A method for evaluating reliability of an ev aggregated capacity participation reserve market

Abstract

The application provides an EV aggregated capacity participation standby market reliability evaluation method, first, through analyzing the historical charging data of the electric vehicle user, the electric vehicle user is classified; second, based on the compound Poisson process, the available discharge capacity and reliability model of the aggregator participating in the day-ahead auxiliary service bidding is provided; based on the model, the aggregator determines the V2G electric quantity of any time slot bidding of t+1 day according to the reliability requirement of the standby market on the day-ahead signed standby capacity; and then the compound Poisson probability model is introduced to calculate the power supply reliability of the EV aggregator at a certain moment within a day. The application is helpful for the optimization of the power grid resources, guarantees the safe and stable operation of the power grid, and has important reference significance for the EV aggregator participating in the power market auxiliary service with the distributed energy storage.

Description

Technical Field

[0001] This invention relates to a reliability assessment method for EV aggregators, and more particularly to a reliability assessment method for EV aggregator capacity participating in the standby market, belonging to the field of new energy technology. Background Technology

[0002] Electric vehicle (EV) clusters have a promising future in backup services, offering higher reliability and flexibility compared to traditional electricity market backup services. EV clusters participating in V2G (Vehicle-to-Grid) are essential for achieving the goals of "carbon peaking and carbon neutrality." Considering that EVs spend most of their time parked, EV aggregators can utilize these resources to participate in ancillary services such as regulating grid voltage and frequency, thereby further improving grid reliability and the economic benefits for EV users. Based on the grid load demand at each moment, EVs participating in V2G are uniformly allocated, and a composite Poisson probability model is provided for the V2G power supplied by EV aggregators at each moment during peak demand periods, assessing the reliability of the power supply capacity of EV aggregators at each peak load moment. This assessment result contributes to the coordinated optimization of grid voltage and is of great significance for the safe and stable operation of the grid.

[0003] The risk for EV aggregators utilizing EV clusters for backup services stems from the possibility that multiple contracted EV users may be unable to participate in V2G at designated times, meaning they refuse to provide V2G service while their contracted EVs are in motion. If the number of defaulting users is large at a given moment, and the EV aggregator has committed a high amount of electricity to the grid at that moment, it will be unable to fulfill the contracted V2G electricity quota. Due to the uncertainty of EV users' driving behavior, they may not be able to participate in V2G at a given moment, and the V2G electricity quota loss caused to the EV aggregator by each defaulting user is also uncertain.

[0004] Therefore, a reliable model is needed to analyze the reliability of EV aggregators. Summary of the Invention

[0005] The purpose of this invention is to provide a method for evaluating the reliability of EV aggregate capacity in the standby market.

[0006] To achieve the above objectives, the present invention provides the following solution: A method for reliability assessment of the aggregated capacity of electric vehicles participating in the standby market includes the following steps: Step S1: Extract the charging characteristics of each EV user from the historical charging data of EV users in the EV user cluster. The charging characteristics include charging time period, expected daily average charging amount and charging time interval. Classify EV users based on the charging characteristics. Step S2: Calculate the available V2G power for each EV user at the beginning of each day based on the EV user's battery capacity and average daily power consumption. Step S3: Based on the annual historical data of the EV user cluster, determine the probability distribution type of the V2G capacity available for the EV user cluster, and evaluate the results. Step S4: Based on the composite Poisson distribution, establish a reliability model of the V2G capacity of the EV user cluster at any time of day; Step S5: Based on the EV user cluster's subscribed standby capacity before the current time of day, calculate the available V2G capacity and reliability of the EV user cluster at any time of day.

[0007] Furthermore, in step S1, EV users are divided into three categories: short-distance commuters, long-distance commuters, and home-based users. Daily charging demand expected by short-distance commuters 3~6 kWh ; Daily charging volume expected by long-distance commuters Greater than 6 kWh ; Daily charging demand expected by home-based users Less than 3 kWh .

[0008] Furthermore, in step S2, calculating the available V2G power for each EV user at the initial moment of each day includes the following steps: Step S201, calculate the confidence level of EV users. Average daily power consumption: (1) in, For EV users M Average daily charging capacity For confidence intervals Statistical value For EV users M Standard deviation of daily charging capacity; Step S202, calculate the available V2G capacity for each EV user at the initial time of each day: (2) in, Battery capacity for EV users.

