Intelligent power grid adjusting method for electric vehicle charging load prediction and analysis

Abstract

The invention relates to the technical field of power grid adjustment, and discloses an intelligent power grid adjustment method for electric vehicle charging load prediction analysis, and the method comprises the steps: carrying out the prediction of the charging time period and the charging amount of an electric vehicle for a user through combining the charging information data of the electric vehicle; calculating the charging concentration degree and the charging load of each charging station in different time periods; performing time period overlap ratio analysis on the charging concentration ratios of different time periods; and carrying out charging station charging strategy optimization in combination with the power grid load on the time periods of which the time period overlap ratio is higher than a preset threshold value. According to the method, the charging concentration ratios of different charging stations in different time periods are calculated by using the charging behaviors of the electric vehicle in each time period in each charging process, the time period overlap ratio analysis is performed on the charging concentration ratios of the charging stations in different time periods to obtain the time period with the maximum charging load, and the strategy optimization model is constructed by combining the power grid load to obtain the maximum charging load. The charging strategy of the charging station is optimized, the power grid load in a specific time period is reduced, and the power grid operation efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power grid regulation, and in particular to an intelligent power grid regulation method for predicting and analyzing charging load of electric vehicles. Background Art

[0002] With the acceleration of the global energy transformation process and the promotion of the low-carbon economic model, electric vehicles, as a clean and efficient means of transportation, are gradually replacing traditional internal combustion engine vehicles and becoming an important development direction of the global automotive industry. According to data from the International Energy Agency (IEA), the global ownership of electric vehicles has grown rapidly in the past decade, and it is expected that by 2030, the global ownership of electric vehicles will exceed 200 million. However, the rapid popularization of electric vehicles has also posed new challenges to the power system, the most important of which is the impact of its charging behavior on the grid load. The charging behavior of electric vehicles usually exhibits randomness and volatility, and has significant temporal and spatial aggregation characteristics. For example, users may tend to make appointments for charging, resulting in a surge in load during the evening peak hours. The concentration and uncertainty of this charging behavior pose huge challenges to the dispatching and stable operation of traditional power systems. As the development direction of the new generation of power systems, smart grids can deeply integrate advanced information and communication technologies with power systems to achieve real-time perception, prediction and control of the power grid, providing technical support for the intelligent management of electric vehicle charging loads. Summary of the invention

[0003] In view of this, the present invention provides an intelligent grid regulation method for electric vehicle charging load prediction and analysis, which effectively alleviates the impact of load fluctuations on the power system and improves the operating efficiency and stability of the power grid by accurately predicting the charging behavior of electric vehicles and combining it with a dynamic adjustment strategy.

[0004] To achieve the above object, the present invention provides a smart grid regulation method for predicting and analyzing electric vehicle charging load, comprising the following steps:

[0005] S1: Collect the user's electric vehicle charging information data, and use the electric vehicle charging information data to predict the user's electric vehicle charging period and charging amount;

[0006] S2: Based on the predicted electric vehicle charging period and charging amount, the charging concentration and charging load of each charging station in different periods are evaluated, and the grid load in different periods is predicted. The higher the charging concentration, the higher the charging demand of the charging station in that period.

[0007] S3: Analyze the time period overlap of the charging concentration of each charging station in different time periods;

[0008] S4: For time periods where the time period overlap is higher than a preset threshold, the charging strategy of the charging station is optimized in combination with the grid load to improve the grid operation efficiency.

[0009] As a further improvement method of the present invention:

[0010] Optionally, collecting the user's electric vehicle charging information data in step S1 includes:

[0011] The electric vehicle charging information data includes the charging date, charging location, charging start time, charging end time, the electric power of the electric vehicle at the charging start time, and the electric power of the electric vehicle at the charging end time of each charging process of the electric vehicle.

[0012] Optionally, the predicting of the charging period and charging amount of the electric vehicle for the user in combination with the electric vehicle charging information data includes:

[0013] A day is divided into 24 time periods, each of which is 1 hour long, and the starting and ending points of each time period are both on the hour. The charging behavior of the electric vehicle in each time period is counted, and the charging frequency of the user in each time period is calculated. The calculation process of the charging frequency of the user in any time period is as follows:

[0014] Calculate the charging period of the electric vehicle during each charging process, wherein the charging period is the period between the charging end time and the charging start time;

[0015] Mapping the charging period to 24 periods, wherein the period of the charging period in the 24 periods is marked as 1, and the other periods are marked as 0, to obtain a charging period sequence in each charging process of the electric vehicle, wherein the length of the charging period sequence is 24, corresponding to the marking results of the 24 periods in sequence;

[0016] All charging time period sequences of the electric vehicle are added together to form a charging behavior sequence of the electric vehicle in 24 time periods. The length of the charging behavior sequence is 24, which corresponds to the marking result frequency of the electric vehicle in the 24 time periods. The higher the marking result frequency, the more times the electric vehicle is charged in the time period.

