weighted calculation on the historical charging data according to the environmental temperature to obtain a weighted reference power curve; determining the charging stage of the vehicle according to the real-time power sequence, the current state of charge and the stage determination condition; predicting the expected charging duration and the expected power curve of each charging stage in the entire charging process of the vehicle based on the charging stage, the weighted reference power curve, the static charging feature and the dynamic charging feature; and generating the charging power curve based on the expected charging duration and the expected power curve of each charging stage.

[0101] It should be noted that the charging stage of the vehicle generally includes a constant current stage and a constant voltage stage. In the initial charging stage, the constant current stage is automatically entered, and in this stage, the charging pile adopts a mode of keeping the current constant to perform fast charging. When the battery capacity reaches a certain degree, the constant voltage stage is automatically entered, and in this stage, the charging pile adopts a mode of keeping the voltage constant to perform charging. The charging speed in this stage is relatively slow, but is more friendly to the battery life, so as to maintain the stability of the chemical reaction inside the battery and avoid damage caused by overcharging of the battery. In addition, in the present embodiment, the charging power in the constant current stage can be kept constant continuously for at least a period of time, and the charging power in the constant voltage stage continuously decays.

[0102] In the present embodiment, first, an LSTM model is adopted to input a real-time power sequence, an environmental temperature and historical charging data. Then, an attention mechanism is adopted to perform weighted calculation on the historical charging data to obtain a weighted reference power curve as a reference charging curve of the current vehicle. Then, a static feature fusion mechanism is utilized to add static charging features and dynamic charging features of the vehicle into a prediction process to perform dynamic modeling, so as to realize stage-by-stage correction of the duration and power curve of each charging stage. Finally, the remaining charging time and the stage-by-stage power change curve are obtained.

[0103] Therefore, the charging station load prediction method based on the artificial intelligence prediction model of the present embodiment realizes stage-by-stage fine modeling of the dynamic change law of the charging power through a multi-step prediction mechanism, solves the prediction deviation problem caused by the lack of stage-by-stage dynamic modeling, and further improves the timing accuracy and reliability of the power curve. The prediction result can more intuitively reflect the charging process of the vehicle, improve the matching of the charging power curve and the actual charging stage, and further improve the accuracy of the charging power curve prediction, thereby further ensuring the accuracy of the overall power load prediction of the charging station.

[0104] In some embodiments, the historical charging data can be obtained based on the vehicle type and the charging strategy type of the current vehicle, and can include a plurality of historical charging curves and historical temperatures recorded at each historical charging curve

[0105] In a specific embodiment, the weight calculation formula for the weighted calculation of the historical charging data can be:

[0106]

[0107] In the above weight calculation formula, W i represents the weight of the i-th historical charging curve in the historical charging data, T env is the environmental temperature, The historical temperature at the time of recording the i-th historical charging curve is recorded. Thus, each historical charging curve can be weighted based on the historical temperature and the current ambient temperature, so that the historical charging curve closer to the current ambient temperature has a greater weight, that is, the historical charging curve closer in temperature is more relevant.

[0108] Further, the formula of the weighted reference power curve can be:

[0109]

[0110] In the formula of the weighted reference power curve, n is the number of historical charging curves, represents the power value of the i-th historical charging curve at time t.

[0111] Thus, the charging station load prediction method based on the artificial intelligence prediction model of the embodiments of the present application realizes the construction of the weighted reference power curve based on historical data and real-time data through attention mechanism weighting, and provides a reference benchmark for the subsequent construction of the charging power curve.