[0009] Furthermore, in step S3, determining the probability distribution type of V2G capacity available to the EV user cluster and evaluating the results includes the following steps: Step S301: Establish evaluation criteria for the V2G capacity fitting effect of EV user clusters: Calculate the confirmation coefficient R-squared: (3) (4) (5) in, J The total number of sample data. For the first j The vertical axis value of the sample data of the available V2G capacity of an EV user cluster. for The fitted theoretical value, The mean of the sample data; Step S302: Construct the probability density function of the mixture Gaussian distribution of the available V2G power consumption of the EV user cluster at any time of day: (6) (7) (8) in , representing the k The weighting coefficients of the Gaussian components. For the first k The probability density function of Gaussian components, For sample vectors, Let Gaussian be the parameter vector of the mixture distribution. Let the parameter vector of the Gaussian distribution be the first... k One portion, and The distribution parameters are those of a mixture Gaussian distribution. K Let Gaussian be the order of the mixture model. T The number of time slots; Step S303, calculate the EV user cluster in the... Discharge amount in time slot Distribution function: (9) = in, For EV clusters in the first The minimum power supply or bidding power for the time slot. K Let the order of the Gaussian mixture model be denoted by . Step S304: Calculate the EV users in the EV user cluster. i In the Available V2G power in time slots: (10) in, For EV users i exist t The electricity that has already participated in the bidding and has been locked and allocated by the EV cluster before the time slot.

[0010] Furthermore, in step 301, the parameters are estimated. The method includes the following steps: Step A: Initialize parameters: For parameters , Assign initial values, Step B: Estimate the probabilistic model parameters for V2G power samples in each EV cluster: (11) Step C: [The text appears to be incomplete and contains several grammatical errors. A more accurate translation would require the full context The parameters at which the maximum value is obtained are used as the parameters for the new model: (12) (13) (14) Step D: Calculate the log-likelihood function: (15) Step E: Determine whether the parameters or the log-likelihood function converge. If neither converges, proceed to step B.

[0011] Furthermore, the reliability model of the V2G capacity of the EV user cluster at any time of day in step S4 consists of the following steps: Step 401: Establish a composite Poisson distribution model: (16) (17) in, n The number of EVs participating in V2G at time t; Let i be the discharge amount of EV user i in time slot t; V2G quota allocated by the EV aggregator to EV user i in time slot t; For Poisson parameters, The number of participants in V2G in time slot t. for t V2G power supply available from time-slot EV clusters.

[0012] Step S402: Calculate the expected number of EV users participating in V2G for each time slot, and use it as the Poisson parameter for the corresponding time slot. : (18) in, For the first t The initial number of EV users selected during V2G sorting to participate in the discharge activity. For the first t Number of EV users who missed their scheduled appointments This represents the probability of the corresponding number of EV users defaulting on their orders appearing in the annual statistics; when At that time, it is determined that the EV aggregator has successfully completed the contracted volume promised to the power grid dispatch center in time slot t.

[0013] By adopting the above technical solution, the present invention discloses the following technical effects: 1. This invention combines a large amount of charging data from EV users to explore the regularity of EV user behavior, classifies them reasonably, establishes a reliability model for EV aggregators based on V2G auxiliary services, and reasonably evaluates the V2G capacity available for EVs at each moment. 2. This invention determines the probability distribution of V2G power available to EV clusters at corresponding times by fitting the data, and uses a composite Poisson model to calculate the reliability change of EV aggregators under different bidding values ​​at each time. This method can provide EV aggregators with V2G bidding ranges at different times, and EV users can interact with EV aggregators in real time through smart terminals, which can fully improve the reliability and flexibility of EV clusters participating in V2G. Attached Figure Description

[0014] Figure 1 This is a flowchart of the present invention. Detailed Implementation

[0015] Example 1: The purpose of this invention is to provide an analytical method for quantifying the reliability of electric vehicle aggregation capacity in the day-ahead market when electric vehicle aggregation capacity participates in the standby market. This method comprehensively considers EV user categories and various participation conditions, providing the power grid dispatch center with the reliability of EV aggregators at each moment, thereby improving the reliability of EV clusters participating in V2G.

[0016] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0017] A method for reliability assessment of the aggregated capacity of electric vehicles participating in the standby market includes the following steps: Step S1: Based on the historical charging data of EVs, extract the charging characteristics of each EV user, including their usual charging time, average daily charging amount, and charging time interval. Based on this, EV users are classified.

[0018] Step S2: Based on the EV's battery capacity and average daily power consumption, calculate the available V2G power for each EV user at the beginning of the day.

[0019] Step S3: Based on the historical data of the EV cluster throughout the year, determine the probability distribution type of the V2G capacity available for the EV cluster, and evaluate the results.

[0020] Step S4: Based on the composite Poisson distribution, establish a reliability model for the V2G capacity of the EV cluster at any time of day.