[0017] Based on the charging behavior sequence of the electric vehicle in 24 time periods, the charging frequency of the user in each time period is calculated. The charging frequency of the user in any time period is: the frequency of the marked results in the charging behavior sequence at that moment divided by the number of times the electric vehicle is charged;

[0018] According to the charging frequency of the user in each time period, a time period with a charging frequency higher than a preset frequency threshold is selected, and the selected time period is converted into a continuous time period sequence as the user's electric vehicle charging time period prediction result;

[0019] Extracting the charging date of each charging process of the electric vehicle, the power of the electric vehicle at the start time of charging, and the power of the electric vehicle at the end time of charging from the electric vehicle charging information data, and calculating the charging power of each charging process of the electric vehicle, wherein the charging power of each charging process of the electric vehicle is the difference between the power of the electric vehicle at the end time of charging and the power of the electric vehicle at the start time of charging;

[0020] The charging date and charging power of each charging process of the electric vehicle are used to form charging time series data, and the charging time series data are time-series predicted to obtain the charging power prediction result of the electric vehicle on a future date.

[0021] Optionally, in step S2, the charging concentration and charging load of each charging station in different time periods are evaluated, including:

[0022] Obtaining the prediction results of the charging time period and charging amount of electric vehicles of the main charging users at each charging station, wherein the main charging users of the charging station are users whose charging times at the charging station are higher than a preset charging times threshold;

[0023] Based on the prediction results of the charging time periods of the electric vehicles of the main charging users, the charging concentration of each charging station in the 24 time periods is calculated. The charging concentration of the charging station in the i-th time period is Value(i):

[0024]

[0025] in:

[0026] Sum represents the number of main charging users at the charging station;

[0027] Count(i) indicates the number of main charging users in the i-th period included in the prediction result of electric vehicle charging period;

[0028] w i Indicates the sequence position information of the prediction result of the i-th period in the electric vehicle charging period. The closer the sequence position of the prediction result of the i-th period in the electric vehicle charging period is to the middle, the closer the charging power provided by the charging station to the electric vehicle battery in the i-th period is to the rated power of the charging station, and the greater the sequence position information of the i-th period is;

[0029] The prediction results of the charging amount of electric vehicles of the main charging users are added together to obtain the discharge amount of the charging station in the future. The charging concentration of the charging station in each time period is converted into the time period prediction weight. The time period prediction weight is used to weight the discharge amount of the charging station in the future as the charging load of the charging station in different time periods in the future.

[0030] Optionally, the charging loads of all charging stations at different time periods on future dates are added up as the total charging loads at different time periods on future dates, and the power loads of other power-consuming equipment at different time periods on future dates other than the charging stations are predicted, including:

[0031] The method for predicting the power load of other power-consuming equipment at different time periods on future dates is as follows: constructing a load prediction model with the power load sequence as input, wherein the load prediction model is an LSTM network structure, and the LSTM network includes an input layer, a memory unit, and a gating mechanism, wherein the input layer is used to receive the power load sequence, and the memory unit is used to store the memory information of different time periods in the power load sequence, and the gating mechanism includes a forgetting gate, an input gate, and an output gate, and the forgetting gate is used to selectively forget the memory information stored in the memory unit, and the input gate is used to receive the power load of the current input time period and select the new information to be written into the memory unit, and the output gate is used to extract the old information and the new information retained in the memory unit, perform weighted processing, and obtain the hidden state of the current input time period for output as the power load of the next time period;

[0032] By obtaining the power load sequence of other power-consuming equipment, the load forecasting model is used to output the power load of other power-consuming equipment at different time periods in the future.

[0033] The total charging load and the power load are added together to form the grid load at different times in the future.

[0034] Optionally, in step S3, performing time period overlap analysis on the charging concentration of each charging station in different time periods includes:

[0035] The time period overlap degree of the i-th time period is

[0036]

[0037] in:

[0038] M represents the number of charging stations;

[0039] Num(i) represents the number of charging stations whose charging concentration in the i-th period is higher than the preset concentration threshold value;

[0040] The higher the overlap of the time periods is, the more charging stations with a high charging concentration are in operation during the time period, and the power grid needs to provide more charging loads.