[0112] In some embodiments, the stage determination condition can include: P <δ s , wherein σ P is the standard deviation of the real-time power sequence, δ s is the standard deviation threshold corresponding to the charging strategy type; and SOC now <θ b , wherein SOC now is the current state of charge, and θ bThe battery turning point state of charge of the vehicle. It should be noted that the standard deviation threshold corresponding to different charging strategy types is different, and the standard deviation threshold corresponding to each charging strategy type can be set according to the actual power change of the vehicle under the charging strategy type. Specifically, in one specific embodiment, the standard deviation threshold corresponding to the super charging strategy is greater than the standard deviation threshold corresponding to the fast charging strategy, and the standard deviation threshold corresponding to the fast charging strategy is greater than the standard deviation threshold corresponding to the slow charging strategy. Specifically, the standard deviation threshold corresponding to the super charging strategy can be selected from any value in 4.5kW-10kW, for example, the standard deviation threshold corresponding to the super charging strategy is set to 5kW. The standard deviation threshold corresponding to the fast charging strategy can be selected from any value in 2.5kW-4kW, for example, the standard deviation threshold corresponding to the fast charging strategy is set to 3kW. In addition, the standard deviation threshold corresponding to the slow charging strategy can be selected from any value in 0.5kW-2kW, for example, the standard deviation threshold corresponding to the slow charging strategy is set to 1kW. In addition, the battery turning point state of charge can be set according to the charging condition of different types of batteries at different states of charge. Specifically, the battery turning point state of charge can be selected from any value in 70%-95%. In one specific embodiment, for ternary lithium batteries, the turning point state of charge can be set to 80%, and for iron lithium phosphate batteries, the turning point state of charge can be set to 90%.

[0113] Based on this, the above step of determining the charging phase of the vehicle according to the real-time power sequence, the current state of charge and the phase determination condition can include the following steps: judging whether the real-time power sequence and the current state of charge satisfy the phase determination condition in turn; in the case where the judgment results are both yes, determining that the vehicle is in the constant current phase; in the case where the judgment results are both no, determining that the vehicle is in the constant voltage phase.

[0114] That is, only in the case where the real-time power sequence and the current state of charge of the vehicle both satisfy the phase determination condition, it is determined that the vehicle is currently in the constant current state, and in the case where the real-time power sequence and the current state of charge of the vehicle both do not satisfy the phase determination condition, it is determined that the vehicle is currently in the constant voltage state.

[0115] Therefore, the charging station load prediction method based on the artificial intelligence prediction model of the embodiment of the application realizes the rapid and accurate determination of the current charging phase of the vehicle by judging whether the real-time power sequence and the current state of charge of the vehicle satisfy the phase determination condition, and provides a basis for subsequent generation of charging power curves in stages.

[0116] In addition, the constant current phase of the vehicle can include multiple constant current phases with different charging powers. For example, the constant current phase includes a first sub-constant current phase and a second sub-constant current phase, wherein the charging power of the first sub-constant current phase is greater than the charging power of the second sub-constant current phase. In order to further determine whether the vehicle switches the charging phase in the constant current phase, the LSTM model of the present application is also configured with a phase switching condition: wherein, represents the charging power drop rate of the vehicle within a preset time, and the coefficient -0.5 KW / min represents the power drop rate threshold. It should be noted that the coefficient -0.5 KW / min is only one specific example of the power drop rate threshold in this embodiment, and the power drop rate threshold can be set according to the power change rate in the actual charging process of the vehicle. It should be noted that the duration of the preset time of can be set to be less than the duration required to obtain the real-time power sequence. For example, the real-time power sequence is a power sequence obtained within the last 5 minutes, and the phase conversion is triggered when the power drop rate is less than the power drop rate threshold for consecutive 3 minutes.

[0117] Based on the above steps of determining the charging phase of the vehicle according to the real-time power sequence, the current state of charge, and the phase determination condition, the step of determining the charging phase of the vehicle can include: determining whether σ P < δ s is satisfied; if σ P < δ s is satisfied, determining whether SOC now < θ b is satisfied; if SOC now < θ b is satisfied, determining that the vehicle is in the constant current phase; if SOC now < θ b is not satisfied, determining whether is satisfied; if is satisfied, determining that the vehicle is in the constant voltage phase; if is not satisfied, determining that the vehicle is in the switching phase between the constant current phase and the constant voltage phase; if σ P < δ s is not satisfied, determining whether SOC now < θ b is satisfied; if SOC now < θ b is satisfied, determining that the vehicle is in the switching phase between the two sub-constant current phases; and if SOC now < θ b is not satisfied, determining that the vehicle is in the constant voltage phase.

[0118] Therefore, the charging station load prediction method based on the artificial intelligence prediction model in the embodiment of the present application further refines the determination process of the charging stage through the stage switching condition, and improves the accuracy of the charging power curve generated in stages.

