Electric vehicle time sequence classification aggregation frequency modulation optimization model considering responsivity

Abstract

The invention discloses an electric vehicle time sequence classification aggregation frequency modulation optimization model considering responsivity. The model comprises the following steps: establishing a power energy operation boundary for performing energy exchange of an EV; classifying each single EV according to the EV network access time and the network access time period, and aggregating the power operation boundary and the energy operation boundary of the single EVs with the same network access time period to obtain an EVA time sequence classification aggregation model; depicting the relationship between different types of EV responsivity and price excitation through the Weber-Fechner law according to the EV classification to obtain an EVA time sequence classification aggregation model considering the responsivity under charging and discharging excitation; and by considering the day-ahead and real-time two-stage income and cost of the EV participating in the energy-frequency modulation market after EV classification aggregation, and by taking the maximum EVA operation income as a target, establishing an EVA time sequence classification aggregation frequency modulation optimization model capable of giving consideration to both precision and efficiency. The model is used for accurately and efficiently evaluating the EVA frequency modulation capability, formulating an incentive mechanism and optimizing an EVA operation strategy.

Description

Technical Field

[0001] The present invention relates to the technical field of power system operation control, and in particular to an electric vehicle time series classification aggregation frequency modulation optimization model considering responsiveness. Background Art

[0002] The global electric vehicle (EV) industry is experiencing rapid growth. As a key source of flexibility in the new power system, EVs possess time-scalable charging and discharging characteristics. Regulating their charging and discharging behavior can enhance grid flexibility and promote renewable energy consumption. By coordinating and managing the charging and discharging behavior of large-scale EVs, EV aggregators (EVAs) can further tap into their regulatory potential and provide effective power support during grid frequency regulation.

[0003] Existing EVA models primarily include the traditional EVA model and the independent EVA model. Both models are based on a single-EV model that includes power and energy boundaries. The traditional EVA model directly constructs the EVA model by summing the energy and power operating boundaries of all single EVs at each moment, obtaining the energy and power operating boundaries of the EVA. The number of constraints is fixed and independent of the number of EVs. The independent EVA model, on the other hand, directly uses the energy and power boundaries of individual EVs in the full single-EV model as EVA constraints. The number of constraints is proportional to the number of EVs. The traditional EVA model is the most common. Because its fixed number of constraints allows for fast solutions to problems involving a large number of EVs, however, the simplified constraints can lead to errors in the resulting solutions. The independent EVA model produces accurate solutions, but because its number of constraints is proportional to the number of EVs, it encounters a large number of constraints, resulting in high computational complexity and slow solutions when solving problems involving a large number of EVs. Therefore, existing research still lacks an EVA model that strikes a balance between computational accuracy and solution efficiency, and further research is urgently needed on an EVA modeling method that can balance accuracy and efficiency.

[0004] Furthermore, the regulatory capacity of EVA is influenced by the willingness of EV users to respond. To fully tap the regulatory potential of EVs, it is necessary to design appropriate incentive mechanisms to guide user response to scheduling. Some existing studies assume that all EV users will actively respond to incentive schemes and fail to delve into the relationship between incentive mechanisms and user response willingness. Other studies, while considering the impact of EV user response willingness on regulatory capacity and the relationship between incentive mechanisms and user response willingness, fail to consider the differences in responsiveness of EVs for different uses to price incentives.

[0005] In the prior art, a method for incentivizing electric vehicles for grid power dispatch, disclosed in publication number CN117639039A, is employed to directly establish an EVA model by summing the energy and power operating boundaries of all individual EVs. This method employs a traditional EVA model with a fixed number of constraints. Because the constraints for EVs at different grid access times are simplified by direct superposition when establishing the EVA model, errors exist in the solution. Furthermore, the scheme does not provide an accurate relationship for EV responsiveness and fails to account for differences in responsiveness caused by differences in EV usage under the same incentive. Consequently, the scheme's control results are insufficiently refined, making it difficult to accurately assess and fully utilize the EVA's regulatory capabilities, and balancing solution speed and accuracy. Summary of the Invention

[0006] 1. Technical problems to be solved:

[0007] In response to the above technical problems, the present invention provides an EVA time-series classification aggregation frequency modulation method considering user responsiveness. First, EVs are classified according to the time period of their access to the network, and EVs in different time periods of access to the network are aggregated separately and then constraints are established, and an EVA time-series classification aggregation modeling method is proposed; then, taking into account the responsiveness differences under the incentives of EVs for different purposes, the relationship between the user responsiveness of EVs for different purposes and price incentives is characterized by the Weber-Fechner law, and an EVA time-series classification aggregation model considering responsiveness is obtained; considering the day-ahead and real-time two-stage benefits and costs of EVs participating in the energy-frequency modulation market after classification and aggregation, an EVA time-series classification aggregation frequency modulation optimization model is established with the goal of maximizing EVA operating benefits, which is used to accurately and efficiently evaluate EVA frequency modulation capabilities, formulate incentive mechanisms, and optimize EVA operating strategies.