[0041] Optionally, in step S4, for the time periods with a time period overlap higher than a preset threshold, the charging station charging strategy is optimized in combination with the power grid load, including:

[0042] The charging station charging strategy optimization method is as follows:

[0043] Obtain time periods with a time period overlap higher than a preset threshold, and obtain the location of the charging station, the grid load of the time period on a future date, the charging load of the charging station during the time period on a future date, and the number of main charging users at different charging stations during the time period;

[0044] A policy optimization model is constructed by using a reinforcement learning method. The policy optimization model includes a policy network and a value network. The policy network includes a state space and an action space. The state space includes all states of the charging station, and the action space includes all actions that the charging station can take. The state is the location of the charging station, the grid load at the time period on a future date, the charging load of the charging station, and the number of major charging users. The action is the optimization strategy that the charging station can take. The value network is used to measure the strategy value of taking different actions in the current state, and then select the action with a higher strategy value as the optimal optimization strategy of the charging station. The results to be solved in the policy network and the value network are the strategy values ​​of taking different strategies under different states, and the solution is based on the reward function.

[0045] The strategy optimization model receives the location of the charging station, the grid load at the future date, the charging load of the charging station, and the number of main charging users as the state, and selects the action with the highest strategy value from the strategy network as the optimal optimization strategy that the charging station can take, and optimizes the charging strategy of the M charging stations in the power grid to reduce the grid load.

[0046] Optionally, the reward function of the policy optimization model is:

[0047]

[0048] in

[0049] Reward(S|A) represents the expected reward value for taking action A in state S, and Reward(·) represents the reward function;

[0050] f(S|A) represents the grid load after taking action A in state S, and f represents the preset grid load threshold;

[0051] δ represents the load penalty coefficient, and τ represents the load reward coefficient.

[0052] In order to solve the above problem, the present invention provides an electronic device, the electronic device comprising:

[0053] A memory storing at least one instruction;

[0054] Communication interface, enabling electronic equipment to communicate; and

[0055] The processor executes the instructions stored in the memory to implement the smart grid regulation method for electric vehicle charging load prediction and analysis as described above.

[0056] In order to solve the above problems, the present invention also provides a computer-readable storage medium, in which at least one instruction is stored. The at least one instruction is executed by a processor in an electronic device to implement the above-mentioned smart grid regulation method for electric vehicle charging load prediction and analysis.

[0057] Compared with the prior art, the present invention proposes a smart grid regulation method for predicting and analyzing electric vehicle charging load, which has the following advantages:

[0058] First, this scheme proposes a method for predicting the charging load of electric vehicles. The charging behavior of electric vehicles in each time period during each charging process is obtained by dividing the time period, and the charging frequency of users in each time period is calculated. The higher the charging frequency, the greater the probability that the user will charge in this time period. Then, the charging concentration of different charging stations in different time periods is calculated. The higher the charging concentration, the greater the charging load that the charging station needs to provide in this time period, and the greater the number of charging users. In the process of calculating the charging concentration, combined with the rule that the charging process is fast first and then slow, the electric vehicle will be closest to the rated power in the middle charging period. At this time, the charging load that the charging station needs to provide is also greater. The sequence position information is introduced for weighting to improve the characteristic meaning of the charging concentration, effectively represent the size of the charging load in the charging station, and extract the periodic characteristics of the sequence data based on the charging time series data of the electric vehicle. The sequence prediction is performed to obtain the charging amount of the electric vehicle on a future date, thereby realizing the prediction of the charging load of the electric vehicle.

[0059] At the same time, this scheme proposes a grid regulation method. By predicting the load of other power-consuming equipment, the grid load in different time periods is obtained, and the charging concentration of each charging station in different time periods is analyzed for time period overlap. The time period with the largest charging load of each charging station is selected, and a strategy optimization model is constructed in combination with the grid load. A reward function is constructed with the goal of reducing the grid load, and the strategy values ​​of different charging optimization strategies under different states are obtained. The charging strategy of the charging station is optimized, the grid load in a specific time period is reduced, and the grid operation efficiency is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 A schematic flow chart of a smart grid regulation method for predicting and analyzing electric vehicle charging load provided by one embodiment of the present invention.