[0119] In some embodiments, for the prediction of the expected charging duration of the constant current stage, the steps of predicting the expected charging duration and the expected power curve of the vehicle in each charging stage in the entire charging process based on the charging stage, the weighted reference power curve, the static charging feature and the dynamic charging feature can include the following steps: according to the charging stage and the weighted reference power curve, the expected starting power of the constant current stage and the model prediction duration of the constant current stage are predicted; based on the model prediction duration, the static charging feature and the dynamic charging feature, the expected charging duration of the constant current stage is calculated; and the expected power curve of the constant current stage is generated according to the expected starting power and the expected charging duration of the constant current stage.

[0120] It should be noted that the weighted reference power curve is used as a reference curve, and according to the power-time curve and the statically input battery capacity C, the SOC at any time t in the curve can be obtained. Therefore, according to the dynamically input SOC now and the current charging stage of the vehicle, the historical time t1 matching the current charging state of the vehicle can be found in the weighted reference power curve, so as to obtain the historical charging power of the current charging stage. In combination with the historical time t1 and the historical end time t2 of the weighted reference power curve in the charging stage, the historical remaining charging duration of the charging stage is obtained. Based on this, the historical remaining charging duration of the current charging stage can be used as the model prediction duration T LSTM , which is predicted by the LSTM model, and the historical charging power of the current charging stage can be used as the expected starting power of the current charging stage, which is predicted.

[0121] In some embodiments, the calculation formula of the expected charging duration of the constant current stage can be:

[0122]

[0123] In the above calculation formula, T1 is the expected charging duration of the constant current stage, T LSTM is the model prediction duration, SOC target is the target state of charge, SOC now is the current state of charge, C is the battery capacity of the vehicle, the coefficient 60 represents the time conversion coefficient, P avg is the current average power of the vehicle.

[0124] Therefore, based on the model prediction duration in combination with the dynamic charging feature and the static charging feature, the prediction of the expected charging duration of the constant current stage is realized.

[0125] In some embodiments, for the prediction of the expected charging duration of the constant-voltage phase, the above-mentioned step of predicting the predicted charging duration of each charging phase and the power curve of the entire charging process of the vehicle based on the charging phase, the weighted reference power curve, the static charging feature, and the dynamic charging feature can include the following steps: predicting the expected initial power of the constant-voltage phase according to the charging phase and the weighted reference power curve; generating the expected power curve of the constant-voltage phase according to the expected initial power of the constant-voltage phase and a preset power decay model, wherein the power decay model dynamically corrects the decay coefficient based on the ambient temperature; and calculating the expected charging duration of the constant-voltage phase based on the expected power curve of the constant-voltage phase and the charging termination power.

[0126] Further, the formula of the above-mentioned power decay model can be:

[0127] P(t) = P0·e -k·t ,

[0128] In the formula of the power decay model, P0 is the expected initial power of the constant-voltage phase, and k is the decay coefficient.

[0129] In addition, the dynamic correction formula of the decay coefficient can be:

[0130]

[0131] In the dynamic correction formula, k base is the preset initial decay coefficient, T env is the ambient temperature, and C is the battery capacity of the vehicle.

[0132] In the formula, 0.5 is only a specific example of the SOC optimal interval, and the SOC optimal interval can be adjusted according to the actual energy storage capacity of the vehicle battery. In a specific embodiment, the SOC optimal interval can be selected from any value in the range of 0.4-0.8. Specifically, as shown in the embodiment, the SOC optimal interval can be set to 0.5.

[0133] In addition, 25 is only a specific example of the optimal charging temperature, and the optimal charging temperature can be adjusted according to the actual charging state of the vehicle battery at different temperatures. In a specific embodiment, the optimal charging temperature can be selected from any value in the range of 20-27. Specifically, as shown in the embodiment, the optimal charging temperature can be set to 25 (unit: °C).

[0134] Therefore, in the embodiment, the calculation formula of the expected charging duration of the constant-voltage phase can be:

[0135]

[0136] In the calculation formula, T2 is the expected charging duration of the constant-voltage phase, Pcutoff The charging termination power is the power at which the charging is terminated. It is noted that the charging termination power can be calculated according to the rated charging power, for example, the charging termination power can be set as 10% of the rated charging power.