[0008] 2. Technical solution:

[0009] An electric vehicle time series classification aggregation frequency modulation optimization model considering responsiveness is characterized in that: its construction process includes:

[0010] Step 1: Establish the power operation boundary and energy operation boundary of individual EVs for energy exchange; classify each individual EV according to the EV grid access time and grid access period, aggregate the power operation boundary and energy operation boundary of individual EVs with the same grid access period to obtain the EVA time series classification aggregation model;

[0011] Step 2: Classify EVs by usage and analyze the differences in responsiveness under charging and discharging incentives for different EV types. Use the Weber-Fechner law to characterize the relationship between user responsiveness and price incentives for different EV types, and derive an EVA time-series classification aggregation model that considers responsiveness under charging and discharging incentives.

[0012] Step 3: Considering the day-ahead and real-time benefits and costs of EVs participating in the energy-frequency regulation market after EV classification and aggregation, with the goal of maximizing EVA operating benefits, an EVA time-series classification and aggregation frequency regulation optimization model that takes into account both accuracy and efficiency is established.

[0013] Furthermore, step one specifically includes:

[0014] S11: When a single EV is connected to the grid, it is considered as an energy storage unit that can exchange energy with the grid in both directions. Assume that the period of energy exchange between the single EV and the grid is [t in ,t end ], the energy operation boundary of the energy exchange between a single EV and the grid is as follows:

[0015]

[0016] In the above formula, are the upper and lower limits of the energy of the i-th EV at time t; Δt is the research time interval; e in,i is the initial energy of the i-th EV; t in,i and t end,i are the time when the i-th EV enters and leaves the grid; p c,max and p d,max are the maximum charge and discharge power of EV respectively; η c and η d are the charging and discharging efficiency of EV respectively; C is the battery capacity of EV; e min The minimum energy to prevent EV from over-discharging; e need,i The energy required to complete charging of the i-th EV;

[0017] The power operating boundary of energy exchange between a single EV and the grid is as follows:

[0018]

[0019] In the above formula: are the upper and lower limits of the power that the i-th EV can use to exchange energy with the grid at time t;

[0020] S13: Based on the time and duration of EV access, aggregate the EVs with the same time and duration to obtain the aggregate energy boundary and aggregate power boundary of all EVs in different access periods as shown in the following formula, thus obtaining the EVA time series classification aggregation model;

[0021]

[0022] In the above formula, N ΔT,τIt represents the number of EVs with a duration of ΔT and a time of τ when connected to the network; are the upper and lower bounds of the energy at time t for an EV with a grid access duration of ΔT and a grid access time of τ; are the upper and lower bounds of the power at time t for an EV with a grid access duration of ΔT and a grid access time of τ; are the upper and lower limits of energy consumption of the ith EV at time t, whose duration of access to the network is ΔT and the time of access to the network is τ; are the upper and lower bounds of the power of the i-th EV at time t, whose access time is ΔT and access time is τ.

[0023] Furthermore, step 2 specifically includes:

[0024] S21: EV responsiveness differences analyzed using response price incentives; response price incentive types include: response charging price incentive mechanism (CIM) and response discharge incentive mechanism (DIM);

[0025] For EV users who respond to charging price incentives, the charging incentive price r is described using the Weber-Fechner law as follows: CS Relationship with user responsiveness;

[0026] Q(r CS )=αlnr CS +c (4)

[0027] In the above formula: r CS is the charging incentive price; Q(r CS ) is the incentive electricity price r CS EV response rate when ; c is EV response constant; α represents response coefficient; α and c are determined by fitting the survey data;

[0028] S22: For EV users who respond to the discharge incentive mechanism, additional compensation is required for the battery loss cost caused by discharge. The discharge incentive price given by EVA is set based on the battery loss cost and the charging incentive price:

[0029] r DS =r CS +∑(r0)

[0030] Q(r DS )=αln(r DS -∑(r0))+c (5)

[0031] In the above formula, r DS is the discharge incentive price; Q(r DS ) is the incentive electricity price r DS ∑(r0) is the EV response rate during the entire grid connection period;

[0032] S23: EVs are divided into multiple types according to their uses, and the response coefficients and response constants of the upper and lower limits of each type of EV response are preset. These are then substituted into equations (4) and (5) to obtain the relationship between the charge and discharge incentive electricity price and the upper and lower limits of the charge and discharge response of EVs for each use;

[0033] S24: Based on the relationship between the charging and discharging incentive electricity price and the upper and lower limits of the charging and discharging response of each type of EV, the upper and lower limits of the comprehensive response rate of users participating in EV frequency regulation are obtained by substituting the following formula:

[0034]

[0035] In the above formula, Q c,max (r CS ), Q c,min (r CS ), Q d,max (r DS ), Q d,min (r DS ) are the upper and lower bounds of the charging response rate and the upper and lower bounds of the discharging response rate of EV participating in frequency modulation, respectively, which constitute the comprehensive user response rate of EV participating in frequency modulation;

[0036] Since the response ratios of EV to CIM and DIM in EVA are evenly distributed between their upper and lower limits, the preset discharge subsidy price r DS and charging incentive price r CS When , the response ratio of EV in EVA to CIM and DIM can be obtained as follows:

[0037]

[0038] In the above formula: Q c (r CS ) and Q d (r DS ) represent the proportion of EVs responding to CIM and DIM in EVA, respectively; U[·] represents uniform distribution;

[0039] S25: The Monte Carlo method is used to sample the response boundary of each EV to obtain the upper and lower boundaries of the energy and power response capabilities of the EV aggregate at each grid access time and grid access period as shown in the following formula;

[0040]

[0041] In the above formula, σ ΔT,τ,i =1 means that the i-th EV among the EVs with a network access duration of ΔT and a network access time of τ responds to CIM, otherwise σ ΔT,τ,i =0;υ ΔT,τ,i =1 means that the i-th EV among the EVs with a duration of ΔT and a time of τ responds to DIM, otherwise υ ΔT,τ,i =0.

[0042] Furthermore, the objective function constructed in step 3 with the goal of maximizing EVA operating income is as follows:

[0043] max F=-F1-F2+F3+F4+F5 (9)

[0044] In the above formula, F is the operating income of EVA; F1 is the charging cost of EVA in the day-ahead energy market; F2 is the operating cost of EVA in the real-time energy market; F3 is the frequency regulation capacity income of EVA; F4 is the frequency regulation mileage income of EVA; F5 is the electricity sales income obtained by EVA from users and the incentive fee given to users. The specific expression of the five costs is:

[0045] S31: Charging cost F1 is:

[0046]

[0047] In the above formula: F1 is the charging cost of EVA in the day-ahead energy market; r Chr (k) is the electricity price in the day-ahead energy market at time k; P Chr,ΔT,τ (k) is the charging plan power for an EV with a grid connection time of ΔT and a grid connection time of τ in time period k; K is the entire study period; Δk is the time period interval. In this formula, the grid connection time is divided into 24 grid connection times, 0-23, and each grid connection time is divided into 23 grid connection time durations, 1-23, forming a total of 23*24 grid connection time periods, i.e., 23*24 categories. EVs in each category are aggregated to obtain 23*24 energy and power aggregation boundaries, which serve as constraints of the model.

[0048] S32: EVA regulates the EVs participating in real-time frequency regulation at different access times, causing changes in charging power. The resulting energy costs are settled according to the real-time energy price. The operating cost F2 of EVA in the real-time energy market is as follows:

[0049]

[0050] In the above formula, r RT (k) is the electricity price in the real-time energy market during period k; P UP,ΔT,τ (k) and P DN,ΔT,τ (k) represents the upward and downward frequency modulation power of the EV with a network access time of ΔT and a network access time of τ in the k period;

[0051] S33: The capacity benefit F3 of EVA participating in the frequency regulation market is:

[0052]

[0053] In the above formula: r RC(k) is the frequency regulation capacity electricity price in period k; P RC,ΔT,τ (k) is the frequency modulation capacity of an EV with a network access time of ΔT and a network access time of τ in time period k; λ is the performance score;

[0054] S34: EVA regulation The mileage income F4 obtained by EVs participating in the frequency regulation market at different network access times is:

[0055]

[0056] In the above formula: r M (k) is the frequency regulation mileage electricity price in the k period; in the above formula, m UP (k) and m DN (k) are the upward and downward frequency modulation output mileage of the EV in each grid access period in the k period; specifically:

[0057]

[0058] In the above formula, N m is the number of time intervals of the signal in the k period; A(j,t) is the FM indication signal released by the FM market at time t, A(j,t)∈[-1,1], t∈k, j∈N m ;

[0059] S35: During each access period, the EV is connected to the grid and charged, responding to the regulation of the EVA. The EVA obtains the charging benefits provided by the EV and feeds back the discharge subsidy to the EV. The charging benefits and subsidy fee F5 provided by the EVA to the EV are as follows:

[0060]

[0061] Where: P ΔT,τ (k) is the net power of the EV that responds to the stimulus in the k period when the grid access time is ΔT and the grid access time is τ; P0(k) is the net power of the EV that does not respond to the stimulus; P EVA (k) is the net power of EVA response; r C (k) is the charging electricity price published by EVA in time period k.