[0061] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0062] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0063] The embodiment of the present application provides a smart grid regulation method for predicting and analyzing charging loads of electric vehicles. The execution subject of the smart grid regulation method for predicting and analyzing charging loads of electric vehicles includes but is not limited to at least one of the electronic devices such as a server and a terminal that can be configured to execute the method provided by the embodiment of the present application. In other words, the smart grid regulation method for predicting and analyzing charging loads of electric vehicles can be executed by software or hardware installed on a terminal device or a server device, and the software can be a blockchain platform. The server includes but is not limited to: a single server, a server cluster, a cloud server or a cloud server cluster, etc.

[0064] Reference Figure 1 , Embodiment 1 of the present invention is:

[0065] A smart grid regulation method for predicting and analyzing electric vehicle charging load, comprising the following steps:

[0066] S1: Collect the user's electric vehicle charging information data, and use the electric vehicle charging information data to predict the user's electric vehicle charging period and charging amount.

[0067] The step S1 collects the user's electric vehicle charging information data, including:

[0068] The electric vehicle charging information data includes the charging date, charging location, charging start time, charging end time, the electric power of the electric vehicle at the charging start time, and the electric power of the electric vehicle at the charging end time of each charging process of the electric vehicle.

[0069] The method of predicting the charging period and charging amount of an electric vehicle for a user in combination with the electric vehicle charging information data includes:

[0070] A day is divided into 24 time periods, each of which is 1 hour long, and the starting and ending points of each time period are both on the hour. The charging behavior of the electric vehicle in each time period is counted, and the charging frequency of the user in each time period is calculated. The calculation process of the charging frequency of the user in any time period is as follows:

[0071] Calculate the charging period of the electric vehicle during each charging process, wherein the charging period is the period between the charging end time and the charging start time;

[0072] Mapping the charging period to 24 periods, wherein the period of the charging period in the 24 periods is marked as 1, and the other periods are marked as 0, to obtain a charging period sequence in each charging process of the electric vehicle, wherein the length of the charging period sequence is 24, corresponding to the marking results of the 24 periods in sequence;

[0073] All charging time period sequences of the electric vehicle are added together to form a charging behavior sequence of the electric vehicle in 24 time periods. The length of the charging behavior sequence is 24, which corresponds to the marking result frequency of the electric vehicle in the 24 time periods. The higher the marking result frequency, the more times the electric vehicle is charged in the time period.

[0074] Based on the charging behavior sequence of the electric vehicle in 24 time periods, the charging frequency of the user in each time period is calculated. The charging frequency of the user in any time period is: the frequency of the marked results in the charging behavior sequence at that moment divided by the number of times the electric vehicle is charged;

[0075] According to the charging frequency of the user in each time period, a time period with a charging frequency higher than a preset frequency threshold is selected, and the selected time period is converted into a continuous time period sequence as the user's electric vehicle charging time period prediction result; as an embodiment of the present invention, if the charging frequencies of the i-th time period and the i+2-th time period are both higher than the preset frequency threshold, the i+1-th time period is also selected to form a continuous time period sequence;

[0076] Extracting the charging date of each charging process of the electric vehicle, the power of the electric vehicle at the start time of charging, and the power of the electric vehicle at the end time of charging from the electric vehicle charging information data, and calculating the charging power of each charging process of the electric vehicle, wherein the charging power of each charging process of the electric vehicle is the difference between the power of the electric vehicle at the end time of charging and the power of the electric vehicle at the start time of charging;

[0077] The charging date and charging power of each charging process of the electric vehicle are used to form charging time series data, and the charging time series data is time-series predicted to obtain the charging power prediction result of the electric vehicle on a future date. As a preferred embodiment of the present invention, the charging power prediction process of the electric vehicle on a future date is:

[0078] The charging date and charging power of each charging process of the electric vehicle constitute charging time series data, and the representation form of the charging time series data is L:

[0079] L=(L1,L2,...,L n ,...,L N )

[0080] in:

[0081] L1,L2,...,Ln ,...,L N represents the charging capacity of the electric vehicle on N consecutive dates, L n represents the charging power of the electric vehicle on the nth date. If the electric vehicle is not charged on the nth date, then L n =0;

[0082] The R-order timing information a of the charging timing sequence data L is calculated:

[0083]

[0084] in:

[0085] represents the r-th order timing information of the charging timing sequence data L, represents the r-th order periodic information of the charging timing sequence data L, Represents the r-th order standardized periodic information of the charging timing sequence data L, r∈[0,R-1], R is less than N;