[0137] Therefore, based on the model prediction duration, the dynamic charging characteristics and the static charging characteristics are combined to realize the prediction of the expected charging duration in the constant voltage stage. In addition, the total remaining charging duration T total of the vehicle can be calculated as follows: T total = T1+T2.

[0138] Specifically, in the step of predicting the expected charging duration, the charging station load prediction method based on the artificial intelligence prediction model further has a temperature compensation mechanism and a physical limit constraint mechanism. The temperature compensation mechanism is shown in Table 1. When the ambient temperature is in different temperature intervals, the expected charging duration in the constant current stage predicted can be modified by a coefficient, and the decay model can be modified by a coefficient.

[0139] Table 1: Parameter settings for temperature compensation

[0140] Temperature interval Constant current duration correction coefficient Decay coefficient correction coefficient T env <0℃]] ×1.15 ×0.9 0 °C < T env ≤ 25 °C ×1.0 ×1.0 T env >25°C ×0.95 ×1.1

[0141] In Table 1, each constant data is only a specific example in this embodiment, and can be adjusted according to the actual running state of the vehicle under different ambient temperatures.

[0142] In addition, the physical limit constraint condition can include:

[0143] P(t)≤R max ×C, wherein R max is the maximum charging rate of the vehicle, and C is the battery capacity of the vehicle; and

[0144] wherein P max is the rated charging power of the vehicle.

[0145] Therefore, the above physical limit constraint condition can be used to ensure that the charging power does not exceed the battery limit of the vehicle, and the expected charging duration can meet the minimum charging demand.

[0146] Therefore, the charging station load prediction method based on the artificial intelligence prediction model of the embodiment of the present application realizes the accurate quantification of the starting power, duration and decay characteristics of each charging stage by using the static charging characteristics and the dynamic charging characteristics, and realizes the accurate generation of the single vehicle's stage-by-stage charging power curve under multiple constraints, thereby providing more accurate input data for the total power load aggregation of the charging station.

[0147] In some embodiments, after the above step S106, the charging station load prediction method based on the artificial intelligence prediction model of the present application can further include the following steps: in the case that the charging strategy type of the vehicle is a slow charging strategy or a fast charging strategy, the charging power curve of the vehicle is optimized according to the preset stay duration. That is, for the slow charging strategy and the fast charging strategy, the predicted charging power curve of the vehicle can be further optimized according to the preset stay duration set by the user, for example, by considering the time-of-use electricity price or the overall electricity load of the charging station, to avoid charging too many vehicles at the peak of the electricity price or the electricity load, to support the time-of-use electricity price strategy coupling, to predict the load transfer potential in different electricity price periods, so as to realize the further optimization of the power grid scheduling, and thus improve the economy.

[0148] In some embodiments, the above step S108 can include the following steps: using a station load aggregation algorithm, linearly superimposing the charging power curves of all vehicles in the charging station to obtain the overall electricity load curve of the charging station.

[0149] Therefore, the charging station load prediction method based on the artificial intelligence prediction model of the present application realizes the accurate prediction of the charging station load by predicting the charging power curve of a single vehicle and then linearly superimposing the charging power curves of all vehicles to obtain the overall electricity load curve in the future preset time period.

[0150] Figure 2 is a control block diagram of the LSTM model prediction in the charging station load prediction method based on the artificial intelligence prediction model according to an embodiment of the present application, Figure 3 is a schematic diagram of a charging power curve in the charging station load prediction method based on the artificial intelligence prediction model according to an embodiment of the present application, Figure 4 is a schematic diagram of the overall electricity load curve of the charging station in the charging station load prediction method based on the artificial intelligence prediction model according to an embodiment of the present application. As Figures 2-4 shown, taking a charging station equipped with 4 super-charging piles and 8 fast-charging piles as an example, the specific prediction process of the charging station load prediction method of the present application is as follows:

[0151] Step one, receiving the VIN code, battery capacity C, charging request power, preset stay duration, initial SOC, target SOC, maximum charging rate R, and rated charging power P of vehicle A max and other static charging characteristics of vehicle A; max

[0152] Step two, judging the charging strategy type according to the charging request power and the battery type of vehicle A, calling the corresponding LSTM model, analyzing the battery type through the VIN code, and calling the historical charging data of the corresponding battery type in the historical database;​

[0153] Step three, real-time acquisition of the current environmental temperature T of the charging station env , the real-time power sequence of vehicle A in the actual charging process, and the current state of charge SOC of the vehicle now Dynamic charging characteristics, input static charging characteristics, dynamic charging characteristics and historical charging data into the LSTM model.