[0062] Furthermore, in step 3, EVA participates in the energy-frequency regulation market by utilizing its EV battery energy storage system as an adjustable load to obtain operating benefits. EVA's day-ahead and real-time operating power and energy in the energy-frequency regulation market must remain within their upper and lower bounds. Specifically:

[0063] S36: The planned EV charging power for each access period determined by the EVA in each period and the net energy after the response signal must remain between the upper and lower limits of the energy in that period:

[0064]

[0065] Where: are the upper and lower bounds of the energy of an EV with a network access duration of ΔT and a network access time of τ in time period k.

[0066] S37: EVA responds to changes in the frequency modulation signal by controlling the charge and discharge power of each EV. Therefore, the net power of the EV response during each grid access period must be maintained between the maximum discharge power and the maximum charge power of the EV during each period:

[0067]

[0068] Where: are the upper and lower bounds of the power of an EV with a grid access duration of ΔT and a grid access time of τ in time period k;

[0069] S38: The upward and downward frequency modulation powers of EVs in each grid access period are all non-negative numbers:

[0070]

[0071] Furthermore, in step S25, the Monte Carlo method is used to sample the response boundary of each EV, and the specific process is as follows:

[0072] S251: Set the incentive prices of CIM and DIM, and obtain the EV ratio Q of the response CIM based on formula (7) c (r CS ) and the EV ratio Q in response to DIM d (r DS );

[0073] S252: Monte Carlo sampling is used to generate the purpose of each single EV in the EVA and the initial energy when it enters the network;

[0074] S253: Construct a random number that meets U[0,1] by Monte Carlo sampling, and compare the random number with Q c (r CS ) and Q d (r DS ) to determine whether EV responds to CIM and DIM, thereby determining the 0-1 variable σ ΔT,τ,i 、υ ΔT,τ,i ;

[0075] S254: Obtaining upper and lower bounds of the operating power and energy of the single EV according to the single EV model;

[0076] S255: Repeatedly sample steps S252 to S254 to obtain the upper and lower limits of the comprehensive response rate of users participating in EV frequency modulation.

[0077] Furthermore, in S13, the same network access time and network access time are aggregated according to the network access time and network access time of the EVs, specifically:

[0078] S131: Classify all EVs that have joined the network according to the time of joining the network, and obtain the EVs at each time of joining the network;

[0079] S132: Classify all EVs at each network access time according to the network access duration to obtain EVs in each network access time period at the network access time;

[0080] S133: Aggregate EVs with the same network access time and the same network access period by superimposing energy and power boundaries, and obtain energy and power aggregation boundaries of EVs in each network access period.

[0081] Furthermore, EVs are classified according to their usage, and are divided into three types: taxis, private cars, and other vehicles.

[0082] Furthermore, in the EVA time-series classification and aggregation frequency modulation optimization model, EVA participates in a multi-stage and multi-variety market by regulating the charging and discharging process of EVs in each grid access period classified and aggregated according to the grid access period, thereby obtaining operating income.

[0083] Further,

[0084] 3.Beneficial effects:

[0085] (1) In the electric vehicle time series classification aggregation frequency modulation optimization model considering responsiveness disclosed by this method, individual EVs are classified according to their time periods of access to the grid, and then the EVA models of different time periods of access to the grid are aggregated to establish constraints, thereby obtaining an EV time series classification aggregation model. This model can avoid the problem of decreased solution accuracy caused by the traditional EVA model directly superimposing the boundaries of all EVs. Compared with the EVA independent model, this model can reduce the constraint dimension from the vehicle order level to the time period order level with less loss of accuracy, and can take into account both solution speed and accuracy when dealing with problems involving a large number of EVs.

[0086] (2) In the electric vehicle time series classification aggregation frequency modulation optimization model considering responsiveness disclosed by this method, EVs are classified according to their uses, and the Weber-Fechner law is used to describe the relationship between the responsiveness of EV users with different uses and price incentives. An EVA time series classification aggregation model considering responsiveness is constructed to establish the optimal charging and discharging incentive mechanism.