[0086] μ represents the mean value of the charging timing sequence data L;

[0087] The R-order timing information is converted to obtain the R-order timing parameters. The conversion formula of the R-order timing information is:

[0088]

[0089] in:

[0090] a0,a1,...,a r ,...,a R-1 is the R-order timing parameter;

[0091] Based on the R-order timing parameters, the charging timing sequence data L is predicted, and the charging power of the electric vehicle on the N+1th date is:

[0092] L N+1 =a0L N +a1L N-1 +...+a r L N-r +...+a R-1 L N-R+1

[0093] in:

[0094] L N+1 is the charging capacity of the electric vehicle on the N+1th date.

[0095] S2: Based on the predicted electric vehicle charging period and charging amount, the charging concentration and charging load of each charging station in different periods are evaluated, and the grid load in different periods is predicted.

[0096] In step S2, the charging concentration and charging load of each charging station at different time periods are evaluated, including:

[0097] Obtaining the prediction results of the charging time period and charging amount of electric vehicles of the main charging users at each charging station, wherein the main charging users of the charging station are users whose charging times at the charging station are higher than a preset charging times threshold;

[0098] Based on the prediction results of the charging time periods of the electric vehicles of the main charging users, the charging concentration of each charging station in the 24 time periods is calculated. The charging concentration of the charging station in the i-th time period is Value(i):

[0099]

[0100] in:

[0101] Sum represents the number of main charging users at the charging station;

[0102] Count(i) indicates the number of main charging users in the i-th period included in the prediction result of electric vehicle charging period;

[0103] w i represents the sequence position information of the prediction result of the i-th time period in the electric vehicle charging time period. The closer the sequence position of the i-th time period in the prediction result of the electric vehicle charging time period is to the middle, the closer the charging power provided by the charging station to the electric vehicle battery in the i-th time period is to the rated power of the charging station, and the greater the sequence position information of the i-th time period is. As a preferred embodiment of the present invention, at the moment when the electric vehicle starts to charge, since the battery power is low at this time, in order to protect the battery and avoid damage to the battery by a large current, the allowed charging power is low, and with the change of time, the charging power will gradually increase. When the battery reaches a high SOC, the charging power will gradually decrease to avoid overcharging. Therefore, the electric vehicle will be closest to the rated power in the middle charging period, and the charging load required to be provided by the charging station at this time will be greater, so w i The calculation formula is:

[0104]

[0105] in:

[0106] Len(i; j) represents the number of time periods included in the prediction result of the electric vehicle charging time period for the jth main charging user in the i-th time period;

[0107] loc(i; j) represents the sequence position of the prediction result of the electric vehicle charging period of the jth main charging user in the i-th period, loc(i; j)∈[1,Len(i; j)];

[0108] The higher the charging concentration, the greater the charging load that the charging station needs to provide during this period, and the greater the number of charging users;

[0109] The prediction results of the charging amount of the electric vehicles of the main charging users are added to obtain the discharge amount of the charging station on the future date, and the charging concentration of the charging station in each time period is converted into the time period prediction weight. The discharge amount of the charging station on the future date is weighted by the time period prediction weight as the charging load of the charging station in different time periods on the future date. As a specific embodiment of the present invention, the charging load of the charging station in the i-th time period on the N+1th date is

[0110]

[0111] in:

[0112] The prediction result of the electric vehicle charging period includes the charging power of the electric vehicle of the jth main charging user in the i-th period on the N+1th date;

[0113] exp(·) represents an exponential function with a natural constant as the base.

[0114] The charging loads of all charging stations at different time periods in the future are added up as the total charging loads at different time periods in the future, and the power loads of other power-consuming equipment at different time periods in the future are predicted, including:

[0115] The method for predicting the power load of other power-consuming equipment at different time periods on future dates is as follows: constructing a load prediction model with the power load sequence as input, wherein the load prediction model is an LSTM network structure, and the LSTM network includes an input layer, a memory unit, and a gating mechanism, wherein the input layer is used to receive the power load sequence, and the memory unit is used to store the memory information of different time periods in the power load sequence, and the gating mechanism includes a forgetting gate, an input gate, and an output gate, and the forgetting gate is used to selectively forget the memory information stored in the memory unit, and the input gate is used to receive the power load of the current input time period and select the new information to be written into the memory unit, and the output gate is used to extract the old information and the new information retained in the memory unit, perform weighted processing, and obtain the hidden state of the current input time period for output as the power load of the next time period;

[0116] By obtaining the power load sequence of other power-consuming equipment, the load forecasting model is used to output the power load of other power-consuming equipment at different time periods in the future.