[0154] Step four, prediction by the LSTM model and output of the total remaining charging time T total = T1+T2, and the charging power curve in the future preset time period: [P(t0), P(t1),..., P(t 30 )];

[0155] Step five, dynamic correction of the output results of the LSTM model according to the dynamically detected dynamic charging characteristics.

[0156] Step six, linear superposition of the charging power curves of all vehicles in the charging station predicted by the charging station load aggregation algorithm to obtain the electricity load curve of the charging station in the future preset time period, and the peak load of the charging station in the future preset time period can also be output.

[0157] Therefore, the charging station load prediction method based on the artificial intelligence prediction model can effectively reduce the prediction error by dividing the strategy modeling and dynamically segmenting the power, is compatible with various charging modes, solves the problem of low prediction accuracy caused by not distinguishing the charging strategy modeling, thereby improving the accuracy of the charging station load prediction, and providing accurate load prediction data for the operation management of the charging station and the dispatching of the power grid, which helps to realize the stable operation of the power grid and the reasonable allocation of resources.

[0158] Figure 5 is a flowchart of a control example of the charging station load prediction method based on the artificial intelligence prediction model according to an embodiment of the application. The control process comprises:

[0159] Step S502, acquiring the static charging characteristics of the vehicle. It should be noted that the static charging characteristics can include battery capacity C, charging request power, preset stay time, initial SOC, target SOC, maximum charging rate R max and rated charging power P max , etc.

[0160] Step S504, determining the charging strategy type of the vehicle based on the charging request power of the vehicle and the battery type.

[0161] Step S506, calling the LSTM model corresponding to the charging strategy type.

[0162] Step S508, real-time acquisition of dynamic charging characteristics of the vehicle and historical charging data corresponding to the vehicle. It should be noted that the dynamic charging characteristics can include the current ambient temperature T of the charging station env , the real-time power sequence of the vehicle A in the actual charging process and the current state of charge SOC of the vehicle now and the like. In addition, the historical charging data can be obtained based on the vehicle type and the charging strategy type of the current vehicle, and can include a plurality of historical charging curves and the historical temperature recorded when each historical charging curve

[0163] Step S510, inputting the static charging characteristics, dynamic charging characteristics and historical charging data into the LSTM model.

[0164] Step S512, weighting calculation of the historical charging data according to the ambient temperature, to obtain a weighted reference power curve.

[0165] Step S514, determining the charging phase of the vehicle according to the real-time power sequence, the current state of charge and the phase determination condition. It should be noted that the charging phase of the vehicle includes the constant current phase and the constant voltage phase.

[0166] Step S516, based on the charging phase, the weighted reference power curve, the static charging characteristics and the dynamic charging characteristics, the predicted expected charging time and the expected power curve of each charging phase of the vehicle in the entire charging process are predicted.

[0167] Step S518, based on the expected charging time and the expected power curve of each charging phase, a charging power curve is generated.

[0168] Step S520, according to the preset time interval, the dynamic charging characteristics obtained in real time are input into the LSTM model periodically, so as to dynamically correct the charging power curve generated by the LSTM model.

[0169] Step S522, superimposing the charging power curves of all vehicles to obtain the overall power load curve of the charging station. At this point, the process ends. Of course, after step S522, the process can be restarted by returning to step S502.

[0170] Therefore, the charging station load prediction method based on the artificial intelligence prediction model provided in the embodiments of the present application can ensure the matching of the charging power curve prediction and the actual charging strategy of the vehicle by dividing the charging strategy of each vehicle and using the LSTM model to predict the charging power curve of each vehicle according to the strategy, solve the problem of low prediction accuracy caused by not distinguishing the charging strategy for modeling, and thus improve the accuracy of the charging station load prediction, and further provide accurate load prediction data for the operation management of the charging station and the dispatching of the power grid, which is helpful to realize the stable operation of the power grid and the reasonable allocation of resources.