[0087] (3) In the electric vehicle time series classification aggregation frequency regulation optimization model considering responsiveness disclosed by this method, an EVA time series classification aggregation frequency regulation optimization model is constructed with the goal of maximizing EVA operating benefits. This model participates in the day-ahead and real-time two-stage energy markets and frequency regulation markets by regulating the charging and discharging processes of EVs in each grid access period classified and aggregated according to the grid access period, thereby obtaining the maximum operating benefits. BRIEF DESCRIPTION OF THE DRAWINGS

[0088] Figure 1 Schematic diagram of constructing the EVA time series classification aggregation model in this method;

[0089] Figure 2 This is the relationship diagram between the charging and discharging incentive electricity price and the upper and lower limits of the charging and discharging responsiveness of EVs for various purposes in this method;

[0090] Figure 3 This is the relationship diagram between the charging and discharging incentive electricity price and the upper and lower limits of the user's comprehensive responsiveness in this method;

[0091] Figure 4 is the relationship between the incentive price and EVA operating income in the comparison ratio;

[0092] Figure 5 This is a relative error diagram of the power optimization results of the EVA traditional model and the EVA time series classification aggregation model of this method in the comparative example;

[0093] Figure 6 Graphs showing the running time of three EVA models in a scenario involving 1,000 EVs in a specific embodiment;

[0094] Figure 7 The specific flow chart of this method is shown in FIG. DETAILED DESCRIPTION

[0095] The present invention will be described in detail below with reference to the accompanying drawings.

[0096] In order to make the technical purpose, technical solutions and beneficial effects of the present invention clearer, the present method will be compared with various existing methods below.

[0097] As attached Figure 7 As shown in Figure 1, an electric vehicle time series classification aggregation frequency modulation optimization model considering responsiveness is constructed, and its construction process includes:

[0098] Step 1: Establish the power operation boundary and energy operation boundary of the single EV for energy exchange; classify each single EV according to the EV grid access time and grid access period, aggregate the power operation boundary and energy operation boundary of the single EV with the same grid access period to obtain the EVA time series classification aggregation model; the structure of the EVA time series classification aggregation model and its construction process are shown in the attached figure. Figure 1 shown.

[0099] Step 2: Classify EVs by usage and analyze the differences in responsiveness under charging and discharging incentives for different EV types. Use the Weber-Fechner law to characterize the relationship between user responsiveness and price incentives for different EV types, and derive an EVA time-series classification aggregation model that considers responsiveness under charging and discharging incentives.

[0100] Step 3: Considering the day-ahead and real-time benefits and costs of EVs participating in the energy-frequency regulation market after EV classification and aggregation, with the goal of maximizing EVA operating benefits, an EVA time-series classification and aggregation frequency regulation optimization model that takes into account both accuracy and efficiency is established.

[0101] Comparative Example:

[0102] This comparative example uses data from the US PJM (Pennsylvania—New Jersey—Maryland) electricity market. Each EV's battery storage capacity is assumed to be 50 kWh, the rated charging power is set to 15 kW, the minimum state of charge (SOC) is set to 20%, the minimum required SOC is set to 80%, the charging and discharging efficiencies are both set to 90%, and the initial SOC follows a normal distribution. Monte Carlo simulations are performed using a one-day study period, using the Gurobi solver in Python to simulate the behavior of each EV.

[0103] In this comparative example, the relationship between the charging incentive electricity price and the upper and lower limits of the charging and discharging responsiveness of EVs for various purposes is shown in the attached figure. Figure 2 As shown in (a), the relationship between the discharge incentive electricity price and the upper and lower limits of the charge and discharge response of EVs for various purposes is shown in the attached figure. Figure 2 (b) shows the relationship between the set charging incentive electricity price and the upper and lower limits of the user's comprehensive responsiveness. Figure 3 As shown in (a), the relationship between the discharge incentive price and the upper and lower limits of the user's comprehensive responsiveness is as follows: Figure 3 (b) Set the relationship between price incentives and comprehensive operating income as follows: Figure 4 As shown in the figure, operating revenue reaches its maximum when the incentive price reaches 40 USD / MWh. This is because the EV response rate has already reached a high level at this point. Further increasing the incentive price has little effect on improving the EV response rate, but instead increases the operating costs of EVA. Therefore, after the incentive price exceeds 40 USD / MWh, operating revenue decreases, providing a reference for actual conditions. Therefore, the charging incentive price is set at 40 USD / MWh, and the discharging incentive price is set at the charging incentive price plus battery loss compensation.

[0104] EVs with different uses have different responsiveness under the same price incentive. In order to fine-tune the responsiveness of EV users for different uses, in this comparison, EVs are divided into taxis, private cars, and other vehicles (buses and logistics vehicles, etc.) according to their uses. Taxis are most sensitive to price incentives and have the highest responsiveness under the same price incentives because their behavior is mainly for profit. Other vehicles have the lowest responsiveness under the same price incentives because their operating trajectories are relatively fixed and their dispatchable flexibility is small. The responsiveness of private cars under the same price incentives is between taxis and other vehicles. The relationship between the incentive electricity price and the upper and lower limits of the charge and discharge responsiveness of EVs for different uses is as follows: Figure 2 The EV proportions for each purpose are shown in the following table.