[0021] Step S5: Based on the EV cluster's contracted standby capacity before the current time of the day, calculate the V2G capacity and reliability that the EV cluster can provide at any time before the current time of the day.

[0022] In step S1, based on historical EV charging data, data features of each EV user are extracted, and EV users are classified according to these data features. Specifically, this includes: Step S101: By analyzing historical EV data, determine the common charging periods, average daily power consumption, and charging intervals for each EV. Based on this, users are divided into three categories: short-distance commuters, long-distance commuters, and home-based users. For short-distance commutes, the average daily distance is 15-35 km, corresponding to an expected daily charging amount of 3-6 kWh. kWh ; Long-distance commuting is defined as an average daily mileage exceeding 35km, corresponding to an expected daily charging volume greater than 6 kWh. kWh ; Users who stay at home travel the least, therefore their charging intervals are the longest. h It has the lowest expected daily charging volume. ; In step S2, the available V2G power for each EV user at the beginning of the day is calculated based on the EV battery capacity and average daily power consumption, specifically including: S201, calculate the average daily power consumption of EV users, and estimate its confidence interval using the normal approximation method: Since EVs can gain additional benefits from participating in V2G, assuming EV users keep their EVs fully charged at the start of the day, the EV user at a confidence level The average daily power consumption is as follows: (1) in, This is the average charging amount calculated by analyzing the annual charging data of EVs. For confidence intervals Statistical value To analyze the standard deviation of the annual charging data, M This refers to the number of days included in the year.

[0023] S202, calculate the available V2G capacity for each EV at the beginning of the day: Considering the direct impact of V2G behavior on EV battery life, 20% of the battery capacity is taken as the minimum daily charge level to prevent EV users from over-discharging. i The available V2G capacity for each EV user at the start of the day is: (2) in, This refers to the EV battery capacity.

[0024] In step S3, based on the historical data of the EV cluster throughout the year, the probability distribution type of the EV cluster's V2G capacity is determined, and the results are evaluated, specifically including: S301, the evaluation criteria for V2G capacity fitting effect of EV clusters of different sizes are as follows: The formula for calculating the sum of squared errors between theoretical values ​​and corresponding points in the sample data is: (3) The formula for calculating the sum of squares of the differences between the sample data and the mean is: (4) The fit confirmation coefficient R-squared is determined by SSE and SST, and its value ranges from [0,1]. Its calculation formula is as follows: (5) in, J The total number of sample data. For the first j The vertical axis value of the sample data of the available V2G capacity of an EV user cluster. for Fitting theoretical values, The mean of the sample data is denoted as . The closer the coefficient of confirmation is to 1, the higher the goodness of fit.

[0025] S302, by analyzing the V2G capacity fitting effect of EV clusters of different sizes, it was found that the V2G power of the EV cluster at a certain moment satisfies a mixed Gaussian distribution.

[0026] The probability density function of the mixture Gaussian distribution is: (6) in ,represent k The weighting coefficients of the Gaussian components. For the first k The probability density function of Gaussian components, For sample vectors, Let Gaussian be the parameter vector of the mixture distribution. For the first K A vector of distribution parameters of a Gaussian distribution. K In this embodiment, the order of the Gaussian mixture distribution model is given. K =1 or 2, k This indicates the number of Gaussian components contained in the mixture Gaussian distribution.

[0027] The EV group is described by a mixture of Gaussian distributions in the 1st... Discharge amount per time slot The distribution function is as follows: (7) (8) (9) in, For EVA Minimum power supply of EVA in each time slot (in this article) Take EVA in the first (Electricity volume bid per time slot), For EVA The maximum discharge capacity of the EVA in each time slot. , , where is the distribution parameter of the mixture Gaussian distribution. K In this embodiment, the order of the Gaussian mixture distribution model is given. K =1 or 2. Different time slots result in different EV volumes and different distribution parameters participating in V2G.

[0028] S303, EV clusters of different sizes have different fitting parameters.

[0029] The method for calculating the fitted samples is as follows: Time slot EV users i The available power for V2G is: (10) in, For EVs t The power that was bid on and locked and allocated by the EV cluster agent EVA before the time slot.

[0030] Since the V2G power released by the EV cluster prior to that time slot must be considered in any time slot, the release of some V2G power will change the subsequent discharge capability of the EV cluster, thus affecting the distribution parameters of the mixture Gaussian distribution. However, the mixture Gaussian distribution is obtained by weighted summation of multiple Gaussian components. When calculating the parameters of each component, due to the difficulty in differentiating the likelihood function and the singularity of the distribution itself, an analytical solution cannot be obtained directly by differentiation. Therefore, this paper uses historical EV charging data and... t The parameters of the Gaussian mixture distribution are obtained by solving the bidding situation before the time slot using the expectation-maximization algorithm.