[0171] Further, the charging station load prediction method based on the artificial intelligence prediction model provided in the embodiments of the present application can make the model be capable of comprehensively predicting various factors by inputting the static charging features, the dynamic charging features and the historical charging data into the LSTM model, generate a charging power curve that is more in line with the actual situation, improve the accuracy of the charging power curve prediction, and at the same time, dynamically correct the prediction result based on the real-time obtained dynamic charging features at regular intervals, improve the real-time performance of the charging power curve prediction, so that the prediction can timely reflect the changes in the vehicle charging process, improve the timeliness and accuracy of the prediction, and thus further improve the accuracy of the charging station load prediction.

[0172] Further, the charging station load prediction method based on the artificial intelligence prediction model provided in the embodiments of the present application can realize the fine modeling of the charging power dynamic change law in stages through the multi-step prediction mechanism, solve the prediction deviation problem caused by the lack of the phased dynamic modeling, and thus improve the timing accuracy and reliability of the power curve, so that the prediction result can more intuitively reflect the charging process of the vehicle, improve the matching of the charging power curve and the actual charging stage, and thus further improve the accuracy of the charging power curve prediction, and further guarantee the accuracy of the overall power load prediction of the charging station.

[0173] The present embodiment also provides a computer program product 10, a computer readable storage medium 20, and a computer device 30. Figure 6 is a schematic diagram of the computer program product 10 according to an embodiment of the present application, Figure 7 is a schematic diagram of the computer readable storage medium 20 according to an embodiment of the present application, Figure 8 is a schematic block diagram of the computer device 30 according to an embodiment of the present application.

[0174] The computer program product 10 comprises a computer program 11 which, when executed by the processor 32, implements the steps of any of the above-described methods of load forecasting of charging stations based on artificial intelligence prediction models. The computer-readable storage medium 20 has stored thereon the above-described computer program 11 which, when executed by the processor 32, implements the steps of any of the above-described methods of load forecasting of charging stations based on artificial intelligence prediction models. The computer device 30 can comprise a memory 31, a processor 32 and the computer program 11 stored on the memory 31 and running on the processor 32.

[0175] The computer program 11 for performing the operations of the present application can be in assemblies instructions, Instruction Set Architecture (ISA) instructions, machine instructions, machine dependent instructions, microcode, firmware instructions, state-setting data, configuration data for integrated circuits, or in source code or object code written in any combination of one or more programming languages and process programming languages.

[0176] The computer program 11 can execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer, or entirely on the remote computer or server. In the latter scenario, the remote computer can be connected to the user's computer through any type of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or the connection can be made to an external computer (for example, through the Internet using an Internet Service Provider). In some embodiments, electronic circuitry including, for example, programmable logic circuitry, field-programmable gate arrays (FPGA), or programmable logic arrays (PLA) can execute the computer readable program instructions by utilizing state information of the computer readable program instructions to personalize the electronic circuitry, in order to perform aspects of the present application.

[0177] For the description of the present embodiment, the computer program product 10 is a relevant product containing the computer program 11.

[0178] For the description of the present embodiment, the computer-readable storage medium 20 is a tangible device that can retain and store computer program 11, which can be any apparatus that can contain, store, communicate, propagate or transport the program 811 for use by or in connection with the instruction execution system, apparatus or device. More specific examples (a non-exhaustive list) of the computer-readable storage medium 20 include the following: portable computer diskette, hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or Flash memory), static random access memory (SRAM), portable compact disc read-only memory (CD-ROM), digital versatile disk (DVD), memory stick, floppy disk, and any suitable combination thereof.

[0179] At this point, those skilled in the art will fully appreciate that the application need not be limited to the exact details of construction, arrangement or exact embodiments described herein, and that many other variations and modifications can be made without departing from the spirit or scope of the present application. Accordingly, the scope of the application should be judged in terms of the claims that follow.