[0105] Table 1 EV share by application

[0106]

[0107] In this comparative example, the traditional EVA model, the EVA time series classification aggregation model of this method, and the EVA independent model are used to solve the examples containing different numbers of EVs. The incentive price is set to the optimal price determined in the previous section. The result of the EVA independent model solution is used as the accurate result. The relative error of the operating power optimization results of the EVA traditional model and the EVA time series classification aggregation model in the scenarios containing different numbers of EVs is calculated as follows: Figure 5 shown.

[0108] Calculations were performed for scenarios with 2 to 1,000 EVs. The EVA time-series classification aggregation model based on this method achieved a lower relative error than the traditional EVA model. In scenarios with 5 to 1,000 EVs, the traditional EVA model exhibited errors, which increased with the number of EVs. The error was most pronounced in the scenario with 1,000 EVs, reaching a relative error of 5.91%. In contrast, the EVA time-series classification aggregation model presented no errors in scenarios with 2 to 100 EVs. However, the error increased slightly with the number of EVs in scenarios with 200 to 1,000 EVs. Errors in the EVA time-series classification aggregation model began to appear when the number of EVs reached 200. The model error was 0.04% for 200 EVs, 0.17% for 500 EVs, and 0.63% for 1,000 EVs, representing only 10.69% of the error of the traditional EVA model in these scenarios. It can be seen that the EVA time series classification aggregation model proposed in the present invention has obvious accuracy advantages over the traditional EVA model when solving problems containing a large number of EVs.

[0109] In a scenario involving 1,000 EVs, the running times of the traditional EVA model, the EVA time series classification aggregation model of this application, and the EVA independent model are as follows: Figure 6As shown. The running time of the model includes the model establishment time and the solution time, among which the EVA independent model has the longest running time, the EVA traditional model has the shortest, and the EVA time series classification aggregation model is in the middle. The EVA time series classification aggregation model can achieve a solution accuracy of 99.37% of the EVA independent model with a running time of 28% of the EVA independent model. It can be seen that under the premise of maintaining a high level of solution accuracy, the EVA time series classification aggregation model proposed in the present invention has obvious advantages in solution speed compared with the EVA independent model.

[0110] Although the present invention has been disclosed above in terms of preferred embodiments, they are not intended to limit the present invention. Anyone skilled in the art can make various changes or modifications without departing from the spirit and scope of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection defined by the claims of this application.

Claims

1. A time-series classification and aggregation frequency modulation optimization model for electric vehicles considering responsiveness, characterized by: The construction process includes: Step 1: Establish the power operation boundary and energy operation boundary of individual EVs for energy exchange; classify each individual EV according to the EV grid access time and grid access period, aggregate the power operation boundary and energy operation boundary of individual EVs with the same grid access period to obtain the EVA time series classification aggregation model; Step 2: Classify EVs by usage and analyze the differences in responsiveness under charging and discharging incentives for different EV types. Use the Weber-Fechner law to characterize the relationship between user responsiveness and price incentives for different EV types, and derive an EVA time-series classification aggregation model that considers responsiveness under charging and discharging incentives. Step 3: Considering the day-ahead and real-time benefits and costs of EVs participating in the energy-frequency regulation market after EV classification and aggregation, with the goal of maximizing EVA operating benefits, an EVA time-series classification and aggregation frequency regulation optimization model that takes into account both accuracy and efficiency is established.

2. The electric vehicle time series classification aggregation frequency modulation optimization model considering responsiveness according to claim 1 is characterized by: Step 1 specifically includes: S11: When a single EV is connected to the grid, it is considered as an energy storage unit that can exchange energy with the grid in both directions. Assume that the period of energy exchange between the single EV and the grid is [t in ,t end ], the energy operation boundary of the energy exchange between a single EV and the grid is as follows: In the above formula, are the upper and lower limits of the energy of the i-th EV at time t; Δt is the research time interval; e in,i is the initial energy of the i-th EV; t in,i and t end,i are the time when the i-th EV enters and leaves the grid; p c,max and p d,max are the maximum charge and discharge power of EV respectively; η c and η d are the charging and discharging efficiency of EV respectively; C is the battery capacity of EV; e min The minimum energy to prevent EV from over-discharging; e need,i The energy required to complete charging of the i-th EV is: In the above formula: are the upper and lower limits of the power that the i-th EV can use to exchange energy with the grid at time t; S13: Based on the time and duration of EV access, aggregate the EVs with the same time and duration to obtain the aggregate energy boundary and aggregate power boundary of all EVs in different access periods as shown in the following formula, thus obtaining the EVA time series classification aggregation model; In the above formula, N ΔT,τ It represents the number of EVs with a duration of ΔT and a time of τ when connected to the network; are the upper and lower bounds of the energy at time t for an EV with a grid access duration of ΔT and a grid access time of τ; are the upper and lower bounds of the power at time t for an EV with a grid access duration of ΔT and a grid access time of τ; are the upper and lower limits of energy consumption of the ith EV at time t, whose duration of access to the network is ΔT and the time of access to the network is τ; are the upper and lower bounds of the power of the i-th EV at time t, whose access time is ΔT and access time is τ.