[0031] The specific algorithm for estimating the expectation is as follows: Step A: Initialize parameters: Step B: Estimate the probabilistic model parameters for V2G power samples in each EV cluster: 1) (11) Step C: [The text appears to be incomplete and contains several grammatical errors. A more accurate translation would require the full context The parameters at which the maximum value is obtained are used as the parameters for the new model: (12) (13) (14) Step D: Calculate the log-likelihood function: (15) Step E: Check whether the parameters converge or the log-likelihood function converges. If they do not converge, return to step B.

[0032] In step S4, the composite Poisson distribution refers to the fact that when an event following a Poisson process occurs, each event is accompanied by a random variable, such as economic loss or equipment damage.

[0033] Given time (0, t Number of events within ) N ( t ) is a ratio of Poisson process Let random variable With fault events Related variables Is it related to the fault? i Related critical faults (e.g., faults) i (The resulting economic losses). Hypothetical variables It is an independent random variable.

[0034] Specifically, it includes: S401, the composite Poisson distribution model is... (16) (17) in, n The number of EVs participating in V2G at time t; Let be the discharge amount of the i-th EV at time t; V2G quota allocated by the EV aggregator to the i-th EV at time t; For a unit of time (15) min The number of EVs participating in discharge within the ) represent t time The distribution function, To analyze the data throughout the year Received t V2G power available for EV clusters at any time Let t be the number of participants in V2G. for t V2G power available from the EV cluster.

[0035] S402, the method for calculating the Poisson parameter λ in the reliability model is as follows: By analyzing EV users' annual charging data and combining it with user types, we can determine their travel habits and statistically analyze data for each time period (15...). min The expected value of the number of EVs participating in V2G is used as the Poisson parameter at each time step. (Take a positive integer).

[0036] (18) in, for t Selected during V2G sorting t The initial number of EVs participating in the discharge activity at any given time. For the first statistical t The number of EVs that fail to show up at the scheduled time This represents the probability of the corresponding number of defaulted EVs occurring in the annual statistical data. At that time, it is determined that the EV aggregator has successfully completed the contracted volume promised to the power grid dispatch center at time t.

[0037] In step S5, based on the EV cluster's contracted backup capacity before the current time, the V2G capacity and reliability that the EV cluster can provide at any time before the current time are calculated.

[0038] This invention provides an analytical method for quantifying the reliability of electric vehicle (EV) aggregation capacity in the day-ahead market when participating in the standby market. It combines extensive charging data from EV users to explore the patterns in EV user behavior and classify them appropriately. The method rationally evaluates the EV's V2G capacity at each moment, determines its probability distribution by fitting the V2G capacity of the EV cluster at the corresponding moment, and calculates the reliability change of EV aggregators under different bidding values ​​at each moment using a load Poisson model. This method can provide EV aggregators with V2G bidding ranges at different times, and EV users can interact with EV aggregators in real time via smart terminals, significantly improving the reliability and flexibility of EV clusters participating in V2G.

[0039] The method for calculating the responsive capacity and reliability of EV aggregators can achieve the following objectives: 1) fully leveraging the V2G potential of EV clusters; 2) improving the management level of EV aggregators over EV clusters; 3) providing the grid dispatch center with the reliability of EV aggregators at each moment, laying the groundwork for subsequent economic rewards and penalties for EV aggregators; 4) while obtaining the reliability at the corresponding moment, EV aggregators also receive a bidding reference range for that moment the following day, maintaining their own V2G reliability and providing a basis for rewards and penalties for individual EVs; 5) the obtained reliability model can be used as part of the objective function or constraint conditions for subsequent microgrid voltage optimization. The calculation method provided by this invention yields the reliability of EV aggregators at each moment, promoting the optimization of power resources and quantifying the indicators of EV cluster participation in V2G reliability.

[0040] Currently, most research on EV clusters participating in V2G focuses on optimized scheduling and the economic benefits brought to all parties involved, with little mention of the reliability of EV clusters participating in V2G. When power grid dispatch centers consider rewarding or penalizing EV aggregators, or when EV aggregators implement rewards or penalties for EV users, corresponding quantitative criteria are needed. Calculating the corresponding capacity and reliability of EV aggregated loads can serve as a standard for measurement, improving the peak-shaving effect of EV clusters. This invention effectively meets this need.