Claims

1. A charging station load prediction method based on an artificial intelligence prediction model, characterized by The method comprises the following steps: acquiring static charging characteristics of each vehicle charging in the charging station; determining a charging strategy type of each vehicle according to the static charging characteristics; calling an artificial intelligence prediction model corresponding to the charging strategy type, and predicting a charging power curve of each vehicle based on the artificial intelligence prediction model; superimposing the charging power curves of all the vehicles to obtain an overall power load curve of the charging station. 2.The charging station load prediction method based on an artificial intelligence prediction model of claim 1, wherein, The step of predicting the charging power curve of each vehicle based on the artificial intelligence prediction model comprises the following steps: acquiring dynamic charging characteristics of the vehicle and historical charging data corresponding to the vehicle in real time, wherein the dynamic charging characteristics include an environmental temperature, a real-time power sequence, and a current state of charge; inputting the static charging characteristics, the dynamic charging characteristics, and the historical charging data into the artificial intelligence prediction model to generate the charging power curve by the artificial intelligence prediction model; periodically inputting the dynamically acquired dynamic charging characteristics into the artificial intelligence prediction model at a preset time interval to dynamically correct the charging power curve generated by the artificial intelligence prediction model. 3.The charging station load prediction method based on an artificial intelligence prediction model of claim 2, wherein, The artificial intelligence prediction models corresponding to different charging strategy types are configured with different phase determination conditions. The step of generating the charging power curve by the artificial intelligence prediction model comprises the following steps: performing weighted calculation on the historical charging data according to the environmental temperature to obtain a weighted reference power curve; determining a charging phase of the vehicle according to the real-time power sequence, the current state of charge, and the phase determination condition, wherein the charging phase of the vehicle includes a constant current phase and a constant voltage phase; predicting a predicted expected charging duration and an expected power curve of each charging phase of the vehicle in the entire charging process based on the charging phase, the weighted reference power curve, the static charging characteristics, and the dynamic charging characteristics; generating the charging power curve based on the expected charging duration and the expected power curve of each charging phase. 4.The charging station load prediction method based on an artificial intelligence prediction model of claim 3, wherein, The static charging characteristics include a battery capacity, a target state of charge, a maximum charging rate, and a rated charging power of the vehicle; and the step of predicting the expected charging duration and the expected power curve of each charging phase of the vehicle in the entire charging process based on the charging phase, the weighted reference power curve, the static charging characteristics, and the dynamic charging characteristics comprises the following steps: predicting an expected initial power of the constant current phase and a model prediction duration of the constant current phase according to the charging phase and the weighted reference power curve; calculating an expected charging duration of the constant current phase based on the model prediction duration, the static charging characteristics, and the dynamic charging characteristics; generating an expected power curve of the constant current phase according to the expected initial power and the expected charging duration of the constant current phase. 5.The charging station load prediction method based on an artificial intelligence prediction model of claim 4, wherein, The calculation formula of the expected charging duration of the constant current phase is: In the calculation formula, T1 is the expected charging duration of the constant current phase, T LSTM is the model prediction duration, SOC target is the target state of charge, SOC now is the current state of charge, C is the battery capacity of the vehicle, the coefficient 60 represents the time conversion coefficient, P avg is the current average power of the vehicle. 6.The charging station load prediction method based on an artificial intelligence prediction model of claim 3, wherein, The static charging characteristics include a battery capacity, a target state of charge, a maximum charging rate, and a rated charging power of the vehicle; and the step of predicting the predicted charging duration and the predicted power curve of each charging phase of the vehicle during the entire charging process based on the charging phase, the weighted reference power curve, the static charging characteristics, and the dynamic charging characteristics includes: predicting an expected starting power of the constant-voltage phase according to the charging phase and the weighted reference power curve; generating an expected power curve of the constant-voltage phase according to the expected starting power of the constant-voltage phase and a preset power decay model, wherein the power decay model is based on a dynamic correction decay coefficient of the ambient temperature; calculating an expected charging duration of the constant-voltage phase based on the expected power curve of the constant-voltage phase and the charging termination power. 