3. The electric vehicle time series classification and aggregation frequency modulation optimization model considering responsiveness according to claim 1 is characterized by: Step 2 specifically includes: S21: EV responsiveness differences analyzed using response price incentives; response price incentive types include: response charging price incentive mechanism (CIM) and response discharge incentive mechanism (DIM); For EV users who respond to charging price incentives, the charging incentive price r is described using the Weber-Fechner law as follows: CS Relationship with user responsiveness; Q(r CS )=αlnr CS +c (4) In the above formula: r CS is the charging incentive price; Q(r CS ) is the incentive electricity price r CS EV response rate when ; c is EV response constant; α represents response coefficient; α and c are determined by fitting the survey data; S22: For EV users who respond to the discharge incentive mechanism, additional compensation is required for the battery loss cost caused by discharge. The discharge incentive price given by EVA is set based on the battery loss cost and the charging incentive price: r DS ＝r CS +∑(r0) Q(r DS )=αln(r DS -∑(r0))+c (5) In the above formula, r DS is the discharge incentive price; Q(r DS ) is the incentive electricity price r DS ∑(r0) is the EV response rate during the entire grid connection period; S23: EVs are divided into multiple types according to their uses, and the response coefficients and response constants of the upper and lower limits of each type of EV response are preset. These are then substituted into equations (4) and (5) to obtain the relationship between the charge and discharge incentive electricity price and the upper and lower limits of the charge and discharge response of EVs for each use; S24: Based on the relationship between the charging and discharging incentive electricity price and the upper and lower limits of the charging and discharging response of each type of EV, the upper and lower limits of the comprehensive response rate of users participating in EV frequency regulation are obtained by substituting the following formula: In the above formula, Q c,max (r CS ), Q c,min (r CS ), Q d,max (r DS ), Q d,min (r DS ) are the upper and lower bounds of the charging response rate and the upper and lower bounds of the discharging response rate of EV participating in frequency modulation, respectively, which constitute the comprehensive user response rate of EV participating in frequency modulation; Since the response ratios of EV to CIM and DIM in EVA are evenly distributed between their upper and lower limits, the preset discharge subsidy price r DS and charging incentive price r CS When , the response ratio of EV in EVA to CIM and DIM can be obtained as follows: In the above formula: Q c (r CS ) and Q d (r DS ) represent the proportion of EVs responding to CIM and DIM in EVA, respectively; U[·] represents uniform distribution; S25: The Monte Carlo method is used to sample the response boundary of each EV to obtain the upper and lower boundaries of the energy and power response capabilities of the EV aggregate at each grid access time and grid access period as shown in the following formula; In the above formula, σ ΔT,τ,i =1 means that the i-th EV among the EVs with a network access duration of ΔT and a network access time of τ responds to CIM, otherwise σ ΔT,τ,i =0;υ ΔT,τ,i =1 means that the i-th EV among the EVs with a duration of ΔT and a time of τ responds to DIM, otherwise υ ΔT,τ,i =0.

4. The electric vehicle time series classification aggregation frequency modulation optimization model considering responsiveness according to claim 1 is characterized by: The objective function constructed in step 3 with the goal of maximizing EVA operating income is as follows: max F=-F1-F2+F3+F4+F5 (9) In the above formula, F is the operating income of EVA; F1 is the charging cost of EVA in the day-ahead energy market; F2 is the operating cost of EVA in the real-time energy market; F3 is the frequency regulation capacity income of EVA; F4 is the frequency regulation mileage income of EVA; F5 is the electricity sales income obtained by EVA from users and the incentive fee given to users. The specific expression of the five costs is: S31: Charging cost F1 is: In the above formula: F1 is the charging cost of EVA in the day-ahead energy market; r Chr (k) is the electricity price in the day-ahead energy market at time k; P Chr,ΔT,τ (k) is the charging plan power for an EV with a grid connection time of ΔT and a grid connection time of τ in time period k; K is the entire study period; Δk is the time period interval; in this formula, the grid connection time is divided into 24 grid connection times, 0-23, and each grid connection time is further divided into 23 grid connection time durations, 1-23, forming a total of 23*24 grid connection time periods, i.e., 23*24 categories. EVs in each category are aggregated to obtain 23*24 energy and power aggregation boundaries, which serve as constraints of the model; S32: EVA regulates the EVs participating in real-time frequency regulation at different access times, causing changes in charging power. The resulting energy costs are settled according to the real-time energy price. The operating cost F2 of EVA in the real-time energy market is as follows: In the above formula, r RT (k) is the electricity price in the real-time energy market during period k; P UP,ΔT,τ (k) and P DN,ΔT,τ (k) represents the upward and downward frequency modulation power of the EV with a network access time of ΔT and a network access time of τ in the k period; S33: The capacity benefit F3 of EVA participating in the frequency regulation market is: In the above formula: r RC (k) is the frequency regulation capacity electricity price in period k; P RC,ΔT,τ (k) is the frequency modulation capacity of an EV with a network access time of ΔT and a network access time of τ in time period k; λ is the performance score; S34: EVA regulation The mileage income F4 obtained by EVs participating in the frequency regulation market at different network access times is: In the above formula: r M (k) is the frequency regulation mileage electricity price in the k period; in the above formula, m UP (k) and m DN (k) are the upward and downward frequency modulation output mileage of the EV in each grid access period in the k period; specifically: In the above formula, N m is the number of time intervals of the signal in the k period; A(j,t) is the FM indication signal released by the FM market at time t, A(j,t)∈[-1,1], t∈k, j∈N m ; S35: During each access period, the EV is connected to the grid and charged, responding to the regulation of the EVA. The EVA obtains the charging benefits provided by the EV and feeds back the discharge subsidy to the EV. The charging benefits and subsidy fee F5 provided by the EVA to the EV are as follows: Where: P ΔT,τ (k) is the net power of the EV that responds to the stimulus in the k period when the grid access time is ΔT and the grid access time is τ; P0(k) is the net power of the EV that does not respond to the stimulus; P EVA (k) is the net power of EVA response; r C (k) is the charging electricity price published by EVA in time period k.