[0117] The total charging load and the power load are added together to form the grid load at different times in the future.

[0118] In the process of training the load forecasting model, public data sets can be used, such as the Open Power System Data containing power load data of different regions, the Individual household electric power consumption data set, the GEFCom2012 data set, etc.; this embodiment uses the GEFCom2012 data set as the training set and the Individual household electric power consumption data set as the test set.

[0119] S3: Perform time period overlap analysis on the charging concentration of each charging station in different time periods.

[0120] In step S3, the charging concentration of each charging station in different time periods is analyzed for time period overlap, including:

[0121] The time period overlap degree of the i-th time period is

[0122]

[0123] in:

[0124] M represents the number of charging stations;

[0125] Num(i) represents the number of charging stations whose charging concentration in the i-th period is higher than the preset concentration threshold value;

[0126] The higher the overlap of the time periods is, the more charging stations with a high charging concentration are in operation during the time period, and the power grid needs to provide more charging loads.

[0127] S4: For time periods where the time period overlap is higher than a preset threshold, the charging station charging strategy is optimized in combination with the grid load.

[0128] In the step S4, for the time periods where the time period overlap is higher than the preset threshold, the charging station charging strategy is optimized in combination with the grid load, including:

[0129] The charging station charging strategy optimization method is as follows:

[0130] Obtain time periods with a time period overlap higher than a preset threshold, and obtain the location of the charging station, the grid load of the time period on a future date, the charging load of the charging station during the time period on a future date, and the number of main charging users at different charging stations during the time period;

[0131] A policy optimization model is constructed by reinforcement learning. The policy optimization model includes a policy network and a value network. The policy network includes a state space and an action space. The state space includes all states of the charging station, and the action space includes all actions that the charging station can take. The state is the location of the charging station, the grid load in the future date, the charging load of the charging station, and the number of main charging users. The action is the optimization strategy that the charging station can take. The value network is used to measure the strategy value of taking different actions in the current state, and then select the action with a higher strategy value as the optimal optimization strategy of the charging station. The results to be solved in the policy network and the value network are the strategy values ​​of taking different strategies in different states, and the solution is based on the reward function. The solution process is as follows:

[0132] Initialize and generate the policy function values ​​for taking different actions in different states, substitute the states and actions into the reward function, and calculate the expected reward values ​​for taking different actions in different states;

[0133] The difference between the policy function value and the expected reward value is used as the action advantage value of taking different actions in different states;

[0134] Based on the action advantage value, the policy function value and the expected reward value, the policy function value is iterated until the policy function value does not change. The iterative formula of the policy function value is:

[0135]

[0136] in:

[0137] G t (S|A) represents the t-th iteration result of the policy function value of taking action A in state S;

[0138] λ t represents the discount factor of the tth iteration, and Max represents the preset maximum number of iterations;

[0139] Reward(S|A) represents the expected reward value for taking action A in state S;

[0140] F t (S|A) represents the policy function value G of taking action A in state S t (S|A) corresponds to the action advantage value; specifically, F t (S|A)=G t (S|A)-Reward(S|A);

[0141] The policy function value obtained by the final iteration is used as the policy value of taking different policies under different states, and a policy optimization model is constructed;

[0142] Specifically, the optimization strategy that charging stations can adopt is to adjust the charging pricing for time periods with a time overlap higher than a preset threshold, adopt a dynamic pricing strategy, and push other nearby charging stations to users to increase their travel time and avoid charging during this period;

[0143] The strategy optimization model receives the location of the charging station, the grid load at that time in the future, the charging load of the charging station, and the number of main charging users as the state, and selects the action with the highest strategy value from the strategy network as the optimal optimization strategy that the charging station can take. The charging strategy of the M charging stations in the power grid is optimized to reduce the grid load in a specific period of time and improve the grid operation efficiency.

[0144] The reward function of the policy optimization model is:

[0145]

[0146] in

[0147] Reward(S|A) represents the expected reward value for taking action A in state S, and Reward(·) represents the reward function;

[0148] f(S|A) represents the grid load after taking action A in state S, and f represents the preset grid load threshold;

[0149] δ represents the load penalty coefficient, and τ represents the load reward coefficient. Specifically, when the grid load is higher than the preset grid load threshold, the larger the grid load, the lower the reward function, which is expressed as a penalty for the action; when the grid load is lower than the preset grid load threshold, the smaller the grid load, the higher the reward function, which is expressed as a reward for the action, thereby obtaining an action to reduce the grid load in a specific period of time.