7.The charging station load prediction method based on an artificial intelligence prediction model of claim 6, wherein, The formula of the power decay model is: P(t) = P0e -k·t , In the formula of the power decay model, P0 is the expected starting power of the constant-voltage phase, and k is the decay coefficient; The dynamic correction formula of the decay coefficient is: In the dynamic correction formula, k base is a preset initial attenuation coefficient, T env is the ambient temperature, and C is the battery capacity of the vehicle. The formula for calculating the expected charging duration of the constant-voltage phase is: In the calculation formula, T2 is the expected charging duration of the constant voltage stage, P cutoff is the charging termination power. 8.The charging station load prediction method based on an artificial intelligence prediction model of claim 3, wherein, The historical charging data includes a plurality of historical charging curves and historical temperatures recorded when each of the historical charging curves is generated; The weight calculation formula for the weighted calculation of the historical charging data is: In the weight calculation formula, W i represents the weight of the i-th historical charging curve in the historical charging data, T env is the ambient temperature, is the historical temperature when the i-th historical charging curve is recorded; and The formula of the weighted reference power curve is: n is the number of the historical charging curves, Pti represents the power value of the ith historical charging curve at time t. 9.The charging station load prediction method based on an artificial intelligence prediction model of claim 3, wherein, The phase determination condition includes: σ P <δ s wherein σ P is a standard deviation of the real-time power sequence, and δ s is a standard deviation threshold value corresponding to the charging strategy type; and SOC now < θ b wherein SOC now is the current state of charge, θ b is the battery turning point state of charge of the vehicle; and The step of determining the charging phase of the vehicle according to the real-time power sequence, the current state of charge, and the phase determination condition includes: sequentially determining whether the real-time power sequence and the current state of charge satisfy the phase determination condition; in the case where the determination results are both yes, determining that the vehicle is in the constant-current phase; in the case where the determination results are both no, determining that the vehicle is in the constant-voltage phase. 10.The charging station load prediction method based on an artificial intelligence prediction model of claim 1, wherein, The static charging characteristics include a battery type and a charging request power of the vehicle, and the charging strategy type includes a plurality of charging strategies with different rated charging powers; and The step of determining the charging strategy type of each vehicle according to the static charging characteristics includes: comparing the charging request power with a preset power threshold interval, wherein different power threshold intervals are set according to the rated charging power of the vehicle under different charging strategies; determining the charging strategy type of the vehicle according to the power threshold interval to which the charging request power belongs and the battery type. 11.The charging station load prediction method based on an artificial intelligence prediction model of claim 10, wherein, The charging strategy type includes a slow charging strategy, a fast charging strategy, and a super charging strategy with the rated charging power increasing in turn; and The step of determining the charging strategy type of the vehicle according to the power threshold interval to which the charging request power belongs and the battery type includes: in the case where the power threshold interval to which the charging request power belongs corresponds to the slow charging strategy or the fast charging strategy, taking the charging strategy corresponding to the power threshold interval to which the charging request power belongs as the charging strategy type of the vehicle; in the case where the power threshold interval to which the charging request power belongs corresponds to the super charging strategy, determining whether the vehicle supports super charging according to the battery type. If yes, it is determined that the charging strategy type of the vehicle is an overcharging strategy; If no, it is determined that the charging strategy type of the vehicle is a fast charging strategy. 12.The charging station load prediction method based on an artificial intelligence prediction model of claim 11, wherein, The static charging feature further comprises a preset stay duration; And After the step of predicting the charging power curve of each vehicle based on the artificial intelligence prediction model, the charging station load prediction method based on the artificial intelligence prediction model further comprises: In the case that the charging strategy type of the vehicle is a slow charging strategy or a fast charging strategy, the charging power curve of the vehicle is optimized according to the preset stay duration.

13. A computer program product comprising a computer program, characterized in that, The computer program, when executed by a processor, implements the steps of the charging station load prediction method based on the artificial intelligence prediction model according to any one of claims 1 to 12.

14. A computer readable storage medium having stored thereon a computer program, characterized in that The computer program, when executed by a processor, implements the steps of the charging station load prediction method based on the artificial intelligence prediction model according to any one of claims 1 to 12.

15. A computer device comprising a memory, a processor, and a computer program stored on the memory, wherein the computer program comprises instructions that, when executed by the processor, cause the processor to perform the method of any one of claims 1-14. The processor executes the computer program to implement the steps of the charging station load prediction method based on the artificial intelligence prediction model according to any one of claims 1 to 12.

Images