[0041] Table 1

Claims

1. A method for reliability assessment of the aggregated capacity of electric vehicles participating in the standby market, characterized in that, Includes the following steps: Step S1: Extract the charging characteristics of each EV user from the historical charging data of EV users in the EV user cluster. The charging characteristics include charging time period, expected daily average charging amount and charging time interval. Classify EV users based on the charging characteristics. Step S2: Calculate the available V2G power for each EV user at the beginning of each day based on the EV user's battery capacity and average daily power consumption. Step S3: Based on the annual historical data of the EV user cluster, determine the probability distribution type of the V2G capacity available for the EV user cluster, and evaluate the results, including the following steps: Step S301: Establish evaluation criteria for the V2G capacity fitting effect of EV user clusters: Calculate the confirmation coefficient R-squared: (1) (2) (3) in, J The total number of sample data. For the first j The vertical axis value of the sample data of the available V2G capacity of an EV user cluster. for The fitted theoretical value, The mean of the sample data; Step S302: Construct the probability density function of the mixture Gaussian distribution of the available V2G power consumption of the EV user cluster at any time of day: (4) (5) (6) in , representing the k The weighting coefficients of the Gaussian components. For the first k The probability density function of Gaussian components, For sample vectors, Let Gaussian be the parameter vector of the mixture distribution. Let the parameter vector of the Gaussian distribution be the first... k One portion, and The distribution parameters are those of a mixture Gaussian distribution. K Let Gaussian be the order of the mixture model. T The number of time slots; Step S303, calculate the EV user cluster in the... Discharge amount in time slot Distribution function: (7) = in, For EV clusters in the first The minimum power supply or bidding power for the time slot. K Let the order of the Gaussian mixture model be denoted by . Step S304: Calculate the EV users in the EV user cluster. i In the Available V2G power in time slots: (8) in, For EV users i exist t Electricity that had already participated in the bidding and was locked and allocated by the EV cluster before the time slot; Step S4: Based on the composite Poisson distribution, establish a reliability model for the V2G capacity of the EV user cluster at any time of day, specifically as follows: Step 401: Establish a composite Poisson distribution model: (9) (10) in, n The number of EVs participating in V2G at time t; Let i be the discharge amount of EV user i in time slot t; V2G quota allocated by the EV aggregator to EV user i in time slot t; For Poisson parameters, The number of participants in V2G in time slot t. for t V2G power supply available from time-slot EV clusters; Step S402: Calculate the expected number of EV users participating in V2G in each time slot, and use it as the Poisson parameter for the corresponding time slot. : (11) in, For the first t The initial number of EV users selected during V2G sorting to participate in the discharge activity. For the first t Number of EV users who missed their scheduled appointments This represents the probability of the corresponding number of EV users defaulting on their orders appearing in the annual statistics; when At that time, it is determined that the EV aggregator successfully completed the contracted volume promised to the power grid dispatch center in time slot t; Step S5: Based on the EV user cluster's subscribed standby capacity before the current time of day, calculate the available V2G capacity and reliability of the EV user cluster at any time of day.

2. The method for assessing the reliability of electric vehicle aggregated capacity in the standby market according to claim 1, characterized in that, In step S1, EV users are divided into three categories: short-distance commuters, long-distance commuters, and home-based users. Daily charging demand expected by short-distance commuters 3~6 kWh ; Daily charging volume expected by long-distance commuters Greater than 6 kWh ; Daily charging demand expected by home-based users Less than 3 kWh .

3. The method for assessing the reliability of electric vehicle aggregate capacity participation in the standby market according to claim 1, characterized in that, In step S2, calculating the available V2G power for each EV user at the beginning of each day includes the following steps: Step S201, calculate the confidence level of EV users. Average daily power consumption: (12) in, For EV users M Average daily charging capacity For confidence intervals Statistical value For EV users M Standard deviation of daily charging capacity; Step S202, calculate the available V2G capacity for each EV user at the initial time of each day: (13) in, Battery capacity for EV users.

4. The method for assessing the reliability of electric vehicle aggregate capacity participation in the standby market according to claim 1, characterized in that, The parameter estimation in step 302 The method includes the following steps: Step A: Initialize parameters: For parameters , Assign initial values; Step B: Estimate the probabilistic model parameters for V2G power samples in each EV cluster: (14) Step C: [The text appears to be incomplete and contains several grammatical errors. A more accurate translation would require the full context.] The parameters at which the maximum value is obtained are used as the parameters for the new model: (15) (16) (17) Step D: Calculate the log-likelihood function: (18) Step E: Determine whether the parameters or the log-likelihood function converge. If neither converges, proceed to step B.

Images