[0150] It should be understood that the embodiment is for illustration only and the scope of the patent application is not limited to this structure.

[0151] It should be noted that the serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages and disadvantages of the embodiments. And the terms "including", "comprising" or any other variants thereof in this article are intended to cover non-exclusive inclusion, so that a process, device, article or method including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, device, article or method. In the absence of further restrictions, an element defined by the sentence "including a ..." does not exclude the presence of other identical elements in the process, device, article or method including the element.

[0152] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes a number of instructions for a terminal device (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present invention.

[0153] The above are only preferred embodiments of the present invention, and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A smart grid regulation method for electric vehicle charging load prediction and analysis, characterized in that: The method comprises: S1: Collect the user's electric vehicle charging information data, and use the electric vehicle charging information data to predict the user's electric vehicle charging period and charging amount; S2: Based on the predicted electric vehicle charging period and charging amount, the charging concentration and charging load of each charging station at different time periods are evaluated, and the grid load at different time periods is predicted; S3: Analyze the time period overlap of the charging concentration of each charging station in different time periods; S4: For time periods where the time period overlap is higher than a preset threshold, the charging station charging strategy is optimized in combination with the grid load.

2. The smart grid regulation method for electric vehicle charging load prediction and analysis as claimed in claim 1, characterized in that: The step S1 collects the user's electric vehicle charging information data, including: The electric vehicle charging information data includes the charging date, charging location, charging start time, charging end time, the electric power of the electric vehicle at the charging start time, and the electric power of the electric vehicle at the charging end time of each charging process of the electric vehicle.

3. The smart grid regulation method for predicting and analyzing charging load of electric vehicles as claimed in claim 2, characterized in that: The method of predicting the charging period and charging amount of an electric vehicle for a user in combination with the electric vehicle charging information data includes: A day is divided into 24 time periods, each of which is 1 hour long, and the starting and ending points of each time period are both on the hour. The charging behavior of the electric vehicle in each time period is counted, and the charging frequency of the user in each time period is calculated. The calculation process of the charging frequency of the user in any time period is as follows: Calculate the charging period of the electric vehicle during each charging process, wherein the charging period is the period between the charging end time and the charging start time; Mapping the charging period to 24 periods, wherein the period of the charging period in the 24 periods is marked as 1, and the other periods are marked as 0, to obtain a charging period sequence in each charging process of the electric vehicle, wherein the length of the charging period sequence is 24, corresponding to the marking results of the 24 periods in sequence; All charging time period sequences of the electric vehicle are added together to form a charging behavior sequence of the electric vehicle in 24 time periods. The length of the charging behavior sequence is 24, which corresponds to the marking result frequency of the electric vehicle in the 24 time periods. The higher the marking result frequency, the more times the electric vehicle is charged in the time period. Based on the charging behavior sequence of the electric vehicle in 24 time periods, the charging frequency of the user in each time period is calculated. The charging frequency of the user in any time period is: the frequency of the marked results in the charging behavior sequence at that moment divided by the number of times the electric vehicle is charged; According to the charging frequency of the user in each time period, a time period with a charging frequency higher than a preset frequency threshold is selected, and the selected time period is converted into a continuous time period sequence as the user's electric vehicle charging time period prediction result; Extracting the charging date of each charging process of the electric vehicle, the power of the electric vehicle at the start time of charging, and the power of the electric vehicle at the end time of charging from the electric vehicle charging information data, and calculating the charging power of each charging process of the electric vehicle, wherein the charging power of each charging process of the electric vehicle is the difference between the power of the electric vehicle at the end time of charging and the power of the electric vehicle at the start time of charging; The charging date and charging power of each charging process of the electric vehicle are used to form charging time series data, and the charging time series data are time-series predicted to obtain the charging power prediction result of the electric vehicle on a future date.