5. The electric vehicle time series classification aggregation frequency modulation optimization model considering responsiveness according to claim 4 is characterized by: In step 3, EVA participates in the energy-frequency regulation market by utilizing its EV battery energy storage system as an adjustable load to obtain operating benefits. EVA's day-ahead and real-time operating power and energy in the energy-frequency regulation market must remain within their upper and lower bounds. Specifically: S36: The planned EV charging power for each access period determined by the EVA in each period and the net energy after the response signal must remain between the upper and lower limits of the energy in that period: Where: are the upper and lower bounds of the energy of an EV with a network access time of ΔT and a network access time of τ in time period k; S37: EVA responds to changes in the frequency modulation signal by controlling the charge and discharge power of each EV. Therefore, the net power of the EV response during each grid access period must be maintained between the maximum discharge power and the maximum charge power of the EV during each period: Where: are the upper and lower bounds of the power of an EV with a grid access duration of ΔT and a grid access time of τ in time period k; S38: The upward and downward frequency modulation powers of EVs in each grid access period are all non-negative numbers:

6. The electric vehicle time series classification aggregation frequency modulation optimization model considering responsiveness according to claim 3 is characterized by: In step S25, the Monte Carlo method is used to sample the response boundary of each EV. The specific process is as follows: S251: Set the incentive prices of CIM and DIM, and obtain the EV ratio Q of the response CIM based on formula (7) c (r CS ) and the EV ratio Q in response to DIM d (r DS ); S252: Monte Carlo sampling is used to generate the purpose of each single EV in the EVA and the initial energy when it enters the network; S253: Construct a random number that meets U[0,1] by Monte Carlo sampling, and compare the random number with Q c (r CS ) and Q d (r DS ) to determine whether EV responds to CIM and DIM, thereby determining the 0-1 variable σ ΔT,τ,i 、υ ΔT,τ,i ; S254: Obtaining upper and lower bounds of the operating power and energy of the single EV according to the single EV model; S255: Repeatedly sample steps S252 to S254 to obtain the upper and lower limits of the comprehensive response rate of users participating in EV frequency modulation.

7. The electric vehicle time series classification aggregation frequency modulation optimization model considering responsiveness according to claim 3 is characterized by: In S13, the same network access time and network access time are aggregated according to the network access time and network access time of the EVs, specifically: S131: Classify all EVs that have joined the network according to the time of joining the network, and obtain the EVs at each time of joining the network; S132: Classify all EVs at each network access time according to the network access duration to obtain EVs in each network access time period at the network access time; S133: Aggregate EVs with the same network access time and the same network access period by superimposing energy and power boundaries, and obtain energy and power aggregation boundaries of EVs in each network access period.

8. The electric vehicle time series classification aggregation frequency modulation optimization model considering responsiveness according to claim 1 is characterized by: EVs are classified according to their usage into three types: taxis, private cars, and other vehicles.

9. The electric vehicle time series classification aggregation frequency modulation optimization model considering responsiveness according to claim 1 is characterized by: In the EVA time-series classification and aggregation frequency modulation optimization model, EVA participates in a multi-stage, multi-variety market by regulating the charging and discharging process of EVs in each grid access period classified and aggregated according to the grid access period, thereby obtaining operating income.

Images