4. The smart grid regulation method for electric vehicle charging load forecasting and analysis as claimed in claim 3, characterized in that: In step S2, the charging concentration and charging load of each charging station at different time periods are evaluated, including: Obtaining the prediction results of the charging time period and charging amount of electric vehicles of the main charging users at each charging station, wherein the main charging users of the charging station are users whose charging times at the charging station are higher than a preset charging times threshold; Based on the prediction results of the charging time periods of the electric vehicles of the main charging users, the charging concentration of each charging station in the 24 time periods is calculated. The charging concentration of the charging station in the i-th time period is Value(i): in: Sum represents the number of main charging users at the charging station; Count(i) indicates the number of main charging users in the i-th period included in the prediction result of electric vehicle charging period; w i Indicates the sequence position information of the prediction result of the i-th period in the electric vehicle charging period. The closer the sequence position of the prediction result of the i-th period in the electric vehicle charging period is to the middle, the closer the charging power provided by the charging station to the electric vehicle battery in the i-th period is to the rated power of the charging station, and the greater the sequence position information of the i-th period is; The prediction results of the charging amount of electric vehicles of the main charging users are added together to obtain the discharge amount of the charging station in the future. The charging concentration of the charging station in each time period is converted into the time period prediction weight. The time period prediction weight is used to weight the discharge amount of the charging station in the future as the charging load of the charging station in different time periods in the future.

5. The smart grid regulation method for electric vehicle charging load prediction and analysis as claimed in claim 4, characterized in that: The charging loads of all charging stations at different time periods in the future are added up as the total charging loads at different time periods in the future, and the power loads of other power-consuming equipment at different time periods in the future are predicted, including: The prediction method of the power load of other power-consuming equipment at different time periods in the future is: Construct a load forecasting model with a power load sequence as input, wherein the load forecasting model is an LSTM network structure, the LSTM network includes an input layer, a memory unit and a gating mechanism, the input layer is used to receive the power load sequence, the memory unit is used to store the memory information of different time periods in the power load sequence, the gating mechanism includes a forget gate, an input gate and an output gate, the forget gate is used to selectively forget the memory information stored in the memory unit, the input gate is used to receive the power load of the current input time period, and select the new information to be written into the memory unit, the output gate is used to extract the old information and the new information retained in the memory unit, perform weighted processing, and obtain the hidden state of the current input time period for output as the power load of the next time period; Obtain the power load sequence of other power-consuming equipment, and use the load forecasting model to output the power load of other power-consuming equipment at different time periods on future dates; The total charging load and the power load are added together to form the grid load at different times in the future.

6. The smart grid regulation method for electric vehicle charging load prediction and analysis as claimed in claim 1, characterized in that: The S3 step includes: The time period overlap degree of the i-th time period is in: M represents the number of charging stations; Num(i) represents the number of charging stations whose charging concentration in the i-th period is higher than the preset concentration threshold value; The higher the overlap of the time periods is, the more charging stations with a high charging concentration are in operation during the time period, and the power grid needs to provide more charging loads.

7. The smart grid regulation method for electric vehicle charging load forecasting and analysis as claimed in claim 6, characterized in that: In the step S4, for the time periods where the time period overlap is higher than the preset threshold, the charging station charging strategy is optimized in combination with the grid load, including: The charging station charging strategy optimization method is as follows: Obtain time periods with a time period overlap higher than a preset threshold, and obtain the location of the charging station, the grid load of the time period on a future date, the charging load of the charging station during the time period on a future date, and the number of main charging users at different charging stations during the time period; A policy optimization model is constructed by using a reinforcement learning method. The policy optimization model includes a policy network and a value network. The policy network includes a state space and an action space. The state space includes all states of the charging station, and the action space includes all actions that the charging station can take. The state is the location of the charging station, the grid load at the time period on a future date, the charging load of the charging station, and the number of major charging users. The action is the optimization strategy that the charging station can take. The value network is used to measure the strategy value of taking different actions in the current state, and then select the action with a higher strategy value as the optimal optimization strategy of the charging station. The results to be solved in the policy network and the value network are the strategy values ​​of taking different strategies under different states, and the solution is based on the reward function. The strategy optimization model receives the location of the charging station, the grid load at the future date, the charging load of the charging station, and the number of main charging users as the state, and selects the action with the highest strategy value from the strategy network as the optimal optimization strategy that the charging station can take, and optimizes the charging strategy of the M charging stations in the power grid to reduce the grid load.

8. The smart grid regulation method for electric vehicle charging load forecasting and analysis as claimed in claim 7, characterized in that: The reward function of the policy optimization model is: in Reward(S|A) represents the expected reward value for taking action A in state S, and Reward(·) represents the reward function; f(S|A) represents the grid load after taking action A in state S, and f represents the preset grid load threshold; δ represents the load penalty coefficient, and τ represents the load reward coefficient